1 Introduction

Kidney transplantation (KT) is the preferred treatment in children with end-stage kidney disease, as it improves survival, growth, cognitive development, and quality of life compared to other renal replacement therapies [1, 2].

Randomized controlled trials (RCT) of immunosuppressive medication in pediatric transplantations are sparse. Thus, evidence primarily comes from observational studies and extrapolation from adult studies [3-5]. This has resulted in a wide range of immunosuppressive protocols. In Europe, there has been an increasing trend toward using maintenance therapy with tacrolimus, mycophenolate mofetil (MMF), and corticosteroids the first 30 days after KT, without induction therapy in approximately 70% of cases [5]. In the United States of America, induction therapy is used in the majority (94%) of pediatric KT, and most commonly triple immunosuppression with tacrolimus, MMF, and corticosteroids is used (54%), followed by dual therapy with tacrolimus and MMF (37%) [6].

Despite the diversity of immunosuppressive protocols, there is a consensus that tacrolimus and MMF are the most effective drugs for maintenance therapy. This is reflected in the guidelines of the Kidney Disease Improving Global Outcomes (KDIGO) [7] and the National Institute for Health and Care Excellence (NICE) [8].

MMF was introduced in the 1990s, replacing azathioprine as the preferred antiproliferative agent. Observational studies in pediatric kidney transplant recipients comparing cyclosporine (CsA), MMF, and corticosteroids to historical cohorts receiving azathioprine showed lower rates of acute rejections [9], while no RCTs comparing MMF and azathioprine have been conducted. Furthermore, evidence of MMF in a tacrolimus-based regimen is limited [3, 4].

The appropriate MMF dosage in pediatric KT has been evaluated based on analyses of the area under the curve (AUC) of mycophenolic acid (MPA), with the target AUC of 30–60 mg h/L extrapolated from the adult population [10-12]. The pharmacokinetics of MPA varies considerably both between subjects and within subjects, and since MPA trough levels do not correlate to the AUC, therapeutic drug monitoring (TDM) with multiple measurements is more accurate but time and cost-demanding [12]. In pediatric recipients, dosing based on body surface area (BSA) is more precise than weight [9]. CsA augments MMF clearance compared to tacrolimus, and dose reduction for tacrolimus MMF combination is advised [13] with a recommended MMF dosage with concomitant use of tacrolimus of 900–1200 mg/m²/day [3, 4, 8].

Tacrolimus was approved in the 1990s and has replaced CsA as the preferred calcineurin inhibitor. A single RCT comparing CsA and tacrolimus in pediatric KT with concomitant azathioprine and corticosteroids demonstrated improved graft survival and fewer rejections with tacrolimus [14].

Several RCTs conducted in pediatric KT have focused on corticosteroid-free or -withdrawal regimens, with corticosteroid-induced growth retardation as a particular concern. A meta-analysis of five RCTs concluded that corticosteroid-free and -withdrawal regimens are safe regarding rejection rates and improve prepubertal growth [15]. Additionally, improved lipid profiles, decreased incidence of posttransplant diabetes mellitus, and improved blood pressure control have been demonstrated [3, 16].

Due to side effects and infections, reduction of the predefined standard immunosuppressive medication is often necessary in pediatric recipients [10, 17, 18]. Reducing immunosuppression increases the risk of rejection and sensitization with human leukocyte antigen (HLA) antibodies. De novo donor-specific (dnDSA) HLA antibodies constitute a significant concern, as they are associated with reduced graft survival [19], increased risk of rejection [20, 21], and complicated retransplantation [22].

Since 1990, we have performed corticosteroid-free pediatric KT at Odense University Hospital [23-25] and have extensive experience in corticosteroid-free KT in adults [26, 27]. Our standard protocol for immunologically uncomplicated transplantations consists of induction with basiliximab and maintenance with MMF and tacrolimus.

However, we have observed a significant need for reducing immunosuppressive therapy due to side effects, raising concerns regarding potential increased alloimmunity.

We, therefore, aimed to describe the alteration in immunosuppressive medication, explore the reasons for the reductions, and assess the potential impact on alloimmunity in a cohort of pediatric kidney transplant recipients.

