Significance Statement
Sustained (or overt) diabetes mellitus after kidney transplantation is strongly associated with hyperglycemia during the early perioperative period. In a multicenter trial with 263 participants randomized to strict blood glucose monitoring and an early basal insulin intervention versus control (a more liberal approach consisting of sporadic corrections of hyperglycemia and otherwise oral antidiabetics), overt post-transplantation diabetes mellitus (PTDM) was ascertained by antidiabetic treatment and an oral glucose tolerance test (2 hour glucose ≥200 mg/dl). The intervention resulted in modestly reduced PTDM rates at 12 months and 24 months at the cost of higher rates of hypoglycemia. In a per-protocol analysis that excluded protocol violators and accounted for baseline differences in polycystic kidney disease, the reduction in PTDM at 12 months was significant, suggesting the approach merits further study.
Keywords: kidney transplantation, diabetes mellitus, clinical trial, diabetes, organ transplant, randomized controlled trials, renal transplantation, hyperglycemia, cardiovascular
Visual Abstract
Abstract
Background
Post-transplantation diabetes mellitus (PTDM) might be preventable.
Methods
This open-label, multicenter randomized trial compared 133 kidney transplant recipients given intermediate-acting insulin isophane for postoperative afternoon glucose ≥140 mg/dl with 130 patients given short-acting insulin for fasting glucose ≥200 mg/dl (control). The primary end point was PTDM (antidiabetic treatment or oral glucose tolerance test–derived 2 hour glucose ≥200 mg/dl) at month 12 post-transplant.
Results
In the intention-to-treat population, PTDM rates at 12 months were 12.2% and 14.7% in treatment versus control groups, respectively (odds ratio [OR], 0.82; 95% confidence interval [95% CI], 0.39 to 1.76) and 13.4% versus 17.4%, respectively, at 24 months (OR, 0.71; 95% CI, 0.34 to 1.49). In the per-protocol population, treatment resulted in reduced odds for PTDM at 12 months (OR, 0.40; 95% CI, 0.16 to 1.01) and 24 months (OR, 0.54; 95% CI, 0.24 to 1.20). After adjustment for polycystic kidney disease, per-protocol ORs for PTDM (treatment versus controls) were 0.21 (95% CI, 0.07 to 0.62) at 12 months and 0.35 (95% CI, 0.14 to 0.87) at 24 months. Significantly more hypoglycemic events (mostly asymptomatic or mildly symptomatic) occurred in the treatment group versus the control group. Within the treatment group, nonadherence to the insulin initiation protocol was associated with significantly higher odds for PTDM at months 12 and 24.
Conclusions
At low overt PTDM incidence, the primary end point in the intention-to-treat population did not differ significantly between treatment and control groups. In the per-protocol analysis, early basal insulin therapy resulted in significantly higher hypoglycemia rates but reduced odds for overt PTDM—a significant reduction after adjustment for baseline differences—suggesting the intervention merits further study.
Clinical Trial registration number: NCT03507829
Data from kidney,1–8 liver,9–12 lung,13,14 and heart15 transplant recipients indicate diabetes mellitus (DM) diagnosed after transplantation is associated with mortality and cardiovascular events. Post-transplantation DM (PTDM) is a unique form of diabetes,16–18 which has been best studied in kidney transplant recipients (KTRs).19,20 KTRs develop early post-transplant hyperglycemia because the abrupt reversal of kidney failure generates higher insulin demand, partly through increased insulin clearance, but also as a result of individual risk factors and transplant-specific mechanisms such as perioperative stress, adding to the diabetogenicity of immunosuppressive therapy (glucocorticosteroids and tacrolimus).21
Hyperglycemia has been shown to be highly prevalent during the first week after kidney transplantation,22 and to be a stronger predictor of overt PTDM at 1 year post-transplant than individual risk factors, such as age and body mass index.7 In an earlier proof-of-concept trial, we assessed whether postoperative hyperglycemia in previously nondiabetic KTRs could be controlled using basal insulin therapy, and found this intervention reduced the odds for PTDM throughout 12 months post-transplant.23 One trial strategy in the intervention group was to treat KTRs with basal insulin according to their afternoon glucose levels, which are generally higher than fasting if glucocorticosteroids are administered in the morning.24
In our earlier trial,23 an afternoon glucose threshold of 140 mg/dl for initiation of basal insulin had the consequence that all treatment group participants received therapy, whereas the control group followed a “watch-and-wait” approach for developing diabetes, in accordance with previous guidelines.25 This multicenter trial systematically investigated, with higher sample size, whether PTDM can be prevented by initiating basal insulin therapy for early postoperative afternoon glucose ≥140 mg/dl.
Methods
Study Design
The Insulin Therapy for the Prevention of New Onset Diabetes after Transplantation trial was an investigator-initiated, open-label, randomized multi-center clinical trial with an unblinded end point evaluation, performed at four transplant centers (Medical Universities of Vienna and Graz, Austria; Hospital del Mar Barcelona, Spain; Charité Universitätsmedizin Berlin, Germany). Adult nondiabetic KTRs (absence of diabetes defined according to American Diabetes Association [ADA] guidelines) receiving standard triple immunosuppressive treatment with tacrolimus, mycophenolate, and glucocorticosteroids were eligible. Induction therapy was allowed per center practice, mycophenolate dose was adjusted to center protocol and adverse effects. Tacrolimus and glucocorticosteroids were tapered according to a predefined scheme (Supplemental Methods, Supplemental Methods Table 3). The study was undertaken with independent external monitoring in accordance with International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use-Good Clinical Practice principles, and the Declaration of Helsinki. Written informed consent was obtained from all patients after approval from the institutional review board at each participating center.
Randomization and Masking
Randomization into the basal insulin intervention versus standard-of-care group was done at a blocked randomization ratio of 1:1 for each transplant center, where sealed envelopes were provided, stratified for first versus repeated transplant. Study inclusion took place before transplantation. Inclusion and randomization shortly after transplantation was not permitted.
Study Interventions
In the basal insulin treatment group, capillary blood glucose monitoring four times daily (fasting, prelunch, afternoon [presupper], evening [postsupper]) started immediately after transplantation. During their postoperative hospitalization, participants were to be trained in glucose self-monitoring after discharge, using glucometers and test strips (Accu-Chek Performa, Roche). Once the postoperative afternoon glucose value was >140 mg/dl, insulin therapy with intermediate acting (neutral protamine hagedorn) insulin (human insulin isophane, Humulin N [Eli Lilly], referred to as “basal” insulin), was to be introduced the next morning, according to a prespecified insulin titration regimen (Supplemental Methods Table 1), targeting a subsequent afternoon capillary glucose level of 110 mg/dl. Short-acting insulin (insulin lispro, Humalog, Eli Lilly) additionally was to be used for correcting prelunch and bedtime hyperglycemia. A prespecified dose adjustment scheme was used for standardized, adaptive basal insulin titration (Supplemental Methods Table 1).
The prespecified trial procedures for the standard-of-care control group were also very detailed (see Supplemental Methods). Briefly, no routine postoperative glycemia monitoring other than once daily fasting plasma glucose was planned. Participants with glucose >200 mg/dl were subsequently monitored and, if hyperglycemia persisted, treated with short-acting insulin according to a sliding scale during their in-hospital stay (Supplemental Methods Table 2). If permanent antihyperglycemic medication became necessary, sulfonylureas were suggested as treatment of choice. Upon discharge, only participants who needed insulin were instructed for glycemia monitoring at home.
