The DIABGRAFT study was conducted in collaboration with the Pediatric Hepatology and Gastroenterology and Specialized Pediatrics (Endocrinology and Nephrology Units) Services of Cliniques universitaires Saint-Luc (CUSL) in Belgium (Brussels). This study was approved by the local ethical committee (CUSL and UCLouvain Hospital-Faculty Ethics Committee; approval number 2019/12MAR/118) and was conducted in accordance with the Declaration of Helsinki. Our study included liver and renal transplant pediatric patients (<18 years of age) at CUSL. Were excluded patients with a history of diabetes (i.e., type 1, type 2, neonatal or monogenic), pancreatitis, Down Syndrome, cystic fibrosis (n = 1), a second organ transplantation for our LT cohort (n = 4; cardiac, renal), patients deceased shortly after transplantation (<1 year, n = 14), and patients with incomplete medical record (n = 8).

DIABGRAFT was constituted of two parts. Its retrospective part (rDIABGRAFT) consisted of collecting data of pediatric patients who benefited from a liver transplant performed at CUSL between April 2012 and April 2019, or that benefited from a renal transplant in our center between September 2005 and April 2019. The prospective part (pDIABGRAFT) of the study consisted of a longitudinal glycemic evaluation of liver and renal transplant children in CUSL between 2020 and 2022 with the use of dynamic endocrine testing (Figure 1). Informed consents were collected from parents and from all children over six years of age.

At CUSL, liver and renal transplant children receive standard immunosuppression protocol as per international guidelines (13). For LT patients, this protocol includes the association of a monoclonal anti-CD 25 antibody (basiliximab, Simulect®) and a calcineurin inhibitor (tacrolimus, Prograft®) (14, 15). For RT patients, this protocol is based on a combination of Tacrolimus, glucocorticoids, monoclonal anti-CD 25 antibody and a cell proliferator inhibitor as mycophenolate mofetil (Cell-Cept®). Doses of glucocorticoids are introduced or increased when a LT/RT patient presents an acute cellular rejection (ACR). Complete treatment protocol is available in Supplementary data (Text S1).

About glucose monitoring, after a liver or renal transplantation at CUSL, glycemia is measured daily during hospitalization (between two weeks and one month) and for LT patients, glucose monitoring is regularly performed during a month until the patient returns to his home (after three months), after what yearly glycemic control is performed. For RT patients, the measure of fasting glycemia continues once weekly until the 6th month post-transplantation, when the control becomes once a month.

We defined hyperglycemia based on guidelines of the international consensus for diabetes of the American Diabetes Association (ADA): patients presented HG when fasting plasma glucose (FPG) or random plasma glucose (PG) levels exceeded respectively 126 mg/dl (7.0 mmol/L) and 200 mg/dl (11 mmol/L) for at least two measurements separated by 24 h, and not under a condition of stress such as the day of the transplant (16).

For our rDIABGRAFT study, the term “transient hyperglycemia” was used to define patients with HG (as described above) without overt diabetes diagnosed and diabetes was notified when patient required a persistent treatment (i.e., insulin or oral antidiabetics). For our pDIABGRAFT study, as we used dynamic testing, we classified our patient based on ADA guidelines: when a patient presented impaired fasting glucose (IFG) and/or impaired glucose tolerance (IGT) and/or HbA_1C from 5.7% to 6.4% (39–47 mmol/mol), we defined a “prediabetes” state. IFG was defined as FPG between 100 and 125 mg/dl (5.6 and 6.9 mmol/L) and IGT as 2h-PG levels during an oral glucose tolerance test (OGTT) from 140 to 199 mg/dl (7.8 and 11.0 mmol/L) (16). Diabetes was defined when a patient presented FPG ≥ 126 mg/dl (7.0 mmol/L) or a random PG or 2-h PG levels during OGTT > 200 mg/dl (11.1 mmol/L) or hemoglobin A1C (HbA_1C) > 6.5% (48 mmol/mol) and/or when the patient presents classic symptoms of HG (16).

After obtaining the consent of pediatric patients and their parents, a glucose sensor (The FreeStyle Libre Flash Glucose Monitoring system, Abbott) was placed on the patient two weeks post-transplant for one month to detect the presence of early dysglycemia. Pro-insulin and C-peptide secretion by enzyme-linked immunosorbent assays (ELISA) was measured after two weeks of transplant to analyze the insulin secretion function of beta-cell, and measures had continued at one, three, six and nine months for RT patients. The enzyme-linked immunosorbent assays used for our analyses were performed as per manufacturer's instructions (Proinsulin 10-1118-01 and C-peptide 10-1136-01 kits, Mercodia). To analyze the insulin sensitivity and secretion over time, an OGTT was performed after one month of transplantation and, for RT patients also at three, six and nine months (Figure 1). Patients were not treated with insulin during tests. The OGTT was performed after 8 h overnight fast with weight-based glucose load (1.75 g/kg for pediatric patient) (17). Glucose and insulin were measured at fasting and at 30, 60 and 120 min after the ingestion of glucose. Insulin resistance (IR) was evaluated with HOMA-IR (for homeostasis model assessments of fasting insulin resistance; Ins⁰_(µU/ml) × Gluc(0)_(mmol/L)/22.5) (18). If the HOMA index is less than 1.6, the result is normal. When the HOMA index is between 1.7 and 2.3, the patient presents a moderate form of IR and if the value is greater than 2.4, he suffers from a severe form of IR.