2 Methods and Material

3 Results

3.2 Alterations in the Immunosuppressive Medication Over Time

Most recipients (78%) received standard low-immunological risk immunosuppression. Eight KT had DSA pretransplant or were ABOi KT and intensified immunosuppression was applied (Table 1). The majority of recipients followed the standard medication protocols, but five had personalized medication based on specific considerations.

All recipients were initially treated with tacrolimus (Table 2). Two ceased tacrolimus during the first year due to EBV, one developed posttransplant lymphoproliferative disorder. After the first year, the median trough level of B-tacrolimus was at the desired target—5.6 ng/mL, and tacrolimus continued to be well-regulated according to the target level of 5 ng/mL in the subsequent years (Figure 1).

TABLE 2. Changes in immunosuppressive therapy from transplantation to year one. (The two recipients with immediate graft loss are not included at year one.)

Maintenance therapy	Transplantation	Year 1
Tacrolimus, n (%)	49/49 (100%)	45/47 (95.7%)
B-Tacrolimus level (ng/mL), median (IQR; range)	13.7 (5.5; 6.5–33.4) (Day30)	5.6 (1.9; 2–13.3)
MMF, n (%)	49/49 (100%)	37/47 (78.7%)
Dose/BSA (mg/m2/day), median (IQR; range)	846.4 (124; 336–2164)	353.9 (419; 50–1007)
Reduced dose or no MMF, n (%)	4/46a (8.7%)	37/47 (78.7%)
Reduced dose/no MMF in age groups, n (%)
0–5 (12)	0/11a (0%)	10/11 (90.9%)
6–11 (15)	2/15 (13.3%)	13/15 (86.7%)
≥ 12 (22)	2/20a (10.0%)	14/21 (66.7%)
p-value	0.67	0.27
Prednisolone, n (%)	10/49 (20.4%)	5/47 (10.6%)
Dose/BSA (mg/m2), median (IQR; range)	55.9 (23.3; 17–60)	2.5 (2.7; 1.6–11.0)
Other, n (%)
Azathioprin	0	1/47 (2.1%)
Number of immunosuppressive drugs, n (%)
One drug therapy	0/49 (0%)	9/47 (19.2%)
Two drug therapy	39/49 (79.6%)	35/47 (74.5%)
Three drug therapy	10/49 (20.4%)	3/47 (6.4%)
Reduced IS (one IS, OR two IS AND MMF < 600 mg/m2/day)	3/47a (6.4%)	34/47 (72.3%)
0–5 (12)	0/12 (0%)	9/11 (81.8%)
6–11 (15)	1/15 (6.7%)	13/15 (86.7%)
≥ 12 (22)	2/20a (10.0%)	12/21 (57.14%)
p-value	0.78	0.13

• Abbreviations: BSA, body surface area; IS, Immunosuppressive medication; MMF, mycophenolate mofetil.
• ^a Three BSA missing at year 0.

Similarly, all recipients received MMF from transplantation, with 79% still receiving it after the first year, but at year one, 73% received a dosage below the desired of 600 mg/m²/day (Table 2; Figure 2). MMF dosage was targeted at 900 mg/m²/day at transplantation, and a median dosage of 846 mg/m²/day (IQR 124 mg/m²/day) was obtained. At year one, the median dosage in the cohort was only 354 mg/m²/day (IQR 419 mg/m²/day) (< 0.001).

Ten recipients were treated with prednisolone from transplantation, with eight tapering to hold during the first year (Table 2). In three recipients, prednisolone was initiated during the first year—one as part of anti-rejection therapy and two to replace tacrolimus due to chronic Epstein–Barr viremia.

After 1 year, 72% of the cohort received either mono- or dual therapy with an MMF dosage below 600 g/m²/day. The reduction was numerically more pronounced in recipients from 1 to 11 years (Table 2). MMF was successfully increased to the standard dose in some, but most continued a reduced dose (Figure 2).

3.4 Infections and Hospitalizations Over Time

During the first year after KT, the median number of hospitalizations due to all causes was one (IQR 3), and infections were the main reason for hospitalization, with 60% being hospitalized at least once with infection. After the first year, the median number of hospitalizations due to infections decreased to zero (Table S1). Recipients under the age of six were significantly more likely to be hospitalized than older age groups (p = 0.002), especially with infections (p = 0.003) (Figure 3a,b).

