Immune reconstitution in children following chemotherapy for acute leukemia

Abstract Although survival rates for pediatric acute lymphoblastic leukemia are now excellent, this is at the expense of prolonged chemotherapy regimens. We report the long‐term immune effects in children treated according to the UK Medical Research Council UKALL 2003 protocol. Peripheral blood lymphocyte subsets and immunoglobulin levels were studied in 116 participants, at six time points, during and for 18‐month following treatment, with 30‐39 patients analyzed at each time point. Total lymphocytes were reduced during maintenance chemotherapy and remained low 18 months following treatment completion. CD4 T cells remained significantly reduced 18 months after treatment, but CD8 cells and natural killer cells recovered to normal values. The fall in naïve B‐cell numbers during maintenance was most marked, but numbers recovered rapidly after cessation of treatment. Memory B cells, particularly nonclass‐switched memory B cells, remained below normal levels 18 months following treatment. All immunoglobulin subclasses were reduced during treatment compared to normal values, with IgM levels most affected. This study demonstrates that immune reconstitution differs between lymphocyte compartments. Although total B‐cell numbers recover rapidly, disruption of memory/naïve balance persists and T‐cell compartment persist at 18 months. This highlights the impact of modern chemotherapy regimens on immunity, and thus, infectious susceptibility and response to immunization.


INTRODUCTION
2011, the majority of pediatric patients in the United Kingdom with ALL were recruited to the MRC UKALL 2003 trial [4]. This protocol, similar to other treatment regimens internationally, entailed 6-12 months of relatively intensive blocks of chemotherapy, followed by maintenance chemotherapy (oral 6-mercaptopurine and methotrexate and four weekly vincristine and steroid pulses) for the remainder of the treatment period. Treatment was stratified according to conventional clinical, cytogenetic, and morphological response criteria, with three treatment regimens (A, B, and C), of increasing intensity.
There have been a number of studies that have reported the immune effects of ALL treatment, but few have comprehensively examined the effects of impact of modern chemotherapy regimens and characterized the immune recovery following cessation of treatment. During the first few months of treatment, children experience significant neutropenia, but this is less common during maintenance chemotherapy [5]. However, lymphopenia, with low levels of B and T cells is common, and is reported to persist for up to 6 months after treatment [6,7]. B cells have been reported to be more profoundly affected than T cells, with naive B-cell numbers falling proportionately more than memory B-cell populations [8][9][10]. After treatment, variable rates of reconstitution of B-cell subpopulations have been reported, with normal counts documented between 3 and 18 months in different studies [5,[9][10][11][12][13][14]. Serum levels of immunoglobulin fall during therapy and loss of protective levels of some specific antibodies in previously immunized children are seen [11,15]. Immunoglobulin levels have been reported to remain low for up to a year after completion of therapy [13]. The reported effects of chemotherapy on T-cell populations are less consistent, but with more significant effects reported on CD4 + T cells and relative modest effects on CD8 + T-cell numbers [7][8][9]. Reports on the effects on natural killer (NK) cells are limited and inconsistent [16,17].
The risk of infection following chemotherapy reflects both loss of pre-existing immunity (including vaccine immunity) as well as inability to mount new immune responses. Dissecting out the relative importance of these effects is important in determining strategies for reimmunization. It has been reported that children demonstrate adequate responses to reimmunization with booster vaccines 6 months following completion of chemotherapy [18]; and this is current UK practice [19]. However, the timing of reimmunization in these children is largely historical, and it may be that immunization sooner after treatment may be possible, potentially restoring vaccine-specific immunity earlier.
Here, we describe the immune function of these children, during maintenance chemotherapy and after treatment, to characterize the effects of current ALL treatment regimens. We performed a prospec-    Patient demographics from samples taken at "early" maintenance (approximately 6 months after completion of delayed intensification), "late" maintenance (approximately 18 months after completion of delayed intensification), at the end of treatment (4 weeks from last dose of oral chemotherapy) and 6, 12, and 18 months following completion of treatment.

Statistical analysis
Absolute counts for total lymphocytes, total B cells, naïve B cells,

unswitched memory B cells, switched memory B cells, CD4 T cells, CD8
T cells, and NK cells were assessed for normality at each time point, using a combination of tests (ie, the Anderson-Darling, Kolmogorov-Smirnov, and Cramer-von Mises tests) along with the graphical representation of the data. Wilcoxon signed-rank tests for non-normal data and paired t-tests for normal data were performed, for each time point to assess if the counts differ from those of healthy children (median normal reference value for patients age [22,23]). For graphical presentation, these counts were expressed as a percentage of median normal reference value for patients' age [22,23] (NB. for each patient the following was calculated absolute count divided by median count for normal patients of this age × 100 − a patient with a normal cell count for their age would have a value of 100%), and summarized using Box-and-Whisker plots by time.
Total lymphocytes and total B cells values (as a %, as calculated above) were summarized using bar graphs and compared using Kruskal-Wallis tests (for non-normal data) and analysis of variance (for normal data) to test for differences among different chemotherapy regimens and between boys and girls.
A two-sided P-value of <0.05 was deemed statistically significant for all analyses.
Immunoglobulin levels and IgG subclasses were summarized using bar graphs by time and by calculating the percentage of children's results which were within normal ranges for their age [24].
Statistical software SAS 9.4 was used for the analyses.