Outcome Measures
The prespecified primary end point was occurrence of PTDM at month 12 post-transplant. PTDM was initially defined according to the ADA diagnostic criteria, which had endorsed an hemoglobin A1c (HbA1c) cutoff ≥6.5%26,27 when the trial was planned. Due to a subsequent consensus guideline emphasizing the oral glucose tolerance test (OGTT) as gold standard for PTDM diagnosis,28 we used HbA1c only in participants who missed or refused their OGTT, otherwise relying on OGTT-derived 2 hour plasma glucose for diagnosing PTDM. OGTT-derived 2 hour plasma glucose was also used for determining impaired glucose tolerance. Participants on antihyperglycemic therapy (insulin or oral antidiabetics) were classified as “patients with diabetes on treatment.”
Prespecified secondary end points included PTDM at month 24, glycemic control throughout the trial period (by HbA1c), capillary blood glucose profiles during the early postoperative in-hospital stay and at all trial visits, kidney function measured by serum creatinine, and patient and graft survival. The most important prespecified safety end point was hypoglycemia, which was considered a serious adverse event if measured capillary blood glucose equaled or fell <40 mg/dl, or if the hypoglycemic event necessitated inpatient hospitalization or prolongation of current hospitalization, in analogy to a large previous trial.29
Participants in the treatment group who were never prescribed basal insulin despite recorded indication (n=3), or refused therapy (n=1), were excluded from the per-protocol population. In addition, participants not following the protocol’s prespecified insulin weaning procedure, despite normalization of glucose metabolism (proven by nondiabetic 2 hour plasma glucose at their OGTT or nondiabetic HbA1c at preceding visits), were also excluded from the per-protocol analysis (n=4; see Results, Figure 1 and Supplemental Table 1, and note [1] that these participants on insulin therapy were wrongfully classified as “patients with diabetes on treatment” that affected the primary end point in the per-protocol analysis; and [2] that these participants were later on [after month 12] taken off their insulin treatment anyhow, as documented in detail in Supplemental Table 1).
Figure 1.
Consolidated standards of reporting trial (CONSORT) flow chart of the trial. Structural outline of the Insulin Therapy for the Prevention of New Onset Diabetes after Transplantation (ITP-NODAT) trial. Reasons for trial discontinuation are listed at the bottom of the figure, for control versus treatment group participants. ITT, intention-to-treat; PP, per-protocol.
Statistical Analyses
The sample size and power calculations revealed that 276 participants would have 85% power to detect a 15% difference in the proportions of participants developing PTDM, assuming the baseline PTDM incidence in the control group would be 30% at month 12.30,31 The power calculation was on the basis of two-sided significance testing with an alpha level of 0.05 (for further details, see the study protocol in Supplemental Methods).
Depending on their distribution, numerical variables were summarized as mean±SD, or medians with interquartile ranges. We used Student’s t test and Mann–Whitney U for between-group comparisons of normally and non-normally distributed numerical variables; and Chi-squared and Mann–Whitney U for nominal and ordinal categorical variables, respectively. We used logistic regression to estimate the odds for requirement of antidiabetic medication or development of PTDM in the treatment versus control group, adjusting for significant baseline differences (polycystic kidney disease and glomerular disease, as specified in the Results). All analyses were carried out for the intention-to-treat and per-protocol population, which excluded 42 dropouts and eight protocol violations. If variables relevant to the logistic regression analyses were missing, the corresponding observations had to be dropped. A P value <0.05 was considered statistically significant.
Data from all treatment group participants were also utilized in an exploratory analysis assessing the association between adherence to the early insulin protocol and PTDM outcomes. The statistical details of this analysis are mentioned in the respective figure legend.
For calculations and logistic regressions, we used IBM SPSS statistics version 25 and 26 for macOS (SPSS Inc., Chicago, IL, USA). Logistic regressions for exploratory adherence analyses were performed using STATA 12.0 (Statacorp LLC, College Station, TX, USA).
Results
Characteristics of the Trial Participants and Immunosuppressive Medication
The Insulin Therapy for the Prevention of New Onset Diabetes after Transplantation trial was performed from November 21, 2012 through May 22, 2018 and included 263 participants. In Figure 1 (modified CONSORT chart), we provide detailed descriptions of the participant flow. Participants’ baseline characteristics are shown in Table 1 (intention-to-treat and per-protocol). Significantly fewer control versus treatment group participants had polycystic kidney disease (9.2% versus 20.3%), whereas significantly more control versus treatment group participants had glomerular disease. In the control versus treatment group, there was also a tendency for lower prevalence of family history of diabetes (10.8% versus 18.0%) and more living donor transplants (23.8% versus 16.5%), whereas body mass index tended to be higher (26.5±5.5 versus 25.7±4.9 kg/m2). The immunosuppressive therapy that was initiated at baseline (Table 1) continued to be similar between control and treatment group participants through the entire follow-up (Supplemental Tables 2 and 3). Specifically, median tacrolimus trough levels in control versus treatment group participants were 7.8, 9.5, 7.1, 6.2, and 5.9 versus 7.5, 9.1, 7.0, 6.0, and 6.1 ng/ml on day 7, months 1, 6, 12, and 24, respectively (all P values are nonsignificant). Median cumulative glucocorticosteroid levels (in mg prednisone equivalent dose) in control versus treatment group participants were 965, 1380, 2602, 3500, and 4951 versus 965, 1373, 2584, 3510, and 4976 mg, respectively (all P values are nonsignificant).
Table 1.