Patient history data included sex, date of birth, height, weight and gestation at birth (i.e., term, pre- or post-term), country of origin, date of death if patient deceased, the presence of hypo- and hyperglycemia in the neonatal period, dysmaturity, any chronic and hormonal treatment before the transplant, presence of dialysis for RT, its duration and type (e.g., hemodialysis, hemodiafiltration, peritoneal dialysis), endocrine or autoimmune diseases, acanthosis nigricans and sickle cell anemia. Also, we collected data about familial history as the presence or absence of consanguinity, metabolic syndrome, diabetes (type 1, type 2, gestational and monogenic), polycystic ovarian syndrome, fetal dystocia and sickle cell anemia.

We included information about the liver or renal transplant such as disease etiology, transplant date, the type of immunosuppressants administrated (tacrolimus, cyclosporine A, sirolimus, glucocorticoids), the use and duration of glucocorticoids in pre- and post-transplant period, the presence and date of liver or renal rejection, the type of transplant (living or cadaveric) and the link with the donor. We also collected anthropometric data in pre- and post-transplant period (at one, three, six and nine months after RT): weight, heigh, body max index (BMI) in standard deviation score (SDS) and Tanner stage. To obtain values in SDS score we used Belgian Flemish reference charts and Cole's Corpulence Curve (19, 20).

We collected glycemia and HbA_1C data before and after transplantation. When a patient presented HG after the first day of the transplantation, we reported the number of its occurrence, the date of its first and last observation and if a treatment was received (e.g., insulin therapy, antidiabetic oral), its doses per day, and its duration. The number of glycemia recorded was obtained by counting all measurements performed from the first consultation at CUSL (pre-transplantation evaluation) until the end of our data collection (in November 2021 for LT and in April 2022 for RT). The duration between the day of the transplantation and the last glycemia recorded was calculated to obtain the glycemia follow-up. We used REDCap (Research Electronic Data Capture) tool to collect and manage study data (21, 22).

Discrete variables are described as numbers and percentages, and continuous variables were presented as medians with interquartile range (IQR). The characteristics of children were compared according to the occurrence or not of HG using Fisher exact test for discrete variables and Student t test or Mann-Whitney test for continuous variables. A binary logistic regression analysis was performed to predict HG occurrence from all potential predictors described in data collection section and results were expressed by estimating odds ratios (OR) with their 95% confidence intervals. Due to the low number of RT patients, only univariate analysis was performed. For our LT patients, covariates with a p-value less than 0.10 in univariate analysis were introduced into a multivariate model (Wald Chi-Square). The potential predictors “graft rejection” and “CMV (cytomegalovirus)” were not introduced in multivariate model due to their interaction with “glucocorticoids” and “infection” respectively. All p-values were two-sided/2-tailed and values less than 0.05 was considered statistically significant. All statistical analyses were performed with Stata® V17 software (Statacorp, Texas, USA).

Characteristics of our rDIABGRAFT LT and RT cohorts are summarized in Tables 1, 2 (Supplementary Tables S1: LT countries and S2: LT pathologies), respectively. We collected data from 195 pediatric patients treated at CUSL with liver transplantation (LT) (Figure 2). The median age of liver recipients was 18 months (10; 36) and the majority (179/195) received a liver from a living donor. All patients were treated lifelong with tacrolimus and 65% (126/195) were temporally treated with glucocorticoids. Regarding acute complications, 44% (86/195) were diagnosed with a viral infection (nCMV = 55/195; nEBV = 42/195) and approximatively half (104/195) of our total LT cohort presented graft rejection, the majority of which was treated with glucocorticoids (91/104) whereas five patients (2.6%) required a second transplantation. About our renal transplantation (RT) cohort, we collected data about 20 pediatric patients (Figure 3). The median age of renal recipients was 12 years (9; 15), seven patients (35%) received a renal transplant from a living donor, all patients were treated with tacrolimus and glucocorticoids after the transplantation, and fourteen (70%) were still under both treatment at the end of data collection. For acute complications, fourteen (70%) presented infection: nine (45%) were diagnosed with a viral infection (nCMV = 4/20 and nEBV = 7/20) and seven (35%) presented bacterial infection. Seven (35%) patients presented a confirmed or borderline graft rejection for which they received shots or/and increased doses of glucocorticoids. Two patients (10%) were re-transplanted.