A median of five infections (IQR 6) requiring either hospitalization or outpatient care was observed during the first year. Recipients below 6 years old had significantly more infections than older age groups (p < 0.001) (Figure 3c). No association was found between baseline immunosuppressive medication and the number of infections when comparing standard immunosuppression (basiliximab, tacrolimus, and MMF) with other regimens (p = 0.19). Bacterial infections, particularly urinary tract infections (UTI), were the most frequent, with 72% experiencing at least one UTI during the first year. After the first year, the number of infections declined to a median of between zero and two per year (Table S1).

Recipients were routinely screened for CMV and EBV DNA replication. During the first year, 32% had CMV infections and 23% had EBV infections. CMV infections were typically asymptomatic or mild, while EBV infections often caused malaise and hospitalizations. One patient developed PTLD during the first year, with a total of two cases during the follow-up period, both associated with EBViremia. After the first year, only two recipients had incident CMV infections and one had an incident EBV infection. Eighteen percent had BK viruria or viremia, but the virus did not cause biopsy-verified BK nephropathy. CMV and BK had an equal distribution among ages; only EBV was more frequent in the youngest recipients.

3.5 Alloimmunological Outcomes

3.5.1 HLA Sensitization

During the follow-up period, 11.4% of the cohort developed dnDSA, and if including recipients with DSA but without pretransplant solid phase assay, a maximum of 17.4% developed dnDSA (Table 4). The time to the first detection of dnDSA was 3.91 years (IQR 2.28), and the 5-year dnDSA-free survival was 87.4% [95% CI 69.5; 95.1] (Figure S2).

TABLE 4. HLA sensitization. Non-donor-specific HLA antibodies, donor-specific HLA antibodies, and de novo donor-specific HLA antibodies. Distribution of HLA classes and age groups. (Two recipients without posttransplant solid phase assay were not included in the assessment of HLA antibodies and dnDSA, and three recipients without pretransplant solid phase assay were not included in the assessment of dnDSA.)

HLA antibodies, n (%)
All	28/47 (59.6%)
HLA Cl I	7/28 (25.0%)
HLA Cl II	9/28 (32.1%)
HLA Cl I & HLA Cl II	11/28 (39.3%)
Unknown	1/28 (3.6%)
Non-DSA	14/47 (29.8%)
DSA	14/47 (29.8%)
Pretransplant	6/14
Posttransplant	8/14

De novo donor-specific antibodies, n (%)
DnDSA during follow-up	5/44 (11.4%)
HLA Cl I	0/5
HLA Cl II	3/5 (60%)
HLA Cl I & HLA Cl II	2/5 (40%)
Time to dnDSA (years), median (IQR; range)	3.9 (2.3; 2.2–9.2)
MFI, first detected dnDSA, median (IQR; range)	1700 (6100; 1400–14 600)
DnDSA in age groups
0–5 (12)	2/12 (16.7%)
6–11 (13)	0/13 (0%)
≥ 12 (19)	3/19 (15.8%)
p-value	0.36

• Abbreviations: dnDSA, de novo donor-specific antibodies; DSA, donor-specific antibodies; HLA, human leukocyte antigen; MFI, mean fluorescence intensity.

All five recipients with dnDSA were categorized as low-immunological risk at the time of transplantation and received maintenance therapy with tacrolimus and MMF. We found no differences in baseline characteristics between recipients with and without dnDSA (Table 1). When statistically examining tacrolimus and MMF levels over time in the recipients with dnDSA compared to the rest, we did not find significantly different levels (Table S3). Neither did we observe a difference between the groups when examining whether their overall immunosuppression was reduced at 1 year (monotherapy or dual therapy with MMF < 600 g/m²/day) (Table S3). However, the medical histories of the recipients with dnDSA revealed periods of reduced immunosuppression, particularly notable for reductions in tacrolimus. One recipient was strongly suspected of nonadherence; two recipients received reduced or no CNI due to PTLD; one recipient had reduced MMF and tacrolimus due to multiple infections; and the last recipient had reduced MMF but maintained normal tacrolimus levels.

