Contemporary analysis of functional immune recovery to opportunistic and vaccine‐preventable infections after allogeneic haemopoietic stem cell transplantation

Abstract Objectives Infections are a major cause of mortality after allogeneic haemopoietic stem cell transplantation (alloHSCT), and immune recovery is necessary for prevention. Novel transplant procedures have changed the epidemiology of infections but contemporary data on functional immune recovery are limited. In this pilot study, we aimed to measure immune recovery in the current era of alloHSCT. Methods Twenty, 13, 11, 9 and 9 alloHSCT recipients had blood collected at baseline (time of conditioning) and 3‐, 6‐, 9‐, and 12‐months post‐alloHSCT, respectively. Clinical data were collected, and immune recovery was measured using immunophenotyping, lymphocyte proliferation, cytokine analysis and antibody isotyping. Results Median absolute T‐ and B‐cell counts were below normal from baseline until 9‐ to 12‐months post‐alloHSCT. Median absolute CD4+ T‐cell counts recovered at 12‐months post‐alloHSCT. Positive proliferative responses to Aspergillus, cytomegalovirus (CMV), Epstein‐Barr virus (EBV), influenza and tetanus antigens were detected from 9 months. IL‐6 was the most abundant cytokine in cell cultures. In cultures stimulated with CMV, EBV, influenza and tetanus peptides, the CD4+ T‐cell count correlated with IL‐1β (P = 0.045) and CD8+ T‐cell count with IFNγ (P = 0.013) and IL‐1β (P = 0.012). The NK‐cell count correlated with IL‐1β (P = 0.02) and IL‐17a (P = 0.03). Median serum levels of IgG1, IgG2 and IgG3 were normal while IgG4 and IgA were below normal range throughout follow‐up. Conclusions This pilot study demonstrates that immune recovery can be measured using CD4+ T‐cell counts, in vitro antigen stimulation and selected cytokines (IFNγ, IL‐1β, IL‐4, IL‐6, IL‐17, IL‐21, IL‐31) in alloHSCT recipients. While larger studies are required, monitoring immune recovery may have utility in predicting infection risk post‐alloHSCT.

Objectives. Infections are a major cause of mortality after allogeneic haemopoietic stem cell transplantation (alloHSCT), and immune recovery is necessary for prevention. Novel transplant procedures have changed the epidemiology of infections but contemporary data on functional immune recovery are limited. In this pilot study, we aimed to measure immune recovery in the current era of alloHSCT. Methods. Twenty, 13, 11, 9 and 9 alloHSCT recipients had blood collected at baseline (time of conditioning) and 3-, 6-, 9-, and 12-months post-alloHSCT, respectively. Clinical data were collected, and immune recovery was measured using immunophenotyping, lymphocyte proliferation, cytokine analysis and antibody isotyping. Results. Median absolute T-and B-cell counts were below normal from baseline until 9-to 12-months post-alloHSCT. Median absolute CD4 + T-cell counts recovered at 12-months post-alloHSCT. Positive proliferative responses to Aspergillus, cytomegalovirus (CMV), Epstein-Barr virus (EBV), influenza and tetanus antigens were detected from 9 months. IL-6 was the most abundant cytokine in cell cultures. In cultures stimulated with CMV, EBV, influenza and tetanus peptides, the CD4 + T-cell count correlated with IL-1b (P = 0.045) and CD8 + T-cell count with IFNc (P = 0.013) and IL-1b (P = 0.012). The NK-cell count correlated with IL-1b (P = 0.02) and IL-17a (P = 0.03). Median serum levels of IgG1, IgG2 and IgG3 were normal while IgG4 and IgA were below normal range throughout follow-up. Conclusions. This pilot study demonstrates that immune recovery can be measured using CD4 + T-cell counts, in vitro antigen stimulation and selected cytokines (IFNc, IL-1b, IL-4, IL-6, IL-17, IL-21, IL-31) in alloHSCT recipients. While larger studies are required, monitoring immune recovery may have utility in predicting infection risk post-alloHSCT.

