Immune reconstitution in children following chemotherapy for haematological malignancies: a long-term follow-up

Authors


Cornelis M. van Tilburg, Department of Paediatric Haematology/Oncology, University Medical Center Utrecht, Room number KC.03·063·0, P.O. Box 85090, 3508 AB Utrecht, The Netherlands. E-mail: c.m.vantilburg@umcutrecht.nl

Summary

Modern intensive chemotherapy for childhood haematological malignancies has led to high cure rates, but has detrimental effects on the immune system. There is little knowledge concerning long-term recovery of the adaptive immune system. Here we studied the long-term reconstitution of the adaptive immune system in 31 children treated for haematological malignancies between July 2000 and October 2006. We performed detailed phenotypical and functional analyses of the various B and T cell subpopulations until 5 years after chemotherapy. We show that recovery of newly-developed transitional B cells and naive B and T cells occurred rapidly, within months, whereas recovery of the different memory B and T cell subpopulations was slower and incomplete, even after 5 years post-chemotherapy. The speed of B and T cell recovery was age-independent, despite a significant contribution of the thymus to T cell recovery. Plasmablast B cell levels remained above normal and immunoglobulin levels normalised within 1 week. Functional T cell responses were normal, even within the first year post-chemotherapy. This study shows that after intensive chemotherapy for haematological malignancies in children, numbers of several memory B and T cell subpopulations were decreased on the long term, while functional T cell responses were not compromised.

Modern intensive chemotherapy for childhood haematological malignancies has led to cure rates approaching 90% (Pui et al, 2009; Veerman et al, 2009), and is accompanied by extensive damage to the immune system (Lehrnbecher et al, 2008; Eyrich et al, 2009) and infectious morbidity during chemotherapy (Te Poele et al, 2007; van Tilburg et al, in press). It is important that immune reconstitution is accomplished rapidly following cessation of chemotherapy. However, it is largely unknown to what extent the cellular and humoral immune system recover after the currently-used more intensive chemotherapy regimens, especially in the long run.

Successive studies on the reconstitution of the B cell compartment after cessation of chemotherapy in children showed a fast recovery of immunoglobulin levels (de Vaan et al, 1982; Alanko et al, 1992; Ek et al, 2005; Eyrich et al, 2009) and B cell numbers recovered within months to normal levels (Alanko et al, 1992; Ek et al, 2005; Eyrich et al, 2009; Lehrnbecher et al, 2009). We showed, however, that levels of several B cell memory subpopulations were still reduced even 1 year after cessation of chemotherapy (van Tilburg et al. 2010). It remains unclear to what extent and at which rate these subpopulations recover in the long run.

The speed of T cell reconstitution after lymphocyte depletion is thought to be strongly dependent on the ability of the thymus to produce new naive T cells (Mackall et al, 1995; Heitger et al, 1997; Haynes et al, 2000; Hakim et al, 2005). Since thymic T cell production has been shown to decline with age (Bertho et al, 1997; Jamieson et al, 1999; Poulin et al, 1999), it is not surprising that children are able to reconstitute the naive T cell compartment faster than adults after intensive cytotoxic chemotherapy (Mackall et al, 1995), stem cell transplantation (SCT) (Weinberg et al, 1995) or antiretroviral treatment for HIV-1 infection (Van Rossum et al, 2001; Cohen Stuart et al, 2002; Vrisekoop et al, 2008). In healthy children, both thymic production and peripheral proliferation of naive T cells contribute to the establishment of the naive T cell compartment (Schonland et al, 2003; Hazenberg et al, 2004; Bains et al, 2009). To what extent both mechanisms contribute to naive T cell recovery in children post-chemotherapy, and whether the speed of recovery in children is age-dependent remains unclear. Memory CD4+ T cell recovery in adults post-chemotherapy has been shown to be faster than naive CD4+ T cell recovery, although even memory CD4+ T cell numbers did not normalise within the 1st year post-treatment (Hakim et al, 1997). Whether memory T cell recovery in children is equally hampered after cessation of chemotherapy has not been determined. If so, it may consequently compromise T cell functionality.

