Our prospective study included 106 pediatric patients from University Hospital La Paz, who received a liver graft between January 2019 and December 2023. All patients gave informed consent, approved by the ethics committee of our institution (reference PI-4000). Eleven patients were withdrawn from the study ( Figure 1 ), resulting in a final cohort of 95 patients. Transplant indication was categorized in five groups ( Table 1 ), according to Díaz Fernandez et al. (19).

Patients were monitored for 1 year. Follow-up periods included a baseline study just before transplantation (PreTx) and five studies post-transplantation at 1, 3, 6, 9 and 12 months after the procedure (1M, 3M, 6M, 9M and 12M). Demographic and clinically relevant information was collected ( Table 1 ). Immune status was assessed at each timepoint. We considered clinically relevant infections, rejection, liver autoimmunity and PTLD as primary outcomes.

The standard induction regimen consisted of basiliximab administered on days 0 and 4 post-transplantation, combined with tacrolimus and corticosteroids. Maintenance immunosuppression mainly consisted of tacrolimus and corticosteroids. In selected cases experiencing rejection episodes, mycophenolate mofetil was added to the regimen.

Antimicrobial prophylaxis included trimethoprim-sulfamethoxazole for Pneumocystis jirovecii, administered for two years post-transplantation, and either ganciclovir or valganciclovir for Cytomegalovirus, prescribed for six months post-transplantation regardless of donor/recipient serostatus.

Infectious events were categorized according to Van Delden et al. (20), and their relevance was defined as proven bacterial, probable/proven fungal and probable/proven viral infections, as well as viral syndromes. Early infections were defined as those occurring within the first month post-transplantation, while infections occurring thereafter were classified as late infections. Autoimmune hepatitis (AIH) was defined by a positive test result for any of the following antibodies: anti-mitochondrial M2, anti-filamentous-actin (F-actin), anti-Liver Cytosol Antigen Type 1 or anti-Liver-Kidney Microsomal antibodies, along with meeting clinical criteria. PTLD diagnosis was based on histopathologic criteria. The histopathological diagnosis of acute allograft rejection was determined based on the Banff criteria (21).

Cellular immune status was evaluated by multiparametric flow cytometry. Briefly, 75µL of whole blood was stained with various monoclonal antibody combinations, using different panels over time due to supplier changes ( Supplementary Table S1 ). Comparative analyses were conducted to ensure that the percentages remained consistent across all panels (data not shown). Cell acquisition was made on a BD FACSCanto™ or a DxFLEX flow cytometer. The resulting data were analyzed by FACSDiva™ (BD, USA) or Kaluza (Beckman Coulter, USA) software.

Immunophenotype of T lymphocytes (CD3⁺, further classified as CD4⁺ and CD8⁺), B lymphocytes (CD19⁺), NK lymphocytes (CD3^-CD16⁺CD56⁺) and NKT cells (CD3⁺CD16⁺CD56⁺) was performed. CD4⁺ and CD8⁺ T lymphocytes were further distributed in naïve (Tn, CD27⁺CD45RO^-), effector (Teff, CD27^-CD45RO^-), central memory (Tcm, CD27⁺CD45RO^+-) and effector memory (Tefm, CD27^-CD45RO^+-) subsets. Additional quantified subpopulations included recent thymic emigrants (RTE, CD4⁺CD45RA⁺CD31⁺), Treg (CD3⁺CD4⁺CD25⁺CD127^low), gamma-delta T lymphocytes (Tγδ, CD3⁺TCRγδ⁺), activated T cells (CD3⁺HLA-DR⁺) and memory B cells (Bm, CD19⁺CD27⁺).

Absolut numbers of T, B and NK lymphocytes were normalized to a patient-specific age range (22), creating a variable called “score”. To calculate the score, the median of the age-specific normal range was subtracted from the absolute number of lymphocytes in the subpopulation. The result was then divided by the difference between the 90^th percentile and the 10^th percentile of the normal range for that age group. Lymphopenia was defined as a score under -0.5 and lymphocytosis as a score over 0.5.

Regarding humoral immunity, levels of immunoglobulins G, A and M (IgG, IgA and IgM) were quantified on serum by nephelometry following manufacturer’s instructions (Siemens, Altona). Hypogammaglobulinemia was defined as values of IgG, IgA or IgM below the lower 95% confidence interval for each age group (23).