INTRODUCTION
Allogeneic haemopoietic stem cell transplantation (alloHSCT) is routinely used to cure a wide variety of haematological malignancies. The immunosuppression, given during conditioning and for graft-versus-host disease (GVHD) prophylaxis, results in delayed immune reconstitution and places alloHSCT recipients at high risk for opportunistic and vaccinepreventable infections. [1][2][3][4] Infections are a leading cause of death in patients undergoing alloHSCT. 1,3 While antimicrobial prophylaxis is effective in reducing morbidity and mortality from infections post-alloHSCT, it has significant limitations including drug-related toxicity, emergence of resistance, breakthrough infections and high cost. 5,6 Currently, risk stratification models are used to guide the timing and duration of antimicrobial prophylaxis. 7 These are based on clinical factors such as age at time of alloHSCT, underlying disease, prior therapy, conditioning regimen, donor type, degree of matching, stem cell source and presence of GVHD. However, such clinically based models only give a qualitative measure of immune function and consequently lack precision. A strategy which quantifies immune function is urgently needed to more accurately predict risk and guide antimicrobial prophylaxis accordingly. Immune-based strategies for cytomegalovirus (CMV) have been examined and appear promising for guiding CMV prophylaxis; 8 however, more broad-based immune function strategies incorporating other important opportunistic pathogens, such as Aspergillus still need to be developed.
The alloHSCT process results in loss of immune memory accumulated from previous vaccinations, and all recipients need to be re-vaccinated post-alloHSCT. 9,10 The new naive T and B cells develop from donor stem cells and require stimulation with vaccine antigens for long-term protection. For vaccination to be useful post-alloHSCT, it must occur at a time when the immune system has adequate function to generate a protective response. Therefore, the optimal timing of vaccination becomes a critical balance between obtaining a protective immune response as early as possible to prevent infections and delaying it until functionally effective immune responses can be generated. Current post-alloHSCT vaccination strategies are based on fixed schedules. 11 However, it is becoming evident that the timing of vaccination would be more appropriately based on each patient's capacity to respond to vaccine antigens. Evolving data suggest that such immune responses can be measured. [12][13][14] However, a more detailed analysis is required prior to developing novel vaccine schedules to better guide effective vaccination post-alloHSCT.
Assays to measure immune function are available including immune cell counts, 15 subpopulations of natural killer (NK) cells, 16 composition of memory T-cell compartments, 17 cytokine profiling 18 and cellular proliferation measurement. 19 Studies performed to examine immune reconstitution post-alloHSCT are limited by contemporary relevance, number of immune markers and number of pathogens examined, or correlation with clinical outcomes. While many studies have provided evidence for the importance of CD8 + T cell-mediated viral-specific immune recovery post-alloHSCT, 20 the recovery of CD4 + T-cell function is less well understood.
The aim of this study was to perform a contemporary and comprehensive examination of immune reconstitution post-alloHSCT including CD4 + T-cell function and cytokine profiling.

Patient characteristics and clinical outcomes
The baseline characteristics are shown in Table 1. Acute myeloid leukaemia was the most common indication for transplantation (5/20; 25%), and nine patients (45%) received a reduced intensity conditioning (RIC) regimen (Table 1).
The recovery of percent fractions of each cell subset was also analysed and T-and B-cell percent fractions were seen to recover earlier than absolute cell numbers. The median percent T-cell values recovered to normal levels 3-to 6-months post-alloHSCT while median percent B-cell levels recovered at 6-to 9-months post-alloHSCT (Supplementary figure 1).
Median absolute NK cells and monocytes were at the lower limit of normal at baseline (0.08 9 10 9 L À1 and 0.2 9 10 9 L À1 , respectively) but recovered to normal levels at the 3-month time-point while the median absolute neutrophil counts did not recover to normal range until the 6-month time-point post-alloHSCT (Figure 1d-f).
The median absolute count of CD4 + T cells was below the lower limit of normal until the 12month time-point (0.32 9 10 9 L À1 ), while median absolute CD8 + T-cell counts recovered to normal levels 3-to 6-months post-alloHSCT (Figures 2a  and 3b). Patients 7, 11 and 17 who had high absolute CD8 + T-cell counts after the 6-month time-point had multiple infections, and patient 17 also had extensive chronic GVHD ( Figure 2b; Table 2). The median absolute CD4 + T-cell counts in RIC-alloHSCT recipients remained < 0.2 9 10 9 L À1 (0.185 9 10 9 L À1 ) at the 12-month time-point. Low numbers of CD4 -CD8double negative T cells, not usually present in the periphery of healthy individuals, were seen in this cohort of alloHSCT recipients (Figure 2c). The CD4 + :CD8 + T-cell ratio was close to normal (1.7:1) at baseline but reversed during the 12 months of follow-up ( Figure 2d).