In the present study, we evaluated the recovery of B and T cell compartments in 31 children longitudinally until 5 years after cessation of chemotherapy for a haematological malignancy. We measured the numbers of the various B and T cell subpopulations, immunoglobulin levels and proliferative T cell responses to a variety of antigens and related the results to reference values for age. Collectively, this study provides a detailed quantitative and qualitative evaluation of the short- and long-term recovery of the adaptive immune system in children treated for haematological malignancies.

Patients and methods

Study population

Thirty-one children treated for haematological malignancies between July 2000 and October 2006 were prospectively analysed. Acute lymphoblastic leukaemia (ALL) patients were treated according to the Dutch Childhood Oncology (DCOG) ALL-9 protocol (Veerman et al, 2009), infant ALL patients in the Interfant 99 trial (van der Linden et al, 2009), acute myeloid leukaemia (AML) patients in the MRC AML12 trial (Burnett et al, 2010), Non-Hodgkin lymphoma (NHL) patients in the NHL 94 trial (adapted from the LMB89 (Patte et al, 2001) and the NHL-BFM (Neth et al, 2000) trials) and Hodgkin lymphoma (HL) patients in the GPOH-HD-95 trial (Schellong et al, 2005; Dorffel et al, 2003). Peripheral blood was obtained 1 to 8 times during follow up. Peripheral blood was drawn from 10 healthy children (age range 4–18 years) to serve as healthy controls in the study of functional T cell responses. The study was approved by the local medical ethical committee and written informed consent was obtained from all study participants or their legal guardians.

Population definitions and flow cytometry

B and T cell subpopulations were defined as described before (van Gent et al, 2009). In short, B cells were defined as CD19+ cells within the total lymphocyte population. Transitional B cells were defined as CD38+IgD+CD10+. Naive B cells and transitional B cells together constitute the CD27IgM+IgD+ B cell population, and naive B cell levels were determined by subtracting transitional B cell levels from the total CD27IgM+IgD+ population. Four different class switched memory B cell subpopulations were identified: CD27+IgG+, CD27+IgA+, CD27IgG+ and CD27IgA+ memory B cells. Non-class switched IgM+ memory B cells were defined as CD27+IgM+IgD+ and the small proportion of IgM-only memory B cells as CD27+IgM+IgD-. Plasmablast B cells were defined as CD38+IgD B cells.

Naive (CD27+CD45RO), memory (CD45RO+) and effector (CD27CD45RO, for the CD8+ T cell compartment only) subsets were identified within the CD4+ and CD8+ T cell compartments (van Gent et al, 2009). Activated CD4+ and CD8+ T cells were identified by combined expression of CD38 and HLA-DR. The level of proliferation within each of the T cell subsets was determined by measuring Ki67 expression (Hazenberg et al, 2000b).

Staining and analyses of B and T cell subpopulations by flow cytometry were performed as described before (van Gent et al, 2009). For B cell subsets, flow cytometry was performed within 24 h after blood was withdrawn. Characterization of the T cell compartment was performed on thawed cryopreserved peripheral blood mononuclear cells (PBMC). Cryopreservation was performed using a computerised freezing device that results in optimal quality of frozen cells (Noorloos et al, 1980). Absolute lymphocyte numbers were determined with a Cell-Dyn SapphireTM Haematology Analyzer (Abbott Diagnostics, Abbott Parc, IL, USA) and were used to calculate numbers of B and T cells and their subsets.

Immunoglobulins

Plasma was isolated and frozen in aliquots at −20°C until testing. Immunoglobulin concentrations were determined by nephelometry (Immage® 800 immunochemistry system, Beckman Coulter, Brea, CA, USA). Reference values for age were used for evaluating the data (de Vries et al, 2000).

TREC analysis

To measure the total number of TRECs and TREC content of CD4+ and CD8+ T cells, subsets were purified from thawed PBMC by magnetic-bead separation using the MiniMACS multisort kit according to manufacturer’s instructions (Miltenyi Biotec Inc, Bergisch Gladbach, Germany).