Participant characteristics at baseline
| Characteristics at Baseline | Control (ITT) |
Treatment (ITT) |
Control (PP) |
Treatment (PP) |
P value (ITT) |
P value (PP) |
|---|---|---|---|---|---|---|
| Total patients, n (%) | 130 (49.4) | 133 (50.6) | 109 (51.2) | 104 (48.8) | — | — |
| Female, n (%) | 47 (36.2) | 54 (40.6) | 40 (36.7) | 42 (40.4) | 0.49 | 0.58 |
| Recipient age, yrs, mean (SD) | 50.7 (14.0) | 51.7 (14.3) | 50.2 (14.1) | 50.0 (14.4) | 0.58 | 0.93 |
| Height, cm, mean (SD) | 169.6 (9.4) | 169.1 (11.2) | 169.7 (9.4) | 169.9 (10.5) | 0.71 | 0.91 |
| Weight, kg, median (IQR) | 76.0 (65.7–85.4) | 70.4 (63.1–82.5) | 76.0 (65.2–85.3) | 70.0 (63–81.1) | 0.07 | 0.06 |
| BMI, kg/m2, mean (SD) | 26.5 (5.5) | 25.7 (4.9) | 26.3 (5.4) | 25.2 (4.7) | 0.18 | 0.11 |
| Family history of diabetes, n (%) | 14 (10.8) | 24 (18.0) | 10 (9.2) | 18 (17.3) | 0.07 | 0.05 |
| Chronic hepatitis C, n (%) | 1 (0.8) | 2 (1.5) | 1 (0.9) | 1 (1.0) | 1 | 1 |
| CMV antibody positive, n (%) | 85 (65.4) | 86 (64.7) | 73 (67.0) | 70 (67.3) | 0.96 | 0.78 |
| CMV high risk, n (%) | 23 (17.7) | 19 (14.3) | 21 (19.3) | 12 (11.5) | 0.45 | 0.13 |
| PRA highest ≥ 10%, n (%) | 9 (6.9) | 17 (12.8) | 9 (8.3) | 15 (14.4) | 0.09 | 0.12 |
| First graft, n (%) | 107 (82.3) | 111 (83.5) | 94 (86.2) | 92 (88.5) | 0.96 | 0.63 |
| Second graft, n (%) | 15 (11.5) | 17 (12.8) | 13 (11.9) | 11 (10.6) | 0.8 | 0.76 |
| >2 grafts, n (%) | 2 (1.5) | 1 (0.8) | 2 (1.8) | 1 (1.0) | 0.62 | 1 |
| Living donor, n (%) | 31 (23.8) | 22 (16.5) | 29 (26.6) | 21 (20.2) | 0.15 | 0.29 |
| Primary kidney disease, n (%) | ||||||
| Glomerular disease | 51 (39.2) | 40 (30.1) | 47 (43.1) | 34 (32.7) | 0.039 | 0.06 |
| Vascular disease | 12 (9.2) | 17 (12.8) | 9 (8.3) | 13 (12.5) | 0.47 | 0.37 |
| Tubulointerstitial disease | 9 (6.9) | 7 (5.3) | 9 (8.3) | 5 (4.8) | 0.47 | 0.27 |
| Polycystic disease | 12 (9.2) | 27 (20.3) | 10 (9.2) | 22 (21.2) | 0.021 | 0.02 |
| Unknown | 24 (18.5) | 29 (21.8) | 19 (17.4) | 21 (20.2) | 0.70 | 0.72 |
| Other | 2 (1.5) | 1 (0.8) | 2 (1.8) | 1 (1.0) | 0.61 | 1 |
| Comorbidities, n (%) | ||||||
| Cardiovascular | 44 (33.8) | 44 (33.1) | 38 (34.9) | 29 (27.9) | 0.8 | 0.53 |
| Respiratory | 7 (5.4) | 8 (6.0) | 5 (4.6) | 5 (4.8) | 0.78 | 0.75 |
| Urinary | 10 (7.7) | 7 (5.3) | 7 (6.4) | 7 (6.7) | 0.42 | 0.67 |
| Endocrinologic | 12 (9.2) | 12 (9.0) | 9 (8.3) | 8 (7.7) | 1 | 0.84 |
| Neurologic | 1 (0.8) | 4 (3.0) | 1 (0.9) | 3 (2.9) | 0.36 | 0.32 |
| Psychiatric | 2 (1.5) | 7 (5.3) | 2 (1.8) | 5 (4.8) | 0.16 | 0.23 |
| Other | 4 (3.1) | 2 (1.5) | 3 (2.8) | 2 (1.9) | 0.68 | 1 |
| Immunosuppression, n (%) | ||||||
| Tacrolimus | 125 (96.2) | 129 (97) | 109 (100) | 104 (100) | — | — |
| Mycophenolate mofetil | 64 (49.2) | 56 (42.1) | 55 (50.5) | 44 (42.3) | 0.14 | 0.16 |
| Mycophenolate sodium | 52 (40.0) | 67 (50.4) | 46 (42.2) | 55 (52.9) | 0.12 | 0.14 |
| Glucocorticosteroid | 125 (96.2) | 128 (96.2) | 109 (100) | 104 (100) | — | — |
| Basiliximab induction | 73 (56.2) | 82 (61.7) | 66 (60.6) | 66 (63.5) | 0.63 | 0.62 |
| Antithymocyte globulin | 8 (6.2) | 9 (6.8) | 8 (7.3) | 7 (6.7) | 0.93 | 0.86 |
| Other | 13 (10.0) | 15 (11.3) | 11 (10.1) | 14 (13.5) | 0.85 | 0.44 |
ITT, intention-to-treat population; PP, per-protocol population; IQR, interquartile range; BMI, body mass index; CMV, cytomegalovirus; PRA, panel reactive antibodies.
Hyperglycemia and Antihyperglycemic Therapy throughout Follow-up
As demonstrated in Figure 1, the population available for analysis differed at the prespecified time points for evaluation of diabetic outcome. By postoperative day 7, 112 of 128 participants in the treatment group (87.5%) had afternoon capillary blood glucose ≥140 mg/dl at least once. Moreover, 15 of 124 participants in the control group (12.1%) versus eight out of 128 participants in the treatment group (6.3%) had fasting plasma glucose levels ≥200 mg/dl at least at one study visit (P=0.12). The detailed course of the antihyperglycemic therapy in both groups is shown in Figure 2, for every postoperative day through month 24. In summary, 21 of 130 control group participants (16.2%) received some type of antihyperglycemic therapy (insulin or oral antidiabetics) at least once, whereas 117 of 133 treatment group participants (88.0%) were initiated on permanent basal insulin therapy.
Figure 2.
Antihyperglycemic therapy and PTDM incidence (framed for month 12 = contributing to the primary end point) in the control versus treatment groups. (A) and (D) Absolute numbers of participants who were treated with antihyperglycemics (insulin and/or oral antidiabetics) through 730 days of follow-up in the control group (dotted line, grey) versus in the treatment group (bold line). (B) and (E) ORs and their 95% confidence intervals for antidiabetic therapy in the treatment versus control group at months 6, 12, and 24. (C) and (F) Absolute numbers of participants who were treated with antidiabetics at months 6, 12, and 24.
PTDM at Month 12 (Primary End Point) and PTDM at Month 24
In the control group, more trial participants received antidiabetic therapy at months 12 and 24, compared with the basal insulin treatment group (9.5% and 10.1% control versus 6.1% and 2.7% treatment in the intention-to-treat population and 10.1% and 10.1% control versus 1.9% and 1.9% treatment in the per-protocol population; see Figure 2 and Table 2). When the OGTT results were added to the requirement for antidiabetic therapy, more trial participants in the control group had PTDM at months 12 and 24, compared with the basal insulin treatment group, but between-group differences became smaller (14.7% and 17.4% control versus 12.2% and 13.4% treatment in the intention-to-treat population; 15.6% and 17.4% control versus 6.7% and 10.6% treatment in the per-protocol population; see Figure 3 and Table 2).
Table 2.