Quantitative variables were compared between two groups using the Mann–Whitney U test, except for the Δscore, for which the Student’s t-test was applied after confirming normal distribution with the Shapiro–Wilk test. When comparing quantitative variables across more than two groups, the Kruskal-Wallis test was used, followed by Dunn’s post hoc test for pairwise comparisons. Survival analysis was performed using the Cox proportional hazards model. The optimal multivariate model was selected using the Akaike Information Criterion, starting with variables with a p-value <0.100 from univariate analysis. The final model retained variables with the best fit. Statistical significance was set at p <0.05. All analyses were conducted with RStudio (version 4.3.3, R Core Team, 2024).

Ninety-five patients were ultimately included in our prospective study ( Figure 1 ), with a median age of 16 (7–88) months. The baseline characteristics of the cohort are detailed in Table 1 . Split graft from a deceased donor was the most common type of donation (41%), with biliary atresia being the predominant indication for LT (63%). Only 6 patients (6%) underwent combined liver-kidney transplantation, while 9 others (9%) required a second transplant due to primary graft failure (n=7), acute rejection (n=1) or tumor recurrence (n=1).

Ninety-seven percent of the patients received the standard induction regimen. Two patients (2%) also received mycophenolate as part of their induction therapy due to a combined transplant with a kidney graft, whereas one patient (1%) with a liver tumor was on chemotherapy at the time of transplantation.

Seven patients (7%) received intravenous immunoglobulin (IVIG). Two (2%) were prescribed IVIG prior to transplantation: one as part of the treatment for Gestational Alloimmune Liver Disease and another in the context of Evans syndrome associated with Autoimmune Lymphoproliferative Syndrome. Two patients (2%) received IVIG post-transplantation for the management of either adenovirus or Epstein–Barr virus infections. Three additional patients (3%) were treated with IVIG due to severe post-transplant hypogammaglobulinemia. Only two of these seven patients (29%) remained free of infections.

Regarding post-transplant outcomes, 6 episodes of acute cellular rejection were diagnosed (6%) along the follow-up (median time 233 [50 – 349] days). Three patients had a diagnosis of AIH before transplantation: one with type 1 AIH, one with seronegative AIH and one with suspected AIH. One patient tested positive for anti-F-actin antibodies at a titer of 1:80 at 6M, though the antibody was undetectable in subsequent tests. Other autoimmune complications included one case of autoimmune neutropenia and one of autoimmune hemolytic anemia. PTLD was diagnosed in 2 patients (2%) at 6M and 9M, respectively. The low number of rejection, AIH or PTLD events reported prevented us from doing statistical analysis.

Regarding infections, most of them occurred within the first month post-transplantation (early infections) (median time 4 [1-12] days), affecting 41 patients (43%) ( Figure 2A ). In subsequent months, the proportion decreased to 17%, 24%, 10% and 9% during their respective follow-up periods ( Figure 2A ). Early infections were predominantly bacterial, accounting for 51% of cases ( Figure 2B ). In contrast, late infections (median time 100 [30-150] days) were primarily viral, comprising 68%, 67%, 47% and 75% of infections during the corresponding follow-up periods ( Figure 2B ). Pathogens causing early and late infections are detailed in Supplementary Table S2 .

IgA hypogammaglobulinemia was detected in 2 (3%) recipients before the procedure, while the rest of patients maintained normal levels of both IgG and IgM ( Supplementary Table S3 ). Transplantation had a negative impact on immunoglobulin levels during the first month post-transplantation, with 18 (20%), 19 (21%) and 17 (19%) recipients developing hypogammaglobulinemia for IgG, IgA, and IgM, respectively. During the subsequent months, immunoglobulin levels gradually increased ( Figures 3A-C ) and, by the end of the follow-up period, most patients had returned to normal levels. However, 7 (11%) patients still had IgM hypogammaglobulinemia, and 2 (3%) patients had hypogammaglobulinemia of either IgG or IgA ( Supplementary Table S3 ).

Lymphopenia was frequent among recipients prior to transplantation (78%) ( Supplementary Table S3 ), and mainly attributable to baseline CD3⁺ T lymphopenia (81%). In contrast, pre-transplant B and NK lymphopenia was observed in lower percentages (42% and 6%, respectively) ( Supplementary Table S3 ). Accordingly, the baseline median score of total lymphocytes ( Figure 4A ) and T lymphocytes ( Figure 4B ) mirrored each other, both being below -0.5 prior to transplantation, including CD4⁺ ( Figure 4C ) and CD8⁺ ( Figure 4D ) T subsets. Upon transplantation, T lymphocytes already increased above -0.5 at 1M ( Figure 4B ), rising from -0.77 (-0.94 to -0.60) to -0.30 (-0.60 to 0.07) (p<0.001). Concomitantly, both CD4⁺ (PreTx -0.76 [-0.89 to -0.57] vs 1M -0.30 [-0.53 to 0.13], p<0.001) and CD8⁺ (PreTx -0.71 [-0.80 to -0.51] vs 1M -0.34 [-0.58 to 0.02], p<0.001) T-cell scores also exceeded -0.5 at 1M ( Figures 4C, D , respectively).

