- Top of page
- Materials and Methods
A retrospective analysis of 381 pediatric heart-transplant recipients was performed to determine the frequency, characteristics, and risk factors for post-transplant diabetes. The rate of post-transplant diabetes was 1.8% with antithymocyte globulin, cyclosporine and azathioprine as primary immunosuppressive therapy. Time from transplant to diabetes was 0.25–13 years. Diabetes was characterized by reversibility, and lack of insulinopenia and autoimmunity. The post-transplant diabetes rate in tacrolimus-converted children (n = 45) was 8.8%. In tacrolimus-converted children, age at transplant, mean and maximum tacrolimus blood levels, and first-year rejection episodes were higher in the post-transplant diabetes group, which also consistently had DR-mismatched transplants and HLA DR3/DR4 haplotypes. Body mass index was not different between diabetic and control tacrolimus-converted children. In conclusion, pediatric post-transplant diabetes may be related to reversible insulin resistance. Tacrolimus levels, HLA DR mismatch, and older age at transplant may predispose to post-transplant diabetes.
- Top of page
- Materials and Methods
Post-transplant diabetes mellitus (PTDM) is now a well-recognized complication of solid organ transplantation in both adults and children (1–3). Overall reported frequencies range from 4 to 40% (4), depending on the transplanted organ, definition of diabetes, and immunosuppressive regimen (5). Risk factors for PTDM include: tacrolimus use (6), age at transplant, obesity, family history of diabetes, pre- and post-transplant glucose intolerance, ethnicity, and occasionally HLA subtypes (7). Post-transplant diabetes is receiving more focused attention because of speculation that it may be a predictor of graft loss, as well as a prelude to significant chronic complications associated with diabetes in a compromised host (8). Despite the observed association of PTDM with use of glucocorticoids (prednisone), tacrolimus (FK506), and less commonly, with cyclosporine (9), the relation of PTDM to various immunosuppressive medications has not been fully elucidated. Moreover, the glycemic, autoimmune and HLA characteristics of children and adolescents developing PTDM have not been adequately described. The objectives of this study were to determine the frequency and characteristics of PTDM in a large cohort of pediatric heart-transplant recipients, and to elucidate the risk factors for PTDM in children treated with tacrolimus.
- Top of page
- Materials and Methods
The majority of patients in this series were transplanted during infancy (Table 1). Of 381 heart transplant recipients, seven patients developed diabetes, making the overall rate of PTDM 1.8% (Table 2). The median time for diabetes development was 7 years, with a wide range (0.25–13 years). If the number of living patients at 7 years post transplant were used as the denominator, the PTDM rate would be 3.6%. As a small risk of developing diabetes persists for the lifetime of a patient on immunosuppression, an alternative analysis was performed looking at the probability of freedom from diabetes. For the entire group of 381 initially started on cyclosporine, the probability of freedom from diabetes was 99% at 1 and 7 years, 97% at 10 years, and 94% at 15 years post transplant. When patients subsequently treated with tacrolimus were censored out and analyzed separately, the probability of freedom from diabetes was 93% at 1 year, and 90% at 7 years of tacrolimus therapy. The first-year rejection rate in the cyclosporine-based regimen was 0.49 per 100 days of patient follow-up. The probability of freedom from rejection was 31% at 1 year, 19% at 7 years, and 18% at 10–14 years post transplant. Freedom from coronary artery disease (chronic rejection) was 78% at 10 years post transplant.