Diabetic outcome, ORs, and numbers needed to treat for the intention-to-treat and the per-protocol population
| Variable | Control | Treatment | ORs (95% CI), NNT |
|---|---|---|---|
| ITT month 6, n | 123 | 120 | |
| Requiring antidiabetic therapy, n (%) | 9 (7.3) | 14 (11.7) | 0.67 (0.70 to 4.03), / |
| Not requiring antidiabetic therapy, n (%) | 114 (92.7) | 106 (88.3) | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 18 (14.6) | 21 (17.5) | 1.21 (0.61 to 2.42), / |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 102 (82.9) | 98 (81.7) | |
| Missing outcome data for diabetes, n (%) | 3 (2.4) | 1 (0.8) | |
| PP month 6, n | 109 | 104 | |
| Requiring antidiabetic therapy, n (%) | 9 (8.3) | 10 (9.6) | 1.18 (0.46 to 3.04), / |
| Not requiring antidiabetic therapy, n (%) | 100 (91.7) | 94 (90.4) | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 16 (14.7) | 16 (15.4) | 1.03 (0.49 to 2.20), / |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 90 (82.6) | 87 (83.7) | |
| Missing outcome data for diabetes, n (%) | 3 (2.8) | 1 (1.0) | |
| ITT month 12, n | 116 | 115 | |
| Requiring antidiabetic therapy, n (%) | 11 (9.5) | 7 (6.1) | 0.62 (0.23 to 1.67), 30 |
| Not requiring antidiabetic therapy n (%) | 105 (90.5) | 107 (93.0)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 17 (14.7) | 14 (12.2) | 0.82 (0.39 to 1.76), 44 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 97 (83.6) | 97 (84.3) | |
| Missing outcome data for diabetes, n (%) | 2 (1.7) | 4 (3.5) | |
| PP month 12, n | 109 | 104 | |
| Requiring antidiabetic therapy, n (%) | 11 (10.1) | 2 (1.9) | 0.18 (0.04 to 0.82), 13 |
| Not requiring antidiabetic therapy, n (%) | 98 (89.9) | 101 (97.1)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 17 (15.6) | 7 (6.7) | 0.40 (0.16 to 1.01), 12 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 90 (82.6) | 93 (89.4) | |
| Missing outcome data for diabetes, n (%) | 2 (1.8) | 4 (3.8) | |
| ITT month 24, n | 109 | 112 | |
| Requiring antidiabetic therapy, n (%) | 11 (10.1) | 3 (2.7) | 0.25 (0.07 to 9.1), 14 |
| Not requiring antidiabetic therapy, n (%) | 98 (89.9) | 108 (96.4)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 19 (17.4) | 15 (13.4) | 0.71 (0.34 to 1.49), 22 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 82 (75.2) | 91 (81.3) | |
| Missing outcome data for diabetes, n (%) | 8 (7.3) | 6 (5.4) | |
| PP month 24, n | 109 | 104 | |
| Requiring antidiabetic therapy, n (%) | 11 (10.1) | 2 (1.9) | 0.18 (0.04 to 0.82), 13 |
| Not requiring antidiabetic therapy, n (%) | 98 (89.9) | 101 (97.1)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 19 (17.4) | 11 (10.6) | 0.54 (0.24 to 1.20), 13 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 82 (75.2) | 88 (84.6) | |
| Missing outcome data for diabetes, n (%) | 8 (7.3) | 5 (4.8) |
95% CI, 95% confidence interval; NNT, number needed to treat; ITT, intention-to-treat population; PP, per-protocol population.
n=1 excluded due to missing data.
Figure 3.
Oral glucose tolerance tests outcomes at 6, 12, and 24 months after transplantation. (A) and (C) Diabetes (red) was defined by requirement for antidiabetic treatment, 2 hour plasma glucose ≥200 mg/dl or HbA1c ≥6.5% in participants who missed their oral glucose tolerance test, as specified in the Methods (n=18 control and n=9 treatment in month 6, n=13 control and n=13 treatment in month 12, n=25 control and n=20 treatment in month 24). Absence of diabetes (tan) was defined by 2 hour plasma glucose <200 mg/dl or HbA1c <6.5% in participants who missed their oral glucose tolerance test. Fasting plasma glucose ≥126 mg/dl in participants with 2 hour plasma glucose <200 mg/dl is shown above the tan part of the bar chart. Missing data for oral glucose tolerance test or HbA1c (white) are presented at the bottom, and a more detailed presentation of the glucose metabolism is provided in Supplemental Figure 1. (B) and (D) ORs and their 95% confidence intervals for diabetes in the treatment versus control group at months 6, 12, and 24 (framed for month 12 = primary end point). Diabetes was defined as described for (A) and (C) (red). (A) and (B) Intention-to-treat population. (C) and (D) Per-protocol population. Adjustment variables: polycystic kidney disease and glomerular disease. Statistically significant ORs are marked (bold).
PTDM outcomes are shown as odds ratios (ORs) for being on antidiabetic therapy in Figure 2 and as ORs for antidiabetic therapy or OGTT-derived 2 hour plasma glucose ≥200 mg/dl in Figure 3, for the basal insulin treatment versus the control groups. All ORs are shown unadjusted and adjusted for polycystic kidney disease and glomerular disease, because there were significant group differences at baseline. The primary end point (occurrence of antidiabetic therapy or OGTT-derived 2 hour plasma glucose ≥200 mg/dl at month 12 after kidney transplantation) is shown in Figure 3 and reached statistical significance only in the per-protocol population, and only when adjusted (unadjusted OR, 0.40; 95% confidence interval, 0.16 to 1.01, number needed to treat [NNT], 12).
Further details on the OGTT results (including impaired glucose tolerance) are provided in Supplemental Figure 1, along with additional ADA-based diabetes definitions (including, e.g., OGTT-derived fasting plasma glucose), which resulted in similar or overall better outcomes for the basal insulin treatment group.
Post-hoc Analysis of the Study Outcomes in a High-risk Population
Patients with family history of diabetes, polycystic kidney disease, age ≥60 years, or age 45–59 years plus (1) triglycerides ≥200 mg/dl; (2) triglycerides 150–200 mg/dl and body mass index >27 kg/m2; (3) triglycerides 150–200 mg/dl and HDL <40 mg/dl (men) or <50 mg/dl (women) were considered to be at high risk for PTDM (these definitions being on the basis of the literature).4,32–34 Patients fulfilling at least one of these criteria constituted the high-risk population (Supplemental Table 4). For the high-risk population, the results on (requirement for) antidiabetic therapy and diabetic outcome, per OGTT-derived 2 hour glucose are shown in Table 3, along with ORs and NNTs. Compared with the entire study population (Table 2), the NNTs were lower in the high-risk population (Table 3).
Table 3.