DNA was isolated using the QIAamp Blood Kit according to manufacturer’s instructions (Qiagen, Hilden, Germany). Signal joint T cell receptor excision circle (TREC) numbers were quantified by real-time polymerase chain reaction (PCR) as described previously (Hazenberg et al, 2000a). TREC content per T cell was calculated by dividing the TREC content per μg DNA by 150 000 (assuming that 1 μg DNA corresponds with 150 000 T cells). Historical TREC data from 40 healthy controls (age range: 4–18 years) (van Gent et al, 2009) served as control values.

Functional responses to antigens

For 7 patients within 1 year after cessation of chemotherapy and for 17 patients later than 1 year after cessation of chemotherapy sufficient material was available to analyse their proliferative response to a variety of antigens. In short, PBMC were labelled with 1·25 μmol/l carboxyfluorescein succinimidyl ester (CFSE) (Sigma) for 8’ at room temperature. After quenching the reaction with 2 ml cold 100% human-pooled serum (HPS) for 1’, cells were washed twice with phosphate buffered saline. PBMC were resuspended in RPMI1640 containing 10% HPS and 1% penicillin/streptomycin (Invitrogen, Breda, The Netherlands), distributed to round-bottom 96-wells plates (200 000 cells per well) and stimulated with the indicated antigens: cytomegalovirus (CMV) lysate, Herpes simplex virus 1 (HSV-1) lysate, Varicella zoster virus (VZV) (Microbix Biosystems, Ontario, Canada), Candida albicans (HAL Allergy, Leiden, Netherlands), tetanus toxoid (TT) or diphtheria (NVI, Bilthoven, Netherlands). After 6 d of culture, the percentage of divided CD3+ cells (as determined by CFSE dilution) was determined by flow cytometry. When the fraction of divided cells was at least three times higher than background levels (cells cultured without addition of stimuli) and exceeded 10%, cultures with respective antigen were scored positive for the presence of functional antigen specific cells. Prior to chemotherapy treatment, patients in the study had been vaccinated according to the Dutch National Immunisation Program (van Lier et al, 2009). Ten age-matched children served as healthy controls.

Statistical analyses

Absolute counts were expressed as a percentage of the median normal reference value for the patients’ age at the time of blood sampling (the reference value was set to 100%) (van Gent et al, 2009). Recovery of T and B cell subset counts was analysed by using a one-sample t test. In the case of non-normally distributed data, the data were logarithmically transformed. Fisher’s Exact Test was used to analyse responses of PBMC from patients and healthy controls following exposure to antigens. Statistical analyses were performed using the Statistical Package of Social Sciences version 15.0 (SPSS Inc, Chicago, IL, USA). A P value of <0·05 was considered statistically significant.

Results

Patients

Thirty-one children were included in the study (Table I). No relapses or serious infectious complications occurred during the study period. The time point of inclusion ranged from the day after, up to 3·2 years after chemotherapy cessation.

Table I.   Patient characteristics.
  1. NHR, non high risk; HR, high risk.

Age, median years (range)8·6 (3·7–16·3)
Sex (male/female)19/12
First sampling, median years after stop chemotherapy (range)1·3 (0·0–3·2)
Last sampling, median years after stop chemotherapy (range)3·5 (0·0–5·1)
No. of samples per patient, median (range)3 (1–8)
ALL-NHR, n (%)15 (48)
ALL-HR, n (%)5 (16)
Infant ALL, n (%)1 (3)
AML, n (%)4 (13)
Hodgkin lymphoma, n (%)3 (10)
Hodgkin lymphoma relapse, n (%)1 (3)
Non-Hodgkin lymphoma, n (%)2 (6)

Reconstitution of lymphocytes and B cells

To take into account that lymphocyte numbers per μl blood change with age during childhood, counts of all subpopulations were expressed as a percentage of the median reference values for age, defined as 100% (van Gent et al, 2009). Total lymphocyte numbers were decreased after chemotherapy cessation but recovered within 3–6 months post-treatment (Fig 1A, Table SI).

Figure 1.

 Reconstitution of lymphocytes and B cell subpopulations, expressed as a percentage of reference values for age. (A) Total lymphocytes; (B) B cells; (C) transitional B cells; (D) naive B cells; (E) CD27+IgG+ memory B cells; (F) CD27+IgA+ memory B cells; (G) CD27IgG+ memory B cells; (H) CD27IgA+ memory B cells; (I) IgM+ memory B cells; (J) IgM-only memory B cells; (K) plasmablast B cells. Boxplot: 2nd and 3rd quartiles with the median marked by the subdivision of the box. Whiskers extend to the most extreme values within ±1·5 interquartile ranges of the quartiles and more extreme values are plotted individually. X-axis: time after cessation of chemotherapy (months/year). *< 0·05; **< 0·01.