Patient demographics
A total of 116 patients had at least one blood sample value avail-

Lymphocyte compartment
In order to assess the effects of chemotherapy on each lymphocytes subset population, absolute numbers were compared to published data for healthy pediatric age groups and expressed as a percentage of median reference range for the relevant age group (as

F I G U R E 2 Naïve and memory B-cell counts during and after treatment. Boxplot showing naïve B cells (blue) and class switched (CD27+IgM/D-, red) and nonclass switched (CD27+IgM/D+, green) memory B cells. Lines within the boxes represent medians and diamonds
represent means (red) and unstitched memory (green). Cell count as a percentage of the median of healthy children, at each time point relative to chemotherapy. Lines within the boxes represent medians and diamonds represent means

Immunoglobulins
Functionality of B cells, as indicated by circulating levels of immunoglobulins was also affected by chemotherapy (Supporting information Figure 1 and Table 1). Total IgG levels were outside the age-specific reference ranges in 52.6% and 43.2% of patients during early and late maintenance respectively, and were still low for 3.5% of children 18 months after completing treatment. Of the IgG subclasses, IgG1 and IgG2 were affected most (Supporting Information Figure 2 and Table 1). IgM levels were also significantly affected, and 13.8% of patients had levels persistently low levels 18 months after completion of treatment.

Natural killer cells
Similar to helper and cytotoxic T cells, NK cell counts were significantly reduced during maintenance chemotherapy to below 30% of normal values at both time points (Supporting information Figure 3). At end of treatment and during follow-up, cell counts steadily improved but were still significantly low 6 months following treatment cessation (P = 0.014). Circulating NK cell numbers were not significantly abnormal by 12 and 18 months after treatment (P = 0.526 and 0.254, respectively).

Effect of age on immune recovery
There did not appear to be any substantial differences in immune recovery between different age groups, although children aged over 10 appeared to show higher levels of immune recovery at the 18-month

Comparison of different chemotherapy regimens
The UKALL 2003 protocol includes treatment stratification based on established risk factors, such that patients receive one of three different regimens, A, B or C, with increasing intensity. Comparisons were made between the three chemotherapy regimes in the UKALL 2003 protocol for the total lymphocytes and B cells and no significant differences were observed (Figure 4).

Comparison of boys and girls
In view of the fact that boys receive significantly longer treatment than girls (3 years compared to 2 years maintenance therapy), comparisons were made between immune recovery in boys and girls. Significant differences in the immune recovery between boys and girls were observed only at late maintenance for both lymphocytes and B cells (P = 0.009 and 0.024, respectively) ( Figure 5), but levels following completion of treatment were similar.

Discussion
In this prospective study, we demonstrated the impact of a contemporary chemotherapy regimen for ALL on long-term immune reconstitution in a large pediatric patient cohort.
During maintenance chemotherapy, B-cell counts were severely suppressed, but exhibited relatively rapid recovery following treat- Although there is a considerable range of treatment intensity (Regimen A, B, and C), we did not see any significant impact of this on the degree or duration of immunosuppression. However, much of the intensity variation may affect myelosuppression, and duration of neutropenia, rather than immunosuppression and lymphopenia. Therefore, a reduction in the exposure to more immunosuppressive agents (eg, by removing corticosteroids from maintenance treatment) may have an impact on immunosuppression and infection rates. Indeed, the Dutch Oncology Group has reported that reduction in intensity of maintenance treatment was associated with reduced infectious complications [28]. The data presented here will provide a comparator for future studies to assess the impact of such reductions. Similarly, we did not see any significant difference between the degree of immunosuppression in boys and girls, despite the significantly longer treatment duration in boys. This suggests that the duration of maintenance therapy may be less important than the actual drugs received, in terms of the degree of suppression and time for recovery.
Supportive care for reducing infection-related mortality is critical [25]. All patients on the UKALL 2003 trial for example, received pneumocystis prophylaxis and no mortalities from this infection were identified. The risk of some other specific infections is also reported to be high, with one study describing the relative risk of invasive pneumococcal infection to be 50-fold that of healthy children [29]. No prophylactic measures are taken to reduce such infections. Options for this would include regular antimicrobial prophylaxis or immunization. Immunization during treatment is an attractive option, given the duration of risk, but may not be feasible given the degree of B-cell suppression observed in this study, even with highly immunogenic conjugate vaccines. However, the rapid recovery in naïve B cells observed following cessation of treatment may suggest that very early (eg, 1 month following completion of treatment) immunization may be possible, rather than the conventional recommendation of 6 months later. This is supported by the results of our study of immunization with PCV-13 in children with ALL, from which these current samples were obtained, that demonstrated protective levels of immunity could be achieved by immunization 1 month following maintenance treatment [20].
In summary, we have clearly demonstrated that the immune system is significantly impaired during treatment during UKALL 2003 therapy and that full recovery may takes up to at least 18 months following treatment. Although the B-cell compartment is particularly severely affected by chemotherapy, there is a rapid rebound proliferative increase in naïve B cells after treatment. This, together with data we have recently published demonstrating protective responses to PCV-13 immunization given 1 month after cessation of treatment in this population, suggests that early reimmunization may be feasible [20].

ACKNOWLEDGMENTS
The authors would like to thank the University of Southampton CRUK NIHR Clinical Trials Unit for study support. This work was financially supported primarily by the NIHR Research for Patient Benefit Programme with additional financial support (vaccine supply) from Pfizer (formally Wyeth).

Author's contribution
JG was Chief Investigator for the clinical trial. Study concept, trial design, and funding application were conducted by AW, JB, JC, SC, SF, PH, and JG. Trial conduct and management was overseen by ED and JG. Patient recruitment, blood sampling, and data collection were conducted by SM. RB, TM, and AG conducted statistical analysis. The manuscript was primarily written by AW, RB, and JG; all authors reviewed final manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.