Diabetic outcome, ORs, and NNT for the high-risk population
| Variables | Control | Treatment | ORs (95% CI), NNT |
|---|---|---|---|
| ITT month, n | 65 | 73 | |
| Requiring antidiabetic therapy, n (%) | 9 (13.8) | 13 (17.8) | 1.35 (0.54 to 3.40), / |
| Not requiring antidiabetic therapy, n (%) | 56 (86.2) | 60 (82.2) | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 18 (27.7) | 19 (26.0) | 0.88 (0.41 to 1.87), / |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 45 (69.2) | 54 (74.0) | |
| Missing outcome data for diabetes, n (%) | 2 (3.1) | 0 (0.0) | |
| PP month 6, n | 54 | 60 | |
| Requiring antidiabetic therapy, n (%) | 9 (16.7) | 9 (15.0) | 0.88 (0.32 to 2.42), / |
| Not requiring antidiabetic therapy, n (%) | 45 (83.3) | 51 (85.0) | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 16 (29.6) | 14 (23.3) | 0.69 (0.30 to 1.59), / |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 36 (66.7) | 46 (76.7) | |
| Missing outcome data for diabetes, n (%) | 2 (3.7) | 0 (0.0) | |
| ITT month 12, n | 59 | 69 | |
| Requiring antidiabetic therapy, n (%) | 11 (18.6) | 7 (10.1) | 0.50 (0.18 to 1.39), 12 |
| Not requiring antidiabetic therapy, n (%) | 48 (81.4) | 61 (88.4)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 17 (28.8) | 12 (17.4) | 0.53 (0.23 to 1.22), 9 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 41 (69.5) | 55 (79.7) | |
| Missing outcome data for diabetes, n (%) | 1 (1.7) | 2 (2.9) | |
| PP month 12, n | 54 | 60 | |
| Requiring antidiabetic therapy, n (%) | 11 (20.4) | 2 (3.3) | 0.14 (0.03 to 0.65), 6 |
| Not requiring antidiabetic therapy, n (%) | 43 (79.6) | 57 (95.0)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 17 (31.5) | 5 (8.3) | 0.20 (0.07 to 0.59), 5 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 36 (66.7) | 53 (88.3) | |
| Missing outcome data for diabetes, n (%) | 1 (1.9) | 2 (3.3) | |
| ITT month 24, n | 54 | 67 | |
| Requiring antidiabetic therapy, n (%) | 11 (20.4) | 3 (4.5) | 0.19 (0.05 to 0.71), 7 |
| Not requiring antidiabetic therapy, n (%) | 43 (79.6) | 63 (94.0)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 18 (33.3) | 12 (17.9) | 0.41 (0.17 to 0.95), 6 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 31 (57.4) | 51 (76.1) | |
| Missing outcome data for diabetes, n (%) | 5 (9.3) | 4 (6.0) | |
| PP month 24, n | 54 | 60 | |
| Requiring antidiabetic therapy, n (%) | 11 (20.4) | 2 (3.3) | 0.14 (0.03 to 0.65), 6 |
| Not requiring antidiabetic therapy, n (%) | 43 (79.6) | 57 (95.0)a | |
| Diabetic patients (OGTT, therapy, HbA1c), n (%) | 18 (33.3) | 9 (15.0) | 0.32 (0.13 to 0.81), 5 |
| Nondiabetic patients (OGTT, therapy, HbA1c), n (%) | 31 (57.4) | 48 (80.0) | |
| Missing outcome data for diabetes, n (%) | 5 (9.3) | 3 (5.0) |
Patients with family history of diabetes; or polycystic kidney disease; or age ≥60 years; or age 45–59 years plus (1) triglycerides ≥200 mg/dl; (2) triglycerides 150–200 mg/dl and body mass index >27 kg/m2; (3) triglycerides 150–200 mg/dl and HDL <50 mg/dl (men)/<40 mg/dl (women) were considered to be at high-risk for PTDM (these definitions being on the basis of the literature4,32–34). Patients fulfilling at least one of these criteria constituted the high-risk population. 95% CI, 95% confidence interval; NNT, number needed to treat; ITT, intention-to-treat population; PP, per-protocol population.
n=1 excluded due to missing data.
Post-hoc Analysis of the Study Outcomes, by Sex
Sex-specific analyses of the study outcomes35 are provided in Supplemental Table 5. In the control group, lower proportions of male versus female participants were diagnosed as having PTDM (judged by OGTT results, and HbA1c, if needed, besides antidiabetic treatment) at months 6, 12, and 24. However, a clearly higher proportion of male versus female participants in the control group received antidiabetic treatment at month 6, and this tendency was maintained through month 24. In the treatment group, lower proportions of male versus female participants were again diagnosed as having PTDM at months 6, 12, and 24. When compared with the control group, the proportions of male versus female participants in the treatment group still receiving antidiabetic therapy at these time points were more similar. Consistent with these findings, almost all unadjusted ORs describing the effect of the early basal insulin intervention were lower, and closer to reaching statistical significance among male rather than among female participants.
Safety of the Basal Insulin Therapy
Two hypoglycemic episodes were recorded in the control group and 15 in the treatment group throughout the entire follow-up of the intention-to-treat population. These events occurred in two out of 130 (1.5%) versus 13 out of 133 (9.8%) participants of the respective trial groups (P=0.004), and the majority occurred within the first 3 months post-transplant (Table 4 and Supplemental Table 6). Five hypoglycemic events were not accompanied by symptoms. In total, 11 hypoglycemic events (seven with blood glucose 41–60 mg/dl and four with blood glucose ≤40 mg/dl) were described as mildly symptomatic (sweating/dizziness at maximum). The one remaining hypoglycemic event occurred in a treatment group participant who had escalated his insulin dose from eight units (on postoperative day 17) to 22 units (by postoperative day 25), without clear indication. This participant experienced an observed period of unconsciousness, which resolved promptly on administration of intravenous glucose by an emergency physician, called by the participant’s wife. The glucose value of this participant, measured on arrival of the emergency physician, was 36 mg/dl. Of these patients presenting with hypoglycemic events, one patient of the control group and three patients of the treatment group experienced a hypoglycemic event during the initial hospitalization period. There was no statistically significant difference in the length of in-hospital stay between patients with hypoglycemic events, when compared with patients without hypoglycemic events (Supplemental Figure 2), and between patients in the control group when compared with patients in the treatment group (Supplemental Table 7).
Table 4.
Hypoglycemic events and other safety end points in the intention-to-treat population
| Variables, n (%) | Month 3a | Month 6 | Month 12 | Month 24 | P value | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Control | Treatment | Control | Treatment | Control | Treatment | Control | Treatment | ||||||
| Hypoglycemic events (in total) | |||||||||||||
| Clinical episodeb | 1 (0.8) | 11 (8.3) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0.01 | / | / | / | |
| Blood glucose 41–60 mg/dl | 1 (0.8) | 9 (6.8) | 0 (0) | 1(0.8) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0.02 | 0.5 | / | / | |
| Blood glucose ≤40 mg/dl | 0 (0.0) | 5 (3.8) | 0 (0) | 0 (0) | 1 (0.9) | 0 (0) | 0 (0) | 0 (0) | 0.04 | / | 1.0 | / | |
| Patients with hypoglycemic events, | |||||||||||||
| Clinical episodeb | 1 (0.8) | 9 (6.8) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0.01 | / | / | / | |
| Blood glucose 41–60 mg/dl | 1 (0.8) | 8 (6.0) | 0 (0) | 1 (0.8) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0.02 | 0.5 | / | / | |
| Blood glucose <40 mg/dl | 0 (0.0) | 4 (3.0) | 0 (0) | 0 (0) | 1 (0.9) | 0 (0) | 0 (0) | 0 (0) | 0.06 | / | 1.0 | / | |
| Mortalityc | 0 (0) | 4 (3.0) | 0 (0) | 0 (0) | 1 (0.8) | 2 (1.7) | 2 (1.7) | 0 (0) | 0.12 | / | 0.62 | 0.50 | |
| Rejection | 8 (6.2) | 8 (6.0) | 0 (0) | 2 (1.7) | 5 (4.3) | 1(0.9) | 2 (1.8) | 2 (1.8) | 1.0 | 0.5 | 0.21 | 1.0 | |
| Graft lossc | 5 (3.8) | 2 (1.5) | 0 (0) | 1 (0.8) | 2 (1.6) | 0 (0) | 2 (1.7) | 2 (1.7) | 0.28 | 0.50 | 0.50 | 1.0 | |
| Hospitalization | 31 (23.8) | 35 (26.3) | 23 (18.7) | 23 (19.2) | 21 (18.1) | 14 (12.2) | 24 (22) | 18 (16.1) | 0.47 | 0.86 | 0.25 | 0.25 | |
| Kidney disorder | 15 (11.5) | 13 (9.8) | 6 (4.9) | 3 (2.5) | 2 (1.7) | 0 (0) | 3 (2.8) | 0 (0) | 0.68 | 0.32 | 0.5 | 0.11 | |
Data presented up to month 3 (percentages calculated from 263 participants).
The number of clinical episodes is lower than the total number of hypoglycemic events, because not all of the recorded low blood glucose values were accompanied by symptoms (please refer to Results section for further details).