CD3⁺ T lymphocytes at 1M and 3M were significantly lower compared to 12M values (1M -0.30 [-0.60 to 0.07] and 3M -0.30 [-0.54 to 0.04] vs 12M -0.01 [-0.24 to 0.20], p=0.005 and p=0.008, respectively), with the CD3⁺ T-cell score remaining comparable from 6M onwards. Regarding CD4⁺ T-cell score, it normalized at 1M and remained stable throughout the follow-up ( Figure 4C ), with no significant differences observed. Kinetics of the CD8⁺ T-cell score paralleled those described for CD3⁺ T lymphocytes ( Figure 4D ). CD8⁺ T lymphocytes at 1M and 3M were significantly lower compared to 12M values (1M -0.34 [-0.58 to 0.02] and 3M -0.24 [-0.49 to 0.05] vs 12M 0.11 [-0.18 to 0.30], p<0.001 and p=0.003, respectively). Normalization was observed at 6M, after which the CD8⁺ T-cell score remained stable.

Regarding B lymphocytes, a significant expansion at 1M was observed (PreTx -0.46 [-0.59 to -0.21] vs 1M -0.04 [-0.36 to 0.57], p<0.001) ( Figure 4E ), showing values over 0.5 score (lymphocytosis) in 27 (30%) patients at that timepoint. B-cell expansion observed at 1M significantly decreased by 6M (-0.25 [-0.49 to -0.14], p<0.001), when normalization was achieved. From that time onwards, B-cell score remained stable, with no significant differences observed.

The impact of transplantation on NK-cell score appeared less pronounced initially ( Figure 4F ). Compared to pre-transplant study, a significant increase was detected at 3M (PreTx -0.33 [-0.43 to -0.21] vs 3M -0.17 [-0.34 to -0.02], p<0.001). After 6M, NK-cell scores normalized and remained stable in the following months.

Firstly, the decrease in Treg subpopulation at 1M ( Table 2 ) was likely an artifact of the technique, as basiliximab (anti-CD25) used in induction therapy interfered with CD25 detection by flow cytometry. Therefore, 1-month Treg frequencies were excluded from our analysis.

Infant patients aged 0–1 year ( Table 2 ) showed a significant increase in Tγδ lymphocyte frequency, rising from 2.63% pre-transplantation and 3.22% at 1M to 5.10% at 3M (p= 0.008 and p =0.001, respectively) and 9.01% at 6M (p=0.015 and p=0.002, respectively). In recipients aged 1–2 years, Tγδ lymphocyte frequency significantly increased from 5.40% pre-transplantation to 9.72% at 9M and 10.62% at 12M (p=0.005 and p=0.003, respectively). This rise was also significant when comparing 1M (3.48%) to 6M (6.37%), 9M and 12M (p=0.008, p<0.001 and p<0.001, respectively), and when comparing 3M (4.60%) to 9M and 12M (p<0.001 for both comparisons).

Regarding NKT lymphocytes in patients aged 1–2 years ( Table 2 ), their frequency increased from 0.25% at 1M to 0.61% at 9M (p= 0.003) and 0.62% at 12M (p=0.006). Conversely, Treg frequency decreased from 8.49% at 3M to 5.10% at 9M (p<0.001) and 5.17% at 12M (p<0.001), while the frequency of RTE declined from 67.20% at 1M to 52.07% at 6M and 52.79% at 9M (p= 0.015 and p= 0.012, respectively). In contrast, activated CD3⁺HLA-DR⁺ T lymphocytes showed an increase from 7.51% at 1M to 17.79% at 9M (p=0.023).

Similarly, patients aged 2–6 years increased their Tγδ and CD3⁺HLA-DR⁺ subsets throughout the follow-up period ( Table 2 ). Interestingly, frequencies of Bm only showed an increase in that age group, rising from baseline 10.67% and 11.74% at 1M to 20.33% at 12M (p=0.007 and p=0.016, respectively). On the other hand, in older patients aged 6–12 years, only an increase in Tγδ lymphocytes from 6.66% at 1M and 6.62% at 3M to 17.56% at 12M (p=0.009 and p=0.008, respectively) was detected. For recipients aged 12–18 years, the frequencies remained stable throughout the entire follow-up period, with multiple comparisons yielding no significant p-values.