At chemotherapy cessation, all B cell subpopulations, except plasmablast B cells, were drastically reduced (Fig 1, Table SI). Total B cell numbers recovered to normal values around 3–6 months post-treatment (Fig 1B, Table SI). Transitional and naive B cells showed a recovery within the first 3 months post-treatment. (Fig 1C-D, Table SI). Remarkably, transitional B cells even tended to overshoot normal levels in the first months, which was most pronounced in young children (data not shown). In contrast, recovery of the different memory B cell subpopulations was slower (Fig 1E-K, Table SI). Numbers of CD27+IgA+ memory B cells and IgM only memory B cells increased to normal levels 1–1·5 years post-treatment, while CD27+IgG+ and IgM+ memory B cells recovered within 2·5–3 years. After an initial rapid recovery, CD27IgG+ and CD27IgA+ memory B cell counts remained at subnormal levels until the end of follow up. In contrast, plasmablast B cell numbers were not reduced after chemotherapy cessation and tended to be higher than normal levels during follow-up.

Changes in relative composition of the B cell compartment

Since the timing and the extent of reconstitution differed for the various subpopulations, the relative composition of the B cell compartment changed over time (Fig 2). At chemotherapy cessation transitional and naive B cells together comprised approximately 40% of the B cell compartment, compared to at least 60% in healthy children (van Gent et al, 2009), while memory and plasmablast B cells were most abundant within the B cell compartment. During follow-up, the naive B cell proportion increased, while the proportion of memory B cell subpopulations and plasmablast B cells gradually decreased.

Figure 2.

 Relative composition of the B cell compartment during reconstitution. X-axis: time after cessation of chemotherapy (months/year).

Immunoglobulins

To evaluate the functional capacity of memory B cells, IgG, IgA and IgM levels were determined. Levels were subnormal only in the first week post-treatment (data not shown). Thereafter, immunoglobulin levels were within normal ranges in all patients except in one patient treated for ALL with persistent low IgM levels.

CD4+ and CD8+ T cell reconstitution

To evaluate the degree of depletion and recovery of the T cell pool the total number of T cells and their subsets were assessed. At chemotherapy cessation, all CD4+ T cell subpopulations were drastically reduced (Fig 3A,C and E, Table SII). Within 3–6 months, total and naive CD4+ T cell numbers recovered to normal levels. Memory CD4+ T cell numbers increased in the first 3 months post-chemotherapy, but remained significantly reduced during follow-up.

Figure 3.

 Reconstitution of T cell subpopulations, expressed as a percentage of reference values for age. (A) CD4+ T cells; (B) CD8+ T cells; (C) naive CD4+ T cells; (D) naive CD8+ T cells; (E) memory CD4+ T cells; (F) memory CD8+ T cells; (G) effector CD8+ T cells. Boxplot: 2nd and 3rd quartiles with the median marked by the subdivision of the box. Whiskers extend to the most extreme values within ±1·5 interquartile ranges of the quartiles and more extreme values are plotted individually. X-axis: time after cessation of chemotherapy (months/year). *< 0·05; **< 0·01.

Directly after chemotherapy cessation, the degree of depletion in the CD8+ T cell compartment was less than in the CD4+ T cell compartment (Fig 3B,D,F and G, Table SII). Numbers of total and naive CD8+ T cells normalised within 6 months post-treatment. Memory CD8+ T cell numbers remained low during reconstitution, while effector CD8+ T cells took at least 2 years to regain normal numbers.

Age did not have any influence on the speed of total CD4+ and CD8+ T cell recovery (data not shown).

Short-term T cell reconstitution

To address the role of thymic output in naive T cell recovery, we measured total CD4+ T cell TREC numbers directly post-treatment (Fig 4A). Total CD4+ T cell TREC numbers per μl blood increased concomitant with naive CD4+ T cell numbers, suggesting that the thymus contributed to naive T cell recovery. Age did not influence the contribution of thymic output to naive T cell recovery, as the CD4+ T cell TREC number per μl blood and TREC content per CD4+ T cell increased with similar kinetics in all patients (Fig 4D,E).