The mortality and graft loss rate was calculated by using the number of participants who died or lost their graft in the numerator and the intention-to-treat population from the trial visit before in the denominator. P values were calculated accordingly.
Besides hypoglycemia, we observed no meaningful differences in adverse events, including mortality, between control and treatment group participants (Table 4, Supplemental Table 6). None of the deaths appeared to be related to the study treatment. HDL concentrations were higher in the treatment group at month 3, but not throughout the entire follow-up of 24 months (Table 5, Supplemental Table 8). We also observed similar rates of rejection episodes in both groups (Table 4), specifically, eight versus eight rejections in control versus treatment group participants up to month 3, an additional zero versus two rejections by month 6, additional five versus one rejections by month 12, and additional two versus two rejection episodes by month 24 (all P values are nonsignificant).
Table 5.
Weight, renal function, and nonglycemia labs: Intention-to-treat analysis
| Variables | Baseline | Month 12 | Month 24 | P value | |||||
|---|---|---|---|---|---|---|---|---|---|
| Control | Treatment | Control | Treatment | Control | Treatment | ||||
| Patients, n (%) | 130 | 133 | 116 | 115 | 109 | 112 | / | / | / |
| Female, n (%) | 47 (36.2) | 54 (40.6) | 44 (37.9) | 47 (40.9) | 40 (36.7) | 46 (41.1) | 0.49 | 0.65 | 0.51 |
| Weight, kg, median (IQR) | 76.0 (65.7–85.4) | 70.4 (63.1–82.5) | 77.7 (66.0–87.1) | 72.5 (65.5–83.3) | 79.1 (65.2–88.0) | 74.0 (66.0–84.3) | 0.07 | 0.12 | 0.13 |
| BMI, kg/m2, mean (IQR) | 26.5 (5.6) | 25.6 (4.8) | 26.9 (5.9.0) | 25.7 (4.6) | 27.4 (6.2) | 26.0 (4.5) | 0.19 | 0.12 | 0.81 |
| Creatinine mg/dl, median (IQR) | 7.2 (5.4–9.2) | 7.2 (5.8–9.4) | 1.4 (1.1–1.7) | 1.4 (1.1–1.7) | 1.3 (1.1–1.7) | 1.4 (1.1–1.8) | 0.61 | 0.48 | 0.71 |
| GFR ml/min per 1.72m2, median (IQR)a | 7.6 (5.5–9.3) | 7.3 (5.2–9) | 58.0 (42.4–70.4) | 54.5 (42.4–67.6) | 56.1 (39.3–71.4) | 53.8 (39.4–66.1) | 0.55 | 0.44 | 0.50 |
| Hemoglobin, g/dl, mean (SD) | 11.7 (1.7) | 12.0 (1.7) | 13.3 (2.2) | 13.7 (1.7) | 13.7 (1.9) | 13.7 (1.5) | 0.15 | 0.16 | 1.0 |
| Potassium, mmol/L, mean (SD) | 4.8 (0.8) | 4.8 (0.8) | 4.2 (0.5) | 4.2 (0.5) | 4.2 (0.5) | 4.3 (0.5) | 0.54 | 0.47 | 0.52 |
| Sodium, mmol/L, mean (SD) | 138.7 (4.6) | 138.8 (3.4) | 140.7 (2.9) | 140.5 (2.8) | 140.8 (2.7) | 140.8 (3.0) | 0.88 | 0.65 | 0.85 |
| Uric acid, mg/dl, mean (SD) | 5.2 (1.9) | 5.1 (2.0) | 6.9 (1.9) | 6.9 (1.7) | 6.9 (1.9) | 6.9 (1.6) | 0.86 | 0.97 | 0.89 |
| Triglycerides, mg/dl, median (IQR) | 141 (94–207.5) | 124.5 (89.3–197.5) | 141 (108.0–204.0) | 129.5 (96.0–159.8) | 151.5 (102.5–191.0) | 135.5 (91.8–204.3) | 0.66 | 0.12 | 0.53 |
| HDL, mg/dl, mean (SD) | 45.7 (16.5) | 52.0 (17.1) | 54.8 (18.7) | 54.8 (14.0) | 53.4 (17.6) | 54.7 (15.7) | 0.02 | 0.97 | 0.58 |
| LDL, mg/dl, mean (SD) | 95.4 (40.1) | 100.9 (34.4) | 121.0 (34.4) | 122.5 (42.1) | 117.5 (34.3) | 123.1 (57.6) | 0.35 | 0.77 | 0.42 |
| C-reactive protein, mg/dl, median (IQR) | 0.3 (0.1–1.1) | 0.2 (0.1–0.8) | 0.2 (0.1–0.7) | 0.2 (0.1–0.4) | 0.3 (0.1–0.6) | 0.2 (0.1–0.5) | 0.7 | 0.30 | 0.45 |
| Protein creatinine ratio, mg/g, median (IQR) | / | / | 110.0 (73.7–231.5) | 132.0 (72.3–273.3) | 113 (74.2–218.0) | 96 (69.6–175.4) | / | 0.61 | 0.20 |
| Urine albumin, mg/L, median (IQR) | / | / | 15.0 (5.0–47.9) | 14.2 (5.0–40.0) | 14.2 (4.3–62.6) | 11.6 (5.0–31) | / | 0.39 | 0.33 |
IQR, interquartile range; BMI, body mass index.
CKD epidemiology collaboration.
Blood Glucose Profiles and HbA1c Development (Additional Secondary End Points)
Fasting plasma glucose levels did not differ between the treatment and the control groups during the first 7 days, either in the intention-to-treat or the per-protocol population (Supplemental Figure 3, A and D). Although per this study’s design, more treatment than control group participants received insulin, the per-patient dose was similar between the control and treatment populations, except for day 3 (Supplemental Figure 3, B and E). Median HbA1c levels in the control and the treatment group increased by 0.2% and 0.3% from baseline to month 3, respectively, but did not differ significantly between groups at any other time point (Supplemental Figure 3, C and F; Supplemental Tables 9 and 10). Mean capillary blood glucose profiles in the basal insulin treatment group at the four daily time points are shown in Figure 4. On average, afternoon capillary blood glucose levels were higher than evening, followed by lunchtime, and fasting glucose levels were the lowest. These differences were especially prominent during the first month post-transplant. The difference between fasting glucose levels and the other three time points prevailed beyond the first month post-transplant.
Figure 4.
Postoperative capillary blood glucose. Mean±SD capillary blood glucose levels >730 days of follow-up in treatment group participants (intention-to-treat population). The color coding of the four different time points (fasting [A], lunch [B], afternoon [C], evening [D]) represents the mean capillary blood glucose course of the respective time point, described at the y-axis. The grey lines are for comparison of the other time points.
Association between Trial Adherence and PTDM
Treatment group participants who were not promptly insulinized despite hyperglycemia and/or who were insufficiently monitored for afternoon capillary blood glucose (the determining measurement for insulin initiation in this study) during the first postoperative week, had higher odds for PTDM at month 6 (OR, 1.56; 95% confidence interval, 1.03 to 2.38), at month 12 (OR, 6.47; 95% confidence interval, 2.32 to 18.05), and at month 24 (OR, 2.51; 95% confidence interval, 1.42 to 4.44). In simple terms, for each day of nonadherence to the protocol the odds for PTDM increased by the factor of the corresponding ORs in these models (Figure 5, Supplemental Figure 4).