The distribution by age of CD4⁺ and CD8⁺ Tn, Teff, Tcm, and Tefm lymphocyte subpopulations throughout the follow-up period remained comparable ( Supplementary Figures S1 , S2 ). However, in patients aged from 2–6 years, median frequencies of CD8⁺ Tn significantly decreased from 1M to 12M (75.61% vs 54.20%, p=0.012). Conversely, CD8⁺ Teff and Tefm subsets in this age group significantly increased in the same period (CD8⁺ Teff 1.71% vs 11.34%, p=0.005; CD8⁺ Tefm 3.81% vs 10.00%, p=0.010) ( Supplementary Figure S2 ).

When segregated according to the time of infection (early/late), statistical analysis showed that infected patients were significantly younger at transplant and primarily received split grafts ( Table 1 ). To better assess immunological parameters post-transplantation, we subtracted each subpopulation score from pre-transplant study to the one obtained at 1M (Δscore). The higher the Δscore, the better the normalization of lymphocyte subpopulations. Patients that remained free from late infections had higher Δscore for T CD3⁺ (0.524 vs 0.263, p=0.018) and T CD4⁺ (0.452 vs 0.287, p=0.036) than those who developed late infections ( Table 3 ). Interestingly, T CD8⁺ Δscore was also higher in non-infected patients, although this increase nearly reached statistical significance (0.483 vs 0.178, p=0.054).

Subsequently, a survival analysis was performed to explore the relationship between pre-transplant immunological status and the risk of early infections ( Table 4 ). The univariate analysis identified a significant association between the risk of post-transplant infections and both the age at transplantation and the type of graft. Patients aged 0–1 years (hazard ratio [HR] 5.23, p=0.027) or 1–2 years (HR 5.29, p=0.034) had a significantly higher risk of infection. Transplantation using a split graft was associated with a threefold risk for infection (HR 3.02, p=0.071), although this correlation was not statistically significant in the univariate analysis. None of the immunological variables analyzed were associated with the risk of early infection. Interestingly, in the multivariate analysis, only transplantation with a split graft was independently associated with an increased risk of early infection (HR 3.42, p=0.047).

We next analyzed how immune status at 1M influenced the likelihood of remaining free from late infection ( Table 5 ). The univariate model revealed that infants aged 0–1 years had a significantly higher risk of late infection (HR 3.49, p=0.046). Conversely, patients who received a whole graft from deceased donor had a significantly lower risk (HR 0.28, p=0.044). In terms of immunological status, CD3⁺ or CD8⁺ T lymphopenia (HR 2.48, p=0.005 and HR 2.38, p=0.008, respectively) and hypogammaglobulinemia (IgG, IgA or IgM), were associated with a higher risk of late infection (HR 2.18, p=0.036 and HR 2.40, p=0.011 and HR 2.61, p=0.006, respectively). The multivariate model showed that only lymphopenia T CD3⁺ was independently associated with an increased risk of late infection (HR 2.13, p=0.030). Kaplan-Meier curves for patients with or without T lymphopenia are graphed in Figure 5 . Patients with CD3⁺ T lymphopenia showed significantly higher infection rates after the first month post-transplantation (p=0.005) ( Figure 5A ). While CD4⁺ T lymphopenia did not show a statistically significant association with infection rates ( Figure 5B ), the presence of CD8⁺ T lymphopenia was significantly associated with higher infection rates, highlighting the differential impact of T-cell subsets on infection risk ( Figure 5C ).

Furthermore, we stratified patients into four groups based on IgG levels and CD3⁺ T-cell counts at 1M: normal IgG/normal CD3⁺ (n=50), normal IgG/CD3⁺ T lymphopenia (n=22), IgG hypogammaglobulinemia/normal CD3⁺ (n=6) and IgG hypogammaglobulinemia/CD3⁺ T lymphopenia (n=11). The risk of late infection was significantly higher in patients with normal IgG/CD3⁺ T lymphopenia (HR 3.03, 95% CI 1.46–6.30, p=0.003), IgG hypogammaglobulinemia/normal CD3⁺ (HR 4.16, 95% CI 1.36–12.74, p=0.013) and IgG hypogammaglobulinemia/CD3⁺ T lymphopenia (HR 2.86, 95% CI 1.10–7.44, p=0.031), compared to the normal IgG/normal CD3⁺ group.