Out of 195 LT pediatric patients, 25.1% (49/195) developed transient HG (Figure 2 and Table 3) and for most of them (92%) HG appeared during the first two weeks after transplantation (Figure 4 and Supplementary Figure S1). No overt diabetes was observed but a third (16/49) of our HG-positive LT cohort was treated with insulin.

In univariate analysis, the use of glucocorticoids (OR 2.64 95% CI, 1.23–5.71) and the presence of critical condition such as graft rejection (OR 3.18 95% CI, 1.56–6.48) and viral infection (OR 2.54 95% CI, 1.31–4.93), in particularly Cytomegalovirus (OR 2.79 95% CI, 1.41–5.53) were significantly associated with the onset of HG as shown in Table 4. After adjustment with multivariate logistic regression analysis (Wald Chi-Square tests), incidence of transient HG after LT was higher in children who received glucocorticoids (2.96, 95% CI, 1.32–6.61) and presented a viral infection (OR 2.20, 95% CI, 1.09–4.44) (Table 4).

As we observed with rDIABGRAFT that our LT cohort presented transient HG early after transplantation (i.e., 0–14 days), we performed dynamic testing close to transplantation (day 14 and day 30) in four LT children. Table 5 presents the patients, treatments and characteristics of the pDIABGRAFT LT cohort.

All patients presented fasting glucose and c-peptide level in normal range (Table 5) whereas glucose sensor placed at day 14 post-LT for one month showed chronic HG occurring in postprandial afternoon period (Figure 5). Parallelly, all children received high doses of glucocorticoids for graft rejection and required insulin. Moreover, during the OGTT performed at one-month post-LT (n = 3), all presented IR (HOMA-IR > 1.7) while in two of them, glycemia peaked respectively at 212 and 250 mg/dl at 120′ (Figure 6). We thus observed in our LT cohort two patients with diabetes at one-month post-LT.

Out of our 20 pediatric patients with renal transplantation, 55% (11/20) presented HG (Table 6). Out of eleven patients with HG, four of them developed overt diabetes (20% of total cohort, 36% of HG cohort), still treated at the end of data collection with antidiabetic medication (oral antidiabetics in 2/4 and a combination of oral antidiabetics and insulin in 2/4). The remaining seven patients (35% of total cohort, 64% of HG cohort) presented HG without overt diabetes, during a median duration of seven days (6; 12) and four of them (57%) required insulin during a median duration of four days (2; 8).

No precise timing for developing HG was observed with our RT pediatric patients (Supplementary Figure S2), but a concomitance with the occurrence of critical events such as graft rejection and infection has been observed. Indeed, univariate analysis (Likelihood Ratio) was performed to evaluate the association between risk factors and HG, and our analysis showed that graft rejection (OR 14.0, 95% CI, 1.25–156.61) and infections post RT (OR 12.5 95% CI, 1.09–143.43) were significantly associated with a higher occurrence of HG (Table 7). All our patients with a re-transplantation or bi-organ transplantation (4/20; two second RT and two previous LT) presented chronic HG but logistic regression was not possible because no patient in the HG-free RT cohort required another transplant.

For our LT and RT pediatric patients, there was no difference between occurrence of HG and gender of patient, history of overweight/obesity, BMI (pre- and post-transplant) and the use of glucocorticoids before the transplantation, donor status (cadaveric or living), pathology requiring the transplant, weight gain or family history of diabetes.

For our RT cohort, as we did not observe a specific moment of HG occurrence but more a concomitance with the presence of critical events, we analyzed the evolution of glucose over time (at one-, three-, six- and nine-months post RT). Then, we followed four RT pediatric patients (Table 8). One RT patient (TR1) disagreed to use glucose sensor at one month post RT.

Our analyses showed that all our RT pediatric patients presented normal fasting glycemia and HbA_1C levels from all the post-transplant follow-up period (up to 9 months) (Table 8). Moreover, glucose sensor (placed after 2 weeks post-RT) data showed HG in the afternoon as illustrated in Figure 7, with data regrouped on 24 h. OGTT performed at one-month post RT showed that two patients presented IGT with glycemia above 140 mg/dl at the end of the test, suggesting prediabetes, and one presented glycemia above 200 mg/dl (TR1: 221 mg/dl) at 120′, corresponding to overt diabetes (Figure 8). Dosage of pro-insulin and C-peptide showed that no patient presented a β-cell dysfunction whereas HOMA-IR showed severe IR (HOMA-IR >2.4) for all our RT patients (Table 8). At three-, six- and nine-months post-RT, all patients had normalized their glycemia at the end of the test (<140 mg/dl) (Figure 8), but they had continued to present moderate and severe IR except one at nine-month (Table 8).