Figure 5.
Association between exploratory measures of early protocol adherence and diabetic outcome in the treatment group. (A) Diabetes at months 6, 12, and 24 by the sum of days during the first week after transplantation with missed insulin initiation (when basal insulin therapy initiation was mandated per intervention protocol, but not performed) and with missing afternoon glucose measurements in a combined “Treatment Adherence Score,” in the intention-to-treat population. (B) ORs (95% confidence intervals) for PTDM at months 6, 12, and 24 in the intention-to-treat population. ORs were calculated with days of the “Treatment Adherence Score” as the independent variable. In these models, the odds for diabetes rise by the factor of the corresponding OR, for each additional day on the adherence score. PTDM definitions are analogous to Figure 2, as explained in the Methods section. For further methodological details, see Supplemental Figure 4.
Discussion
In this trial, we investigated the hypothesis that early postoperative treatment of hyperglycemia using basal insulin has an effect on development and duration of PTDM beyond month 3 after kidney transplantation. In the intention-to-treat population, we could not detect a statistically significant difference in the primary end point, PTDM at month 12, defined by antidiabetic treatment or OGTT-derived 2 hour glucose ≥200 mg/dl. In the per-protocol population, however, the unadjusted odds for PTDM in the treatment group were 60% lower, when compared with the standard-of-care control group (95% confidence interval, 0.16 to 1.01), and the respective OR became statistically significant (0.21; 95% confidence interval, 0.07 to 0.62) when adjusted for significant baseline imbalances including polycystic kidney disease, a strong risk factor for PTDM.36
To our best understanding of the study data, eight out of 133 (6%) patients with severe protocol deviations obscured the benefit of the early basal insulin intervention in the intention-to-treat population: specifically, four participants in the treatment group were labeled as “diabetics on treatment” at month 12, although they had either a normal OGTT or an HbA1c value <6.0% (and after month 12 were taken off insulin, because they did not require this therapy, as documented in detail in Supplemental Table 1). Moreover, four participants (two of them patients with diabetes at month 12) did not receive basal insulin at all, despite clear indication by their afternoon glucose values. The power of this trial may therefore not have been sufficient to account for these protocol deviations, considering the overall lower than expected PTDM rate. The differences in outcome between the intention-to-treat and the per-protocol analysis may trigger the judgement that this study is not definitively conclusive. At second view, however, these differences are essentially expectable, from the documented protocol deviations.
Analyzing the association between postoperative glycemic monitoring and delay in therapy initiation with PTDM in logistic regression analyses, we observed that consistent (versus sporadic) antihyperglycemic care may be important to prevent PTDM. The delay in early insulin therapy initiation can be seen as evidence for clinical inertia37 in general. Some reverse clinical inertia38 was also observable in our data, in the form of delayed insulin weaning of four treatment-group participants. We suspect these forms of nonadherence are highly prevalent in the PTDM setting, as they are in type 2 diabetes.38
Although PTDM prevalence in this trial was much lower than in our previous proof-of-concept study,23 the resulting ORs for diabetes in the treatment versus control groups were strikingly similar between both trials. The previously higher PTDM incidence could be due to the higher glucocorticosteroid regimen, which resulted in far larger HbA1c increases in both of the earlier study groups.23
As an important study limitation, besides the high drop-out rate, we acknowledge having been forced to adapt the prespecified statistical analysis of the trial protocol, because it had assumed an ever-increasing number of patients with diabetes, whereas the “natural course” of PTDM often presents with patients who switch from overt diabetes to stable phases of normoglycemia and back.39 This natural course was also confirmed in the present study. Thus, logistic regression was used instead of a Cox proportional hazard model.
Among further limitations, the sample size that we originally calculated to detect a 15% difference in PTDM incidence (n=276), assuming a baseline PTDM incidence of 30%, was only closely met in this trial. However, this sample size would have been underpowered to detect anything near to a 15% difference in PTDM, when using ORs and assuming a PTDM rate as low as was actually observed in this trial.
The actual PTDM definition in this trial (on the basis of OGTT-derived 2 hour plasma glucose) differed from the PTDM definition of the study protocol (2003 ADA diagnostic criteria27 combined with HbA1c, according to the 2009 ADA report26). Various combinations of applying the ADA criteria (including HbA1c) in hierarchical and nonhierarchical order on top of the PTDM definition “by antidiabetic treatment” are, however, shown in Supplemental Figure 1 and demonstrate more favorable ORs for PTDM prevalence in treatment versus control group participants. Applying the OGTT results first therefore does not constitute a real limitation, especially in view of the fact that relying on HbA1c first would have been problematic in the setting of kidney disease and transplantation.40
Finally, the PTDM reduction shown in this trial was accompanied by a significantly increased number of hypoglycemic episodes in the basal insulin treatment group. Although the large majority of these episodes were not concerning, one participant experienced unconsciousness, on day 25 post-transplant. The clinical details indicate the insulin dosing was too high, possibly because the prespecified insulin titration scheme was not followed accurately. In a previous analysis, severe hypoglycemia rates (requiring the assistance of another person) ranged from 0.00 to 0.07 episodes per patient-year in individuals with type 2 diabetes and on insulin therapy, who participated in randomized controlled trials.41
In view of these limitations, the question arises whether the early basal insulin approach is viable outside any research protocol, and what resources and patient willingness are needed to adhere to such a regimen in “real life.” The high drop-out rate and the observed protocol violations by the investigators (which mostly concerned patient labeling at month 12) may reflect the complicated end point determination, rather than suggesting that measuring glucose and administering basal insulin is difficult. Our post-hoc analysis showed the number needed to treat to prevent one case of overt PTDM was n=5 for individuals at higher risk of developing hyperglycemia.4,32–34 Among all 130 treatment group participants, the absolute number of those in need of antidiabetic therapy at 1 year was seven (versus 11 controls) and at 2 years was three (versus 11 controls). The overall number of participants saved from antidiabetic treatment was therefore 4–8. At the expense of an increased hypoglycemia risk, the proposed protocol thus appears clinically meaningful, but discussion will go on whether patients should perhaps be selected for it, on the basis of their preposition to develop more severe hyperglycemia,42,43 potential adherence to the insulin administration, and especially their understanding of the hypoglycemia hazard. Last but not least, as an informal post-hoc analysis of the study outcomes by sex hinted at the existence of a potential sex-related treatment bias, we encourage additional awareness and research in this area.44
In summary, the principal outcome of this trial (primary end point) was negative in the intention-to-treat analysis. Among the study limitations, which were fully acknowledged in the data presentation, were 6% protocol violations. Conducting analyses on the per-protocol population allowed accounting for these violations and provided strong additional indications (on top of those from the intention-to-treat population) for an improvement in glucose control long-term, at least up to 2 years, when the initial basal insulin treatment is maintained as scheduled by the protocol. The results of this trial therefore confirm the findings from our earlier proof-of-concept study,23 albeit at a much lower PTDM incidence rate. Ultimately, our findings encourage further studies aimed at tailoring basal insulin therapy to patients who will be expected to maximize their benefit from the proposed prevention strategy.