HLA antibodies were detected in 59.6% of the recipients. Most of the antibodies were weak (MFI < 3000), the majority (39%) were a combination of HLA class I and II antibodies, and the non-DSA HLA antibodies were predominantly class I. In comparison, none of the dnDSA were exclusively class I (Table 4).

3.5.2 Rejections

The cumulative incidence of rejection was 4.1% [95% CI −1.5; 5.5] after 1 year, 10.4% [95% CI 1.6; 18.7] after 5 years, and 14.6% [95% CI 3.0; 25.9] after 10 years when death and graft loss were considered competing risks (Figure 4).

During the follow-up period, seven recipients had eight rejections. Two rejections occurred within 3 months of transplantation, three occurred between one and 2 years after transplantation, and two occurred more than 5 years after transplantation. The median age of the recipients at the time of rejection was 15.5 years (IQR 12.3).

In 75% of the recipients with rejection, the medical records reported a reduction in immunosuppression before the rejection. One-third were not taking their medication as prescribed, one-third had reduced medication due to infections, and one-third had dysregulated B-tacrolimus due to gastrointestinal side effects (early rejections). Two recipients with rejection lost their grafts later, but none of the graft losses were directly related to the rejection episodes.

All the rejections were T-cell-mediated rejections. Five were classified as Banff 1A, one as Banff Ib, one as Banff IIa, and one as Banff IIb.

One recipient with rejection had dnDSA, and 86.3% were sensitized, which was numerically higher than the rest of the population. However, the hazard ratio of rejection was not significantly higher in recipients who were sensitized or who had dnDSA (4.0 [95% CI 0.5; 34.9] and 1.97 [95% CI 0.2; 17.8], respectively). Neither could we statistically show an association between reduced immunosuppression at year one and rejections (only considering rejections occurring after the first year [Table S4]). The baseline characteristics of the recipients with rejection did not differ from the rest of the cohort, except for better HLA compatibilities in the recipients with rejection (data not shown).

3.6 Patient Survival and Graft Function

One recipient died from infection. Five-year patient survival was 100%, while 10-year was 93.8 [95% CI 63; 99].

During the follow-up period, six recipients lost their grafts. The median time from transplantation to graft loss was 3 years (IQR 3.8). Two recipients lost their graft immediately after transplantation: one due to graft thrombosis and the other due to the recurrence of undiagnosed primary hyperoxaluria. The remaining four lost their grafts due to: chronic Parvo B19 infection (after 3 years), multiple UTIs and suspected nonadherence (after 3.5 years), recurrence of membranoproliferative glomerulonephritis (after 3 years), and recurrence of C3 glomerulopathy (dense deposit disease) (after 6 years) (Figure 5).

Death-censored graft survival rates were 95.9% [95% CI 84.7; 99] at year 1, 88.2% [95% CI 73.7; 95] at year 5%, and 84.0% [95% CI 66.7; 92.8] at year 10 (Figure 4; Figure S5). There was no significant difference in graft survival between recipients receiving kidneys from deceased or living donors (p-value = 0.43), in graft survival based on age (p-value = 0.46), gender (p-value = 0.58), HLA-DR mismatches (p = value 0.56), rejection status (p-value = 0.26), or immunological risk pretransplant (p-value = 0.86). However, there was a tendency toward decreased graft survival in recipients with dnDSA (p-value = 0.06).

Evaluating the graft function, the mean eGFR increased to 75.5 mL/min/1.73 m² [95% CI 69.3; 81.5] 6 months after transplantation, after that decreasing with an average of 2.1 mL/min/1.73 m² 73 m² [95% CI 1.3; 2.9] each year (Figure 5). To investigate the impact of dnDSA and HLA sensitization on kidney function, a linear mixed-effect model was applied, and the mean eGFR was 21.0 mL/min/1.73 m² [95% CI 5.1; 36.8] higher in recipients without dnDSA compared to recipients with dnDSA (p = 0.009) (Figure 6a). There was no difference in kidney function between HLA-sensitized and nonsensitized recipients (Figure 6b).