Pathogen-specific PBMC cell responses
T-cell proliferative responses to Aspergillus lysate, tetanus-toxoid and a peptide mix containing MHC Class II binding peptides from CMV, Epstein Bar virus (EBV), tetanus and Influenza (CMV-EBV-Flu-Tet peptide pool) are shown in Figure 3a-d. The median SI for Aspergillus, tetanus-toxoid and the CMV-EBV-Flu-Tet peptide pool was < 2.5 (defined threshold) at baseline, indicating little to no capacity to proliferate. Positive proliferative responses were only observed 9-to 12-months  c Defined as an absolute neutrophil count > 0.5 9 10 9 L À1 sustained over 3 days.   (Figure 3b, c) were statistically significantly higher at 9-months (SI = 1.7 and 2.8, respectively) and 12months (SI = 3.7 and 2.7, respectively) post-alloHSCT as compared with baseline (SI = 1.5 and 1.3, respectively) but tetanus-toxoid-specific proliferation (Figure 3d) was not statistically significantly higher until the 12-month time-point (SI = 1.2 vs. 9.2; P = 0.016) post-alloHSCT.

Cytokine profile of pathogen-specific T cells
In the PBMC culture supernatants, cytokine levels were only analysed at the 9-and 12-month timepoints, as significant proliferation was only detected at these time-points. Median concentrations of cytokines secreted in response to Aspergillus, tetanus-toxoid and the CMV-EBV-Flu-Tet peptide pool are shown in Figure 4a-c. Cytokines representing Th-1 (IFN-c), Th-2 (IL-4) and Th-17 (IL-17, IL-21) subsets and others were detected indicating the presence of a broad T-cell response which was not skewed towards a particular phenotype. Very high levels of IL-6 (> 100 pg mL À1 ) were secreted in response to all antigens and peptides (Figure 4a-c). There was no significant difference between mean total cytokine levels between the 9-and 12-month time-points from Aspergillus (1943 vs. 1953 pg mL À1 ) and tetanus-toxoid (253 vs. 358 pg mL À1) -stimulated cultures (Figure 4a, b), but the mean total cytokine level in CMV-EBV-Flu-Tet peptide-stimulated cultures were significantly higher at 12 months as compared with 9 months (1742 vs. 655 pg mL À1 ; P = 0.008; Figure 4c).

Serum cytokines
The mean concentrations of serum cytokines (Figure 5a-d) were comparable with that reported for healthy donors in other studies. 19,22 IFN-c was not detected at any time-point while the median levels of IL-4 and IL-10 decreased to reach their nadir at 6-9 and 6 months, respectively, and increased thereafter (Figure 5a). Two spikes of the pro-inflammatory cytokine IL-6 were detected at 3-and 12-months post-alloHSCT ( Figure 5b). Median IL-17a and Il-17f levels were highest at 12 months (Figure 5c). IL-31 reached the nadir at 3 months whereas IL-23 and IL-33 declined over the 12 months of follow-up ( Figure 5d).

Antibody isotypes
Median serum levels of IgG1, IgG2, IgG3, IgG4, IgA and IgM are shown in Figure 6a-f. The median concentrations of IgG1, IgG2 and IgG3 were above the lower limit of normal over the 12month follow-up period (Figure 6a-c). 23 Patients showing IgG1 and IgG3 levels above the normal range had no infections diagnosed at the corresponding time-point. Median IgG1 (associated with Th-1 responses) concentrations increased over time and were highest 12-months post-alloHSCT (Figure 6a). Median concentrations of IgG4 (associated with Th-2 responses) remained at the lower limit of normal through the 12 months of follow-up (Figure 6d).
The median concentrations of IgA were below the lower limit of normal throughout the followup period (Figure 6e). 24 Patients who had high IgA levels at baseline and at 3-month post-alloHSCT (patients 8, 10, 13 and 20) had mucosal  Table 2). The median concentration of IgM was within the normal range throughout but highest at 9 months (Figure 6f). 24