Table 1. Demographics of pediatric heart transplant recipient patients at the time of transplant
|Infants (< 1 y)||267||70.1%|
|Children (1–12 y)||95||24.9%|
|Adolescents (13–18 y)||19||5.0%|
Table 2. Profile of patients with post-transplant diabetes
| ||Pt 1||Pt 2||Pt 3||Pt 4||Pt 5||Pt 6||Pt 7|
|Age (year) at diabetes onset||13.5||14.3||11.3||13.3||11.1||12.8||13.6|
|Years from transplant to diabetes||13||10||0.5||0.25||10||7||1.5|
|Family history of diabetes (1st/2nd-degree relative)||+/+||+/+||–/–||–/–||–/–||–/+||–/+|
|BMI at diabetes onset (kg/m2)||35||22.5||14.7||19.4||15||25.9||24.3|
|Immunosuppressants at diabetes onset||Cy,Az||Cy,Az||Cy,My,Pred||Ta,My,Pred||Ta,My||Ta,Az,Pred||Ta,My,Pred|
|Steroid bolus at diabetes onset||–||+||+||–||–||–||+|
|HbA1C% at diabetes onset||6.6||9.5||8.0||11.6||11.4||6.1||8.1|
|Glucose (mg/dL)||245||700||348||952||600||550|| |
|Diabetes treatment: Insulin/Oral agent||–/+||+/+||–/+||+/–||+/–||+/–||+/–|
|DM duration (months)||5||12||10||6||5||8||2 (deceased)|
Four of 45 patients developed diabetes after the institution of tacrolimus therapy (Table 2). Three children developed diabetes within 4 months of tacrolimus treatment, including one who was given a steroid pulse concomitant with starting tacrolimus, and two who were on maintenance steroids. The fourth patient developed diabetes after 27 months of starting tacrolimus. In the latter case, there was no concomitant steroid pulse nor a family history of diabetes at the time of diagnosis. Three children developed diabetes without a history of tacrolimus treatment; all had been on cyclosporine, and two had received bolus glucocorticoids at the time of diagnosis of diabetes. The third case was outstandingly obese with a body mass index of 35 kg/m2, the highest among all seven PTDM cases, at the time of diagnosis of diabetes.
Age at onset of diabetes was 11–14.3 years, BMI range was 14.7–35 kg/m2, and median BMI was 22.5 kg/m2. Four of seven patients had a family history of diabetes (type 2). A positive family history for type 2 diabetes was present in second-degree relatives in four cases, and in first-degree relatives in two cases who did not have post-tacrolimus diabetes. None of the cases had a family history of type 1 diabetes. Plasma glucose at diagnosis of diabetes ranged from 245 to 952 mg/dL, and HbA1C range was 6.1–11.6%. Only one of seven patients presented in ketoacidosis, with blood glucose of > 500 mg/dL, pH of 7.05, bicarbonate level of 5 mmol/L, and HbA1C 11.4%. Six of seven patients with PTDM required insulin therapy at diagnosis, and one was started on metformin, an oral hypoglycemic agent. Treatment was discontinued in five of seven cases, with no additional diabetes therapy, apart from one case who has reactive airway disease and requires insulin periodically in relation to transient intravenous steroid-associated hyperglycemia. The seventh patient with PTDM required insulin, had been on tacrolimus and intravenous solumedrol at the time of diagnosis, and died from chronic rejection and post-transplant coronary artery disease within 2 months of diabetes onset.
Islet cell, GAD, and insulin autoantibodies were negative in all PTDM patients tested (n = 5), and C-peptide levels were 2.1–4.8 ng/mL, normal range being 0.4–2.2 ng/mL. Five patients (of seven) developed diabetes in conjunction with glucocorticoid use. Four of seven patients were positive for either HLA DR3 or DR4, and three were negative for both. The time from transplant to onset of diabetes ranged from 3 months to 13 years. Daily home blood glucose monitoring and quarterly HbA1C measurements were used to monitor diabetes progression or resolution and adjust treatment accordingly. Once both parameters normalize, insulin or oral hypoglycemics are tapered, then discontinued. Quarterly random glucose measurements on all patients, and HbA1C measurements in TTC or PTDM patients, are routinely performed on follow-up. Diabetes duration range in this series was 2–12 months, with evidence of reversal in six cases, and death from an unrelated cause within 2 months in one case.