4 Discussion

This retrospective single-center study examined 49 pediatric kidney transplant recipients, of which 80% were completely corticosteroid-free. By the end of the first year, 72% of the patients received either one immunosuppressive drug or two drugs and MMF below the targeted dose of 600 mg/m²/day. The median dosage of MMF was 354 mg/m²/day, nearly half of the intended dose at year one, and it remained reduced over the following years. However, tacrolimus trough levels were maintained at around 5 ng/mL throughout follow-up. Despite the minimized immunosuppression, rejection rates, dnDSA development, and graft losses were low compared to previously published data. Infection rates were high in the first year but decreased subsequently, likely due to reduced immunosuppression. Side effects like infections, leukopenia, and diarrhea primarily caused the reduction in immunosuppression, but these complications also markedly decreased after the first year. Recipients under six experienced significantly higher infection rates, whereas other side effects were evenly distributed among the age groups.

The deleterious effect of dnDSA was indicated by findings of a lower mean eGFR in recipients with dnDSA. We could not identify any statistically significant risk factors for developing dnDSA. However, a review of the recipients' medical histories revealed a marked reduction in CNI in most.

In our study, MMF primarily caused side effects, a well-known trait of MMF. The rates of side effects were comparable to other studies [10, 31, 32] and declined markedly after the first year. We observed equally distributed side effects among age groups, similar to the study of Bunchmann et al. [10]. In contrast, other studies have found more side effects among the youngest recipients [17, 32, 33]. On the other hand, we showed that infections generally affected children under six significantly more than the oldest (Figure 3), which Hogan et al. also observed in their study [34]. In our cohort, the most abundant number of infections and side effects were observed during the first year, decreasing during follow-up after reductions in immunosuppression. The MMF dosage of 354 mg/m²/day at year one was lower than reported in two studies, reporting 718 mg/m²/day in a tacrolimus-based population and 991–1084 mg/m²/day in CsA-based populations [10, 17]. The recommended target dose of 900–1200 mg/m²/day has been established from a target MPA Area under the curve (AUC of 30–60 mg h/L), extrapolated from MPA AUC in the adult populations [11, 12], but supported by evidence of decreased rejections also in pediatric recipients [3, 9, 31]. Studies of MMF toxicity have not established an upper threshold, and the efficacy of MPA TDM is debatable [11, 12]. We have not used MPA TDM-guided therapy in our center as routine for more reasons. First, measurements for MPA AUC are time-consuming for the recipients and their families. Second, the evidence supporting its clear benefits is controversial [12]. However, we acknowledge that in a corticosteroid-free population, MPA TDM is recommended due to the decreased immunosuppression and the increased MPA exposure without concomitant corticosteroids. The absence of MPA TDM in this study poses a significant limitation, precluding definitive evidence regarding whether our cohort received a lower dosage than recommended. However, our cohort's targeted dosage of 600 mg/m²/day was marginally lower than the recommended [8, 28], and studies have shown a tendency toward underexposure in fixed dosage regimes [35]. On the other hand, the absence of corticosteroids could potentially increase the MPA AUC. Furthermore, the dose-normalized exposure of MPA has been shown to increase in the first months after transplantation [11].

In our cohort, the death-censored 5-year graft survival was 88.2%, comparable to previously published data from Europe, Australia, and the United States [36, 37]. In contrast, our 1-year rejection rate of 4.1% was lower than the rejection rates reported in other studies, which ranged from 10.5% to over 25% [10, 38-40]. Six of the eight rejections occurred after the first year, primarily in adolescents, and were associated with either nonadherence or reduced immunosuppression due to infection.

During the follow-up period, 11.4% of the recipients developed dnDSA, while 59.6% were HLA sensitized. Compared to similar studies, the incidence of dnDSA was low, reporting dnDSA rates ranging from 10% to 73%. However, in most pediatric kidney transplant recipients, the incidence of dnDSA is about 30%–40% using similar MFI cut-offs [21, 41-46]. We speculated whether the HLA compatibility was higher in our cohort, but when reported, the HLA-A, -B, -Cw, -DRB1, and-DQB1 incompatibility varied from 3.9 to 7.4 [42, 43, 46], similar to our findings. A follow-up period of 5 years, with a median time of almost 4 years to the first detection of dnDSA, also seems adequate for assessing dnDSA as an outcome measure. In most studies, the median time to detection of dnDSA is from 2 to 4 years [42, 46-49], similar to our observation of about 4 years.