DISCUSSION
In the modern era of alloHSCT, this proof of concept study comprehensively measures the recovery of immune cell subsets, pathogen-specific immune responses, cytokines and serum antibody isotypes. We report that recovery of functional CD4 + T-cell pathogen responses occurs 9-to 12 months post-alloHSCT and, based on antigenspecific in vitro responses, identify a panel of cytokines (IFN-c, IL-1b, IL-4, IL-6, IL-17, IL-21 and IL-31), with potential for immune-profiling and infection risk prediction.
As with other studies, we found that NK cell and monocyte recovery occurred before T-cell and B-cell recovery. 17,25 CD8 + T cells recovered earlier than CD4 + T cells, with the latter only reaching the reference range 21 12 months post-alloHSCT with reversal of the CD4 + /CD8 + ratio (Figure 2a,   b). 21 This finding is likely due to viral reactivation and more efficient peripheral homeostatic expansion of CD8 + T cells. 11 As CD4 + T cells play a central role in orchestrating all immune responses, CD4 + T-cell counts and function are critical to infection risk post-alloHSCT. Storek et al. 26 demonstrated that long-term infection morbidity is related to CD4 + T-cell reconstitution post-alloHSCT, and antigen-specific responses were reduced if CD4 + T-cell counts were low. Fukuda et al. 27 demonstrated similar rates of invasive aspergillosis especially late, in RIC-as compared with myeloablative-alloHSCT recipients. We have detected low CD4 + T-cell counts at 12 months in our RIC-alloHSCT recipients (median of 0.185 9 10 9 L À1 ) and have demonstrated that recovery of CD4 + T-cell count parallels Aspergillusspecific T-cell responses providing a potential explanation for the findings of Fukuda et al. 27 In the HIV population, a CD4 + T-cell count of < 0.2 9 10 9 L À1 is used as a biomarker in guiding prophylaxis of the fungus Pneumocystis jirovecii. 28 Regular measurement of CD4 + T-cell counts post-alloHSCT may have potential for guiding antimicrobial prophylaxis. We found that median absolute neutrophil counts were below the reference range until 6-months post-alloHSCT. Rommeley et al. 29 examined the recovery of neutrophil function in 30 alloHSCT recipients and demonstrated that 91.8% had recovered the capacity to phagocytose Escherichia coli but only 66.3% had recovered oxidative burst activity by Day 90 post-alloHSCT. Further prospective evaluation of functional neutrophil recovery post-alloHSCT is required.
As data on CD4 + T-cell functional recovery post-alloHSCT are limited, we assessed in vitro T-cell proliferation using PBMC assays. While T-cell subset percent fraction had returned to normal by 6 months, the recovery of absolute T-cell numbers was delayed until 9-months post-alloHSCT. The detection of pathogen-specific PBMC responses at 9-months post-alloHSCT ( Figure 3) aligned closely with recovery of absolute T-cell numbers (Figure 1). Lack of positive proliferative responses until 9-months post-alloHSCT is likely to be due to a combination of factors including low T-cell numbers, ongoing immunosuppressive therapy or its residual effects and limited T-cell repertoire diversity because of low thymic regeneration in adults post-alloHSCT. 30 Proliferative responses to phorbol-12-myristate-13-acetate-Ionomycin were present at all time-points (data not shown), indicating that signal transduction pathways were not affected by alloHSCT procedures. Current guidelines recommend that antifungal prophylaxis continues until Day 75 post-alloHSCT (if no GVHD). 31 However, our findings indicate that larger correlative studies should be performed to determine whether current guidelines should be revisited.
There are very limited data on Aspergillusspecific T-cell responses post-alloHSCT. Perruccio et al. 32 reported that Aspergillus-specific immune recovery occurred in T-cell depleted matched, T-cell depleted haploidentical and T-cell replete alloHSCT recipients at 7-9, 9-12 and 15-18 months, respectively. Beck et al. 33 demonstrated that no significant Aspergillus-specific T-cell responses were detected in the first 4-months post-alloHSCT (n = 15), and in comparison with healthy controls (n = 23), Aspergillus-specific T-cell responses were significantly lower up to 1-year post-alloHSCT. Characterisation of Aspergillus-specific T cells in humans has been challenging because of the low precursor frequency of fungal-specific T cells seen in healthy individuals, 34,35 and the delayed recovery of Aspergillus-specific responses we detected could be attributed to this. However, we found that tetanus-toxoid and CMV-EBV-Flu-Tet peptide pool responses recovered at similar timepoints post-alloHSCT. Also, the timing of recovery of Aspergillus-specific responses in our study was consistent with that of other studies. 32,33 Our findings indicate that recovery of Aspergillusspecific T-cell responses occur late post-alloHSCT and may explain the previously reported late cases of invasive aspergillosis. 36 The high tetanus-toxoid-specific proliferative responses seen at 12 months may be due to vaccination 4-6 months earlier (Supplementary  Higher circulating CD4 + T-cell counts have been reported to correlate with improved pneumococcal conjugate vaccine response. 14 Comparatively, the proliferative response to CMV-EBV-Flu-Tet peptide pool peptides is lower at 6 to 9 months which may be related to the dominance of CD8 + T cells seen in the CMV-EBV-Flu-Tet peptide pool cultures which have lower proliferative capacity than CD4 + T cells. Although the major histocompatibility complex Class II binding CMV-EBV-Flu-Tet peptide pool targets CD4 + T cells, it also contains crossreactive CD8 + T-cell epitopes within its sequence. Avetisyan et al. 12 reported that influenza T-cell responses were broad with multiple epitopes and proteins recognised. Haining et al. 37 showed, in vaccinated paediatric alloHSCT recipients, that CD4 + responses were stronger than CD8 + responses. Measurement of proliferative responses to both the tetanus-toxoid and CMV-EBV-Flu-Tet peptide pool may have value from 6-months post-alloHSCT to better define the full spectrum of vaccine responsiveness and the optimal timing for re-vaccination. The cytokine secretion pattern of PBMC cultures demonstrated a broad range, including those representing Th-1 (IFN-c), Th-2 (IL-4) and Th-17 (IL-17, IL-21) subsets (Figure 4a-c). The cytokine profiles differed between Aspergillus, tetanustoxoid and CMV-EBV-Flu-Tet peptide pool cultures. Further, the median cytokine concentrations were higher in Aspergillus and CMV-EBV-Flu-Tet peptide pool cultures than in tetanus-toxoid cultures (Figure 4). Nilsson et al. 38 stimulated donor T cells with viral antigens and found that IL-1b and IFN-c production was higher in grafts transplanted to recipients who had no post-alloHSCT infectious complications compared to those who did. While further assessment is required, the data presented indicate that cytokine profiles may have potential in infection risk stratification to individualise antimicrobial prophylaxis post-alloHSCT.