The rate of PTDM in tacrolimus-treated children (TTC) (n = 45) was 8.8%. When TTC with and without PTDM were compared (n = 4 and 41, respectively; Table 3) median age at transplant was 10.8 vs. 2.5 years (p = 0.1). Gender, ethnicity, and family history of diabetes were not significantly different. Mean tacrolimus blood levels were 16 vs. 13 ng/mL, and maximum blood levels were 31 vs. 21 ng/mL (p = 0.05). Seventy-five per cent vs. 27% were CMV-positive pretransplant (p = 0.047). The average number of rejection episodes in the first year post transplant was 4 vs. 2.68, and the average time of transplant to first rejection episode was 10.7 vs. 126.5 days. One hundred per cent vs. 50% received transplants where both HLA DR alleles did not match (p = 0.07). One hundred per cent vs. 61% were positive for HLA DR3 and/or DR4. Seventy-five per cent vs. 63% had transient hyperglycemia with a blood glucose over 200 mg/dL in the first week following the transplant. Zero per cent vs. 12% had a BMI z-score below − 2SD, i.e. body mass index was less than two standard deviations below the average for age and gender, at the start of tacrolimus therapy. Solumedrol bolus and/or prednisone maintenance or taper exposure frequencies were higher in the PTDM group. The small number of PTDM patients in this group (4 vs. 41) would not yield statistically significant results for some obvious candidate culprits (e.g. number of rejection episodes). Tacrolimus was routinely discontinued upon recognition of diabetes, and the patients were switched to an alternative, steroid-free regimen.
Table 3. Characteristics of tacrolimus-treated children with and without post-transplant diabetes
|Median age at transplant (year)||10.8||2.5|
|Males : females||3 : 1||28 : 13|
|Reason for transplant: CCHD/CM||1 : 3||22 : 19|
|Duration of tacrolimus therapy (months)||0.7–27.6||26–81|
|Mean tacrolimus blood level (ng/mL)||16||13|
|Maximum tacrolimus blood level||31||21|
|First year rejection episodes||4||2.68|
|Days from transplant to first rejection||10.7||126.5|
|CMV positivity pretransplant||75||27|
|HLA DR mismatched transplant||100%||50%|
|HLA DR3/4 positive||100%||61%|
|Acute post-transplant hyperglycemia||75%||63%|
|BMI z-score < − 2 SD at tacrolimus start||0%||12%|
Because of the small number of patients with PTDM, with or without a history of tacrolimus use (4 vs. 3, respectively), statistical comparison was not conducted between the two groups, as analysis of such small numbers was unlikely to unravel significant trends.
- Top of page
- Materials and Methods
The results of this study contrast with those of other pediatric series (3–6) in that they suggest that post-transplant diabetes is a relatively rare complication in children. This in part reflects differences in the immunosuppression regimen. Many of the reports of an increased incidence of pediatric post-transplant diabetes were based on series where steroids and/or tacrolimus were used as primary immune suppressants (4–6). In this study, steroid avoidance at the outset, and tacrolimus use only in recalcitrant rejection, might have contributed to a better outcome. Moreover, the transplanted population in this series has a relatively large neonatal group; hence the PTDM rates cannot be extrapolated to centers dealing primarily with older children. The PTDM rate in tacrolimus-treated children was comparable to other series (5), albeit for the reversibility of hyperglycemia in all patients which may be a function of immediate discontinuation of tacrolimus upon detection of diabetes. The apparent earlier and more frequent occurrence of rejection in tacrolimus-treated children who develop diabetes may reflect a predominant role for glucocorticoids in the induction of post-transplant diabetes. There may be a synergistic effect between tacrolimus and glucocorticoids. Alternatively, a threshold phenomenon may be operative, such that complete withdrawal rather than weaning of either agent may be necessary to reverse diabetes. It is possible to make an argument for tacrolimus (vs. cyclosporine) as a culprit in PTDM from this study, in which the relative risk for tacrolimus-associated diabetes is 8.8% (4 of 45), compared to 0.08% for cyclosporine (n = 3 of 381; p < 0.0001). However, this cannot be confirmed since there were no comparable patients managed with tacrolimus from the time of transplantation. Thus, the design of the study precludes extrapolation of results to series in which tacrolimus was used as primary immunosuppression. The relatively rare occurrence of PTDM in this series makes it difficult to compare the relative contributions of tacrolimus and cyclosporine (4 of 45 vs. 3 of 381). In PTDM with either drug, there was an approximately 70% frequency of concomitant steroid use, which may suggest that steroids are the primary diabetogenic agent (13). However, the scores of children who received glucocorticoids in some form during cyclosporine treatment without developing diabetes point to the contrary. Moreover, the comparable cumulative steroid exposure frequencies of tacrolimus-treated children with and without diabetes suggest that steroids alone are not the primary causative factor of post-transplant diabetes.