DnDSA is associated with inferior graft survival and rejection [19-21], while the effect of HLA sensitization is debatable. Cioni et al. found in a study of 144 nonsensitized pediatric kidney transplant recipients that 56% developed de novo HLA antibodies, which were not associated with inferior graft outcomes [50]. In a study of 82 nonsensitized pediatric kidney transplant recipients, Ginevri et al. found that 45% developed de novo HLA antibodies, but only the DSA were associated with graft failure [49]. In contrast, Susal et al. found that 42% of 83 adult recipients developed HLA antibodies, which was associated with graft loss [51]. We found a tendency (NS) toward increased graft failure in recipients with dnDSA, but this tendency was not observed with HLA sensitization. However, we did see reduced eGFR in the recipients with dnDSA compared to recipients without dnDSA (p = 0.009), while HLA sensitization did not affect eGFR (Figure 6). Nevertheless, all persistent HLA antibodies complicate retransplantation [22, 52]. To investigate the clinical relevance of HLA sensitization and its potential impact on retransplantation, our definition of HLA sensitization included both pre and posttransplant DSA combined with non-DSA HLA antibodies posttransplant. This broad definition might have contributed to the high sensitization rate. However, on closer examination, the non-DSA HLA antibodies were weaker and predominantly HLA class I. Consequently, the non-DSA HLA antibodies appear less clinically relevant and may not pose a significant challenge in retransplantation.

To our knowledge, no studies have investigated the relationship between the level of MMF and dnDSA in pediatric KT. Two studies have examined the association of dnDSA and corticosteroids, one in adult KT [53] and one in pediatric [54]. Both studies failed to demonstrate a difference in dnDSA between corticosteroid-based and corticosteroid-free/-withdrawal protocols, but the follow-up period of only 24 months in the pediatric study may have been insufficient for evaluating dnDSA formation [54]. In our transplant center, we have performed corticosteroid-free transplantations for more than three decades to avoid the side effects of corticosteroids and enhance the growth and metabolic health of the recipients [25]. With 80% of the cohort being corticosteroid-free (increasing to 89% at year one), our findings do not support excessive dnDSA formation in corticosteroid-free and MMF-reduced recipients.

Tacrolimus, on the other hand, does seem to inhibit the formation of dnDSA. Urzykowska et al. evaluated the effect of triple therapy (tacrolimus/MMF/corticosteroids) on dnDSA and found that only tacrolimus level was associated with dnDSA [55]. The median tacrolimus trough level was 7.9 ng/mL in the dnDSA-negative recipients and 7.1 ng/mL in the dnDSA positive. In a study of extended-release tacrolimus in adult KT on a corticosteroid-free protocol, Gatault et al. concluded that tacrolimus trough level above 7 ng/mL reduced the risk of alloimmunological events [56]. Hiramitsu et al. reported a trough level below 4.9 ng/mL to be associated with dnDSA [57]. In addition, maintaining a stable trough level with minimal variation seems vital in preventing dnDSA [44, 47]. We did not study individual variations in tacrolimus trough levels, but the median level remained close to the desired target over time.

Apart from those mentioned above, our study has certain limitations. One major limitation is that the study was observational and retrospective. Moreover, our cohort was heterogeneous, with diverse pretransplant immunological risk profiles, reflecting the real-life situation. This heterogeneity makes it more challenging to extrapolate our findings to specific subgroups. Pretransplant HLA antibody screening was not routinely performed until 2011, and recipients without were also included. The absence of data regarding pretransplant HLA antibodies affects the interpretation of de novo HLA sensitization, and this complicated the inference of whether the HLA sensitization was clinically relevant. Furthermore, the outcomes of rejection and dnDSA were sparse, limiting the statistical robustness and the ability to adjust for confounders. Furthermore, rejections might be underestimated as surveillance biopsies are not standard care in our center.

Author Contributions

Ann-Maria Gramkow: conceptualization, methodology, validation, formal analysis, investigation, resources, data curation, writing – original draft, writing – review and editing, visualization, project administration, funding acquisition. Emilie T. Gramkow: investigation, writing – review and editing. Per Wittenhagen: writing – review and editing. Johanne H. Baatrup and Pernille Koefoed-Nielsen: resources, writing – review and editing. Helle C. Thiesson: conceptualization, methodology, validation, resources, writing – review and editing, supervision, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.