Even though median levels of proliferation was highest to tetanus-toxoid (12 months), median cytokine secretion was lower compared to Aspergillus cultures, and no correlations were seen between T-cell numbers and cytokine concentration in the tetanus-toxoid cultures. The strong cytokine response generated by Aspergillus lysate is consistent with a polyclonal T-cell response driven by a mix of proteins. Proliferation appears to be a better marker of tetanus-specific immunity than cytokines reflecting the tetanustoxoid-specific immune response which is predominantly antibody-mediated and driven by Th-2 rather than a cell-mediated immune response.
The correlations observed between T-cell and NK-cell numbers and cytokine concentrations in the cultures suggest different cell sources for the various cytokines. In the Aspergillus cultures, IL-6 levels correlated with NK-, CD4 + T-cell and CD8 + Tcell numbers, IL-25 with CD4 + and CD8 + T cells and IL-17, IL-31 and IL-33 with CD8 + T cells. CD4 + T cells and NK cells are known to be important drivers of fungal-specific immunity and Th-17 cells play a central role in Aspergillus clearance. 39,40 Consistent with this, we detected a cytokine profile characterised by a Th-17 driven Aspergillus response. IL-6 is produced by a wide variety of cells including fibroblasts, keratinocytes, endothelial cells, monocytes, macrophages and T cells. As monocytes are possible contributors for IL-6 detected in the PBMC cultures the correlations seen with T and NK cells could be indirect. The cytokine profile from the CMV-EBV-Flu-Tet peptide pool stimulated cultures indicates IFN-c dominance from both CD8 + T cells and NK cells and CD8 + T cells are known to play a critical role in viral immunity post-alloHSCT. 25,41 The CD8 + T-cell contribution observed in Aspergillusstimulated cultures may reflect an adjuvant effect driving bystander CD8 + T-cell activity in these CD4 + T-cell deficient patients. Indeed, many of the patients in our cohort had viral coinfection ( Table 2).
AlloHSCT is known to be associated with substantial changes in cytokine concentrations resulting from mucosal damage, consequent to the procedure. The proinflammatory cytokine IL-6 was the most abundant cytokine in our PBMC cultures (Figure 4a-c). IL-6 is known to play an important role in haematopoiesis and is required for initiation of Th-17 responses. 42 The cytokine profile is also impacted by the kinetics of cytokine secretion and as supernatant was taken at Day 4 from PBMC cultures, cytokines that are secreted earlier (IL-6, TNF-a, IL-1b, IFN-c) are likely to be preferentially detected while those that peak after Day 4 such as IL-17 and IL-22 may be underrepresented. 43 Because the serum cytokine levels detected were low in our cohort ( Figure 5) and are known to be highly variable even in healthy individuals, 19,22,44 we would not advocate their measurement in clinical practice.
Serum antibody isotyping in our patient cohort (Figure 6a-f) showed that the two most abundant subclasses, IgG1 and IgG2, were within the normal range even before B-cell numbers recovered, suggesting functional long-lived plasma cells. Patients who had high levels of IgG3 which is most efficient at complement activation and opsonisation did not have active infections at the time, implying a protective function. In contrast to the other subclasses, IgG4 which is noncomplement activating and known to be produced in response to allergens was low throughout the 12-month period. The stronger IgG1 and low IgG4 responses in this cohort maybe suggestive of a Th-1 bias post-HSCT. 45 Although high IgA levels were detected in patients with mucosal infections in the early time-points, median levels were below normal up to 12months post-alloHSCT. The low levels of some antibody isotypes at the 9-to 12-month timepoint are suggestive of defective Ig-class switching in this cohort. The spike in IgM concentration at 9-months post-alloHSCT may reflect vaccinations introduced around 6 to 8 months.
Our finding of high concentrations of IL-31 in PBMC cultures and serum is novel in the alloHSCT setting. IL-31 is produced by activated Th-2 type T cells, induces production of IL-6, 41 and has been shown to be associated with inflammatory disorders. IL-33, which is produced by necrotic and epithelial cells and has increased function under cellular stress, was seen predominantly in our Aspergillus antigen re-stimulation cultures. IL-33 is known to drive Th-2 responses and play a crucial role in allergy and lung-related inflammatory conditions. 44 Our novel findings in relation to IL-31 and IL-33 warrant further investigation in a larger cohort of alloHSCT recipients as these may serve as biomarkers that can be used for risk prediction of Aspergillus infections post-alloHSCT.
Limitations in the present study include the small sample size, restricted 15-plex cytokine analysis and the narrow antigen range used to assess immune responses. The small sample size prohibited us from effectively determining causeeffect associations, and the correlations observed need to be interpreted with caution. However, this was a pilot study designed to assess feasibility of detecting potential immune biomarkers for improved infection risk prediction post-alloHSCT. The findings provide a framework for larger cohort studies where immune responses to a broader range of vaccine antigens and opportunistic pathogens can systematically be measured along with extensive analysis of cytokine profiles. In addition, PBMC assays have many limitations including lengthy purification process and assay time, and cell loss from cryopreservation. The recently developed flowbased assay using whole blood has the advantage of combining immunophenotyping with detection of antigen-specific proliferation and cytokine secretion (FASCIA). This rapid and practical assay has demonstrated capacity to detect lymphocyte reactivity in primary immunodeficiency and HIV settings. 19,46 Of note, Nilsson et al. 38 used FASCIA to determine immune function in donor grafts to predict infection risk post-alloHSCT.
In conclusion, this proof of concept study indicates that functional immune reconstitution post-alloHSCT can be measured. We have identified CD4 + T-cell counts, antigen-specific T-cell proliferation and cytokine production (IFN-c, IL-1b, IL-4, IL-6, IL-17, IL-21 and IL-31), in response to a variety of opportunistic and vaccine-preventable pathogens, as having potential to be utilised as biomarkers for infection risk prediction. Larger multicentre studies are required for validation and to fully elucidate the clinical utility of these immune assays.