One of the limitations of our study is that it is based in a large international program in which long-term follow-up data may not be available on all patients beyond 1 year of transplantation. Despite the vigilance of the transplant team in maintaining accurate follow-up data on all patients by telephone contact and annual record updates, diabetes may be missed, especially if asymptomatic. Thus, under-reporting of the incidence of diagnosed PTDM would be least likely in the first year after transplant when the transplant team is more pro-actively involved. Long-term follow-up regarding compliance with the immunosuppressive regimen and annual blood glucose testing may similarly be incomplete, compromising the accuracy of the above results. Moreover, the frequency of post-transplant diabetes in this study was based on the total number of transplanted patients, without consideration for postoperative mortality. However, the total number of 1-month pediatric heart-transplant survivors through June 2002 was 350, which would not dramatically affect the overall frequency reported.
To further understand the pathogenesis of post-transplant diabetes, our protocol included testing for endogenous insulin production and insulin resistance by measuring plasma glucose and C-peptide (the connecting peptide of the two endogenous insulin chains). The normal or elevated C-peptide levels suggest that insulin resistance rather than insulin deficiency was the underlying mechanism in most cases (14). The reversibility of hyperglycemia in all cases minimizes the possibility of permanent beta-cell damage. Long-term determination of the frequency of diabetes in tacrolimus-treated patients may further clarify its association with PTDM, and the mechanism thereof.
The mechanism by which tacrolimus may lead to diabetes is a complex one, where islet cell-specific autoimmunity, insulinopenia and insulin resistance have been suggested (15). Experimental data in animal models suggest a possible diabetes-preventive role for tacrolimus, although this effect has been dose-dependent (16). Clinical data have been difficult to interpret due to variability of reported pediatric tacrolimus-associated PTDM definition and frequency, ranging up to 100% in some series (17), and the tendency to use insulin at diabetes onset in most cases, particularly in the context of ketoacidosis (18,19). That such data have been reported in the context of varying regimens of concomitant potential diabetogenic immunosuppressants poses an additional challenge to deciphering the role of tacrolimus alone in the predisposition to diabetes. It is of note that autoimmune diabetes with development of anti-GAD antibodies has occasionally been described in tacrolimus-treated transplanted patients with a diabetes-susceptible HLA haplotype (20). This suggested a beta-cell toxic effect for tacrolimus. Despite the presence of autoimmune diabetes-predisposing HLA subtypes in some patients in our study, C-peptide results and lack of diabetes-related autoantibodies in all PTDM patients tested make tacrolimus-induced autoimmune beta-cell injury highly unlikely. Treatment of PTDM in our series mirrors most others', with a predominant use of insulin at onset due to ketoacidosis or severe hyperglycemia, in which glucotoxicity would interfere with the action of oral hypoglycemic agents.
As more pathogenetic mechanisms are unraveled for diabetes in general, it is now evident that disturbance of glycemic control cannot be viewed as a single time-point occurrence, but rather a dynamic spectrum amenable to acceleration by various stressors. Thus, long-term follow-up is crucial in determining the true incidence of pediatric post-transplant diabetes in our series and others. Current screening of annual random blood glucose measurement may need to be modified to include hemoglobin A1C measurement and basal and stimulated measures of glucose-induced insulin response. Moreover, results pertaining to diabetes in our study are limited by the lack of comprehensive pretransplant glucose tolerance evaluation, especially since insulin-resistant diabetes is asymptomatic in approximately 50% of patients. As diabetes was reversible in all cases, monitoring for diabetes-related chronic complications (neuropathy, nephropathy, retinopathy) has not been undertaken.
The availability of information on PTDM in heart-transplant pediatric recipients from multiple centers, and the development of standardized protocols to define, diagnose, and monitor PTDM might elucidate further immunosuppressive links. Moreover, such data might yield itself to logistic regression analysis, allowing the prediction of post-transplant diabetes in pediatric heart-recipients (7), thus facilitating PTDM prevention.