Figure 4.

 Thymic output, proliferation, activation and functional responses: (A) Representative profiles of the recovery of naive CD4+ T cell counts and CD4+ T cell TREC numbers per L blood in two individuals after cessation of chemotherapy; (B) relation between the fraction of Ki67+ cells in the naive CD4+ T cells and naive CD4+ T cell counts; (C) relation between the fraction of Ki67+ cells in the naive CD4+ T cells and the percentage of activated (CD38+HLA-DR+) cells in the CD8+ T cells; (D) longitudinal analysis of TREC content per CD4+ T cell; (E) CD4+ T cell TREC numbers per L blood; (F) longitudinal analysis of Ki67 expression of naive CD4+ T cells and, (G) percentage of individuals responding to antigenic stimuli among patients and healthy age-matched controls. (A) black symbols = naive CD4+ T cell counts; open symbols = number of TRECs per L blood in CD4+ T cells; vertical dashed lines indicate time point of cessation of chemotherapy. (B,C) n = 7, monitored directly after cessation of chemotherapy; each symbol represents a single individual at given time; dashed lines represent the 95th percentile of normal values (van Gent et al, 2009). (D,E) lines connect data points from a single individual; HC = healthy control values (grey symbols) (van Gent et al, 2009). (F) lines connect data points from a single individual; dashed line represents 95th percentile of normal values (van Gent et al, 2009).

To address whether T cell depletion induced peripheral T cell proliferation, expression of the proliferation marker Ki67 in naive T cells was analysed from the seven individuals monitored directly after chemotherapy cessation. Naive CD4+ T cell proliferation levels were highest directly post-treatment (Fig 4B, circles), but not all individuals with a severely depleted naive CD4+ T cell compartment demonstrated increased proliferation levels. Two to 3 months post-treatment (Fig 4B, squares), naive T cell proliferation levels were only moderately increased, while naive CD4+ T cell counts had hardly restored. When naive CD4+ T cell counts had almost normalised at 6 months post-treatment, proliferation levels were within normal range (Fig 4B, triangles).

Increased naive T cell proliferation levels can also reflect episodes of T cell activation. We measured CD38+HLA-DR+ expression on CD8+ T cells as indication for the in vivo level of T cell activation. Elevated levels of naive CD4+ proliferation coincided with expression of CD38+HLA-DR+ on CD8+ T cells, hence with episodes of T cell activation (Fig 4C).

Long-term T cell reconstitution

We evaluated the relative contribution of thymic output and peripheral proliferation to maintain T cell numbers during the 1st 5 years post-chemotherapy. After an initial rise after chemotherapy cessation, CD4+ (Fig 4D,E) and CD8+ (data not shown) T cell TREC contents and numbers remained stable and within the normal range. Proliferation of naive CD4+ (Fig 4F) and naive CD8+ T cells (data not shown) also returned to normal levels long-term after chemotherapy. Collectively, these data suggest that chemotherapy did not severely affect naive T cell generating mechanisms in the long run.

Functional T cell responses

As a measure for T cell functionality the proliferative capacity of T cells to common antigens such as CMV, HSV-1, VZV, Candida, TT and diphtheria was measured post treatment. Frequencies of patients responding to these antigens were similar to those of healthy controls, both within and after the first year post-treatment (Fig 4G), implying that functional T cell responses were normal.

Discussion

In this study, we showed a fast recovery of naive B and T cell numbers in children after cessation of chemotherapy for haematological malignancies. The memory subpopulations of both lymphocyte compartments recovered more slowly or even remained decreased long-term post-chemotherapy. Nevertheless, there was no evidence for functional impairment.