Patients
Twenty patients aged 18 years or older who underwent alloHSCT at Alfred Health, Melbourne, Australia, between February 2015 and January 2016 were enrolled and followed for 12 months or until death, if earlier. Written informed consent was obtained from all participants within 96 h of commencing conditioning chemotherapy. The study was approved by Alfred Health Human Research Ethics Committee (344.14). All procedures of alloHSCT including immunosuppression regimens, blood product support, and the investigation and treatment of transplant complications were as per standard Alfred Health protocols. All suspected infections were actively investigated using cultures (e.g. blood, urine), radiology (e.g. computed tomography scan), scopes (e.g. bronchoscopy, colonoscopy), tissue biopsy (e.g. open lung biopsy) and blood for molecular and serological testing (e.g. CMV, respiratory viruses, Legionella), as previously described. 47

Study definitions
Cytomegalovirus infection and disease were defined according to published criteria. 48 The European Organization for the Research and Treatment of Cancer/Mycoses Study Group criteria were used to classify invasive fungal disease invasive fungal disease. 49 Published definitions were used to classify bacterial infections and infection-related mortality. 50 Respiratory viral infections were defined as positive PCR results from respiratory specimens together with characteristic respiratory symptoms. 51 Polyomavirus and norovirus infections were defined as positive PCR from blood and disease as detection of virus by PCR or immunohistochemistry in bodily fluids or organ tissue together with symptoms and/or signs in the affected organ. 52,53 The guidelines of Sarmati et al. 54 were used to identify cases of Hepatitis B infection. Acute and chronic GVHD were classified according to the Glucksberg and the National Institute of Health criteria, respectively. 55,56