A considerable inter-individual variation was observed for some lymphocyte subpopulations, especially in the very dynamic first months. Despite these differences, clear reconstitution patterns were observed. At chemotherapy cessation, transitional and naive B cells were severely affected. Because transitional B cells emerge directly from the bone marrow (Sims et al, 2005), these findings point towards severe bone marrow suppression during chemotherapy. In line with the described elevated precursor B cell regeneration in bone marrow post-treatment (van Wering et al, 2000; van Lochem et al, 2000), transitional B cell numbers recovered rapidly post-treatment, and even tended to overshoot normal levels, followed by naive B cell recovery. At this time point, the B cell compartment resembles the B cell compartment as described in young healthy children (van Gent et al, 2009).

In contrast, memory B cell subpopulations dominated the B cell pool at chemotherapy cessation while their reconstitution was much slower. We earlier showed that total CD27+IgG+/IgA+ memory B cell numbers had not completely recovered 1 year post-treatment (van Tilburg et al. 2010). Here we show that CD27+IgG+ memory B cells are largely responsible for this incomplete recovery although these cells eventually recovered. While in our previous study IgM+ memory B cell and IgM only memory B cell levels were still subnormal 1 year post-treatment we here show that they eventually recovered as well. However, for CD27IgG+ and CD27IgA+ memory B cells no complete recovery was observed. A plasmablast predominance was observed already during chemotherapy treatment and we now show that this increase persisted after chemotherapy cessation. These high plasmablast B cell levels may be responsible for a fast recovery of plasma cells, which would explain the rapid recovery of immunoglobulins and the reported good response to revaccinations shortly post-treatment (van Tilburg et al, 2006; Patel et al, 2007).

Naive T cell populations restored within 6 months in our and other studies (Alanko et al, 1994; Eyrich et al, 2009), but full naive CD4+ T cell recovery has been shown to take 6–12 months in studies in which CD4+ T cell counts were more severely affected (Mackall et al, 1995, 1997; Ek et al, 2005). In adults memory CD4+ T cell counts rise quickly after chemotherapy cessation (Hakim et al, 1997) or following SCT (Dumont-Girard et al, 1998; Roux et al, 2000) but do not recover completely 1 year post-treatment. In children, we observed that memory T cells did not fully recover during the entire 5-year follow-up. Consequently, effective T cell reconstitution in children, irrespective of their age, resets the naive/memory composition of the T cell compartment to a composition normally present at a younger age. Nonetheless, functional responses to antigens were not affected, indicating that sufficient numbers of antigen-specific memory T cells had survived or reconstituted.

The speed of naive T cell recovery monitored directly post-chemotherapy was similar for all children and independent of age, which is surprising given that the thymus contributed to T-cell reconstitution, and thymic output declines with age (Bertho et al, 1997; Jamieson et al, 1999; Poulin et al, 1999). Apparently, also later in childhood thymopoiesis is still sufficient to repopulate the naive T cell compartment within months. New naive T cells were also produced by proliferation in the periphery. In mice, homeostatic proliferation which is supposedly emptiness-driven, has been shown to be slow and dependent on the concentration of IL-7 in the microenvironment (Min et al, 2005). The moderately elevated proliferation levels observed here seemed to reflect activation induced proliferation and not a homeostatic response. In concert with the finding that naive T cell proliferation levels in HIV-1 patients normalise after initiation of HAART when naive T cell numbers have not yet recovered (Hazenberg et al, 2000b; Vrisekoop et al, 2008), it remains questionable whether, in humans, naive T cell proliferation is homeostatically elevated in lymphopenic settings.

In conclusion, in children after chemotherapy cessation both B and T cell compartments showed deficiencies, but functional recovery was not affected. The speed of recovery was age-independent. Further studies after intensive chemotherapy are needed to explore the consequences of the altered naive/memory distribution for immunity during ageing. The mechanism behind and the role of the persistent higher numbers of plasmablast B cells has to be elucidated, ideally in concert with revaccination studies.

Acknowledgements

The authors would like to thank all patients and their parents for kindly participating in this study. In addition, we would like to thank Sharda Mahes and Miriam Groeneveld for technical assistance.

Conflict of interest

These authors declare no conflict of interest.

Source of funding

This work was financially supported by KiKa foundation (Children Free of Cancer foundation), and by the Research Council for Earth and Life Sciences (ALW) with financial aid from the Netherlands Organisation for Scientific Research (NWO), grant 836·07·002 to J.A.M. Borghans.

Ancillary