Sample preparation
Whole blood and serum were collected from 20 patients at baseline (within 96 h of commencement of conditioning). Follow-up samples were collected from 13, 11, 9 and 9 patients at 3-, 6-, 9-and 12-months post-alloHSCT, respectively. Peripheral blood mononuclear cells (PBMC) were purified by density gradient centrifugation with Ficoll-Paque (GE Healthcare, USA) according to standard procedures. Briefly, whole blood was diluted 1:1 in sterile PBS and 30 mL layered on to a 15 mL gradient of Ficoll-Paque and centrifuged for 20 min at 400 g. The cell interface layer was carefully harvested and washed twice in PBS (10 min at 500 g) and the pellet resuspended in RPMI 1640 with Glutamax (Invitrogen, USA) and supplemented with 10% (v/v) Human AB serum (Sigma-Aldrich, USA) and 100 U mL À1 Penicillin/Streptomycin solution (Invitrogen).

Flow cytometric analysis of cell subsets
The 8-colour antibody panel, CD3-FITC, CD19-APC, CD14-APC-H7, CD45-PerCP-Cy5.5, CD8-BV421, CD4-PE-Cy7, CD15-BV650 and CD16 + CD56-PE (BD Biosciences, USA), was used for immunophenotyping. Briefly, 100 lL of whole blood was stained with the antibody-conjugate panel, red cells lysed and washed according to manufacturer's instructions and analysed on a BD LSR II flow cytometer (BD Biosciences). Cells were gated using forward and side scatter, doublets excluded, CD45 + gate applied for leucocytes and > 100 000 events collected. Rainbow Calibration Particles with six peaks (BD Biosciences) were used for quality control. The absolute cell number for each subset was calculated by multiplying total leucocyte count for each patient by percent fraction of each cell subset obtained immunophenotyping.