Disclosures
A. Rosenkranz reports receiving speaker honoraria and fees for advisory boards from Astellas Pharma; reports receiving research funding from Baxter and Fresenius Medical Care; reports receiving honoraria from AstraZeneca, Baye, Baxter, Behring, Bristol Myers Squibb, Böehringer, Chiesi, GlaxoSmithKline Mitsubishi, Merck Sharp & Dohme, Novartis, Otsuka, Pfizer, Rigel, Sandoz, and Sanofi. A. Tura reports having consultancy agreements with the Medical University of Vienna, Austria; reports being a scientific advisor or member of Elsevier, Hindawi Publishing Corporation, the International Journal of Endocrinology, Journal of Diabetes Research, and Mathematics and Computers in Simulation. F. Halleck reports receiving honoraria from Novartis. F. Port reports being a scientific advisor or member of Easy Water for Everyone (Non-Governmental Organization). K. Eller reports receiving research funding from Chiesi; and reports receiving honoraria from Alexion, Amicus, AstraZeneca, Chiesi, and Sanofi Aventis. M. Crespo reports receiving speaker honoraria from Astellas, Chiesi, Hansa, and Novartis; reports being a scientific advisor or member as Coordinator of the Transplant Group of the Spanish Society of Nephrology, Member of the Board of the Descartes Group of European Dialysis and Transplant Association. J. Pascual reports receiving honoraria from Chiesi (sporadic as speaker) and Novartis (sporadic as speaker). J. Werzowa reports consultancy agreements with and being a scientific advisor or member of AstraZeneca and Vifor; reports receiving honoraria from Astellas, Amgen, AstraZeneca, Biotest, and Vifor; and reports other interests/relationships in Healthtunes.org. M. Hecking reports having received speaker honoraria from Astellas Pharma, which did not influence the submitted work; reports receiving research funding from Astellas Pharma, Boehringer Ingelheim, Eli Lilly, and Siemens Healthcare; and reports receiving honoraria from Fresenius Medical Care. K. Budde reports receiving research grants, travel support, and honoraria from Abbvie, Alexion, Astellas, Bristol Myers Squibb, Chiesi, Commonwealth Serum Laboratories Behring, Fresenius, Hansa, Hexal, Hookipa Biotech, Merck Sharp & Dohme, Novartis, Otsuka, Pfizer, Quark, Roche, Shire, Siemens, Sandoz, Veloxis Vifor, and Vitaeris; and reports being a scientific advisor or member of Astellas, Bristol-Myers Squibb, Chiesi, Hansa, Hexal, MSD, Novartis, Pfizer, Roche, and Veloxis. P. Song reports being a scientific advisor or member of member of the Editorial Boards of the Canadian Journal of Statistics, Journal of the American Statistical Association and the Journal of Multivariate Analysis. All remaining authors have nothing to disclose.
Funding
This investigator-initiated trial was supported by University of Michigan subcontract 3002300292 to the National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases grant R01DK092475 and by Astellas Pharma and Eli Lilly in the form of contracts with the Medical University of Vienna that did not contain intellectual restrictions on publication.
Supplementary Material
Acknowledgments
K. Budde, K. Eller, A. Frey, M. Hecking, J. Pascual, F. Port, A. Rosenkranz, and P. Song designed the trial, and wrote and submitted the study protocol. M. Bergmann, K. Budde, M. Crespo, K. Eller, A. Frey, H. Hafner-Giessauf, M. Hecking, J. Pascual, M. Perez Saez, A. Rosenkranz, and J. Werzowa enrolled participants into the trial and actively treated trial participants at the four trial sites. L. Bergfeld, M. Bergmann, A. Frey, M. Hecking, S. Kron, A. Kurnikowski, E. Schwaiger, and D. Topitz retrieved the data, actively converting paper case report forms into electronic database. L. Bergfeld, M. Bergmann, A. Frey, M. Hecking, S. Kron, A. Kurnikowski, E. Schwaiger, D. Topitz, and A. Tura analyzed the data. M. Hecking, S. Kron, A. Kurnikowski, and E. Schwaiger wrote the manuscript. F. Bachmann, K. Budde, K. Eller, F. Halleck, S. Kron, J. Pascual, F. Port, R. Ristl, A. Rosenkranz, P. Song, A. Tura, and J. Werzowa corrected the manuscript. All authors approved the final version of the manuscript before submission. We thank Carlos Rodriguez-Torres for monitoring this trial and Bianca Reiskopf for excellent assistance with data management, distribution of trial materials, and blood sample analysis. We also thank Elisabeth Ponweiser and Candace Joefield-Roka for laboratory assistance. We thank the trial coordinators, Andreas Rosenstingl (Vienna), Rene Nadolny (Berlin), Claudia Pflanzl-Knizacek (Graz), and Nuria Cubino Junyent (Barcelona). We thank Prof. Florian Frommlet, Dr. Roman Reindl-Schwaighofer, and Dr. Zeljko Kikic for helpful discussions and statistical advice, and Dr. Walter Hörl (deceased), Dr. Wilfred Druml, and Dr. Rainer Oberbauer for supporting this project. We also acknowledge the important contribution of Dr. Akinlolu Ojo and Dr. Fu Luan, former employees of the University of Michigan, who initially designed the trial, together with M. Hecking and F. Port. We thank Hermann Hayn, Dr. Johannes Franck, and Dr. Markus Grimm for excellent legal advice and support, which enabled bringing this project to completion. As regards completing this project, we also thank Dr. Oswald Wagner and Dr. Josef Smolen for organizing and allowing specimen analysis at the Medical University of Vienna, rather than University of Michigan (as originally planned). Importantly, we thank Dr. Marcus Säemann for igniting the entire study idea and overseeing this trial through 2016.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
See related editorial, “Prevention of Post-Transplantation Diabetes: Small Steps, Big Opportunities,” on pages 1833–1834.
Supplemental Material
This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/10.1681/ASN.2021010127/-/DCSupplemental.
Supplemental Figure 1. Detailed oral glucose tolerance tests outcomes at months 6, 12, and 24 after transplantation.
Supplemental Figure 2. Length of hospitalization among patients who had at least one hypoglycemic event within the initial postoperative hospitalization period (A) or during the entire study period (B), compared with patients without hypoglycemic events.
Supplemental Figure 3. Postoperative fasting plasma glucose, insulinization, and HbA1c.
Supplemental Figure 4. Association between exploratory measures of early protocol adherence and diabetic outcome in the treatment group.
Supplemental Table 1. Details of ITP-NODAT participants who were on antidiabetic therapy at months 6, 12, and 24.
Supplemental Table 2. Immunosuppressive medication: Intention-to-treat analysis.
Supplemental Table 3. Immunosuppressive medication: Per-protocol analysis.
Supplemental Table 4. Characteristics of the high-risk population.
Supplemental Table 5. Diabetic outcome and ORs, by sex (men [males] left, women [females] right—note that we did not distinguish between gender and sex [not self-reported]).
Supplemental Table 6. Hypoglycemic events and other safety end points for the per-protocol population.
Supplemental Table 7. Length of hospitalization and hypoglycemic events.
Supplemental Table 8. Weight, renal function and non-glycemia labs: Per-protocol analysis.
Supplemental Table 9. Postoperative glucose profiles, insulin doses, and HbA1c in the intention-to-treat population.
Supplemental Table 10. Postoperative glucose profiles, insulin doses, and HbA1c in the per-protocol population.
Supplemental Methods Table 1 (from the original study protocol). NPH insulin titration regimen for patients in group A.
Supplemental Methods Table 2 (from the original study protocol). Sliding scale for short acting insulin regimen.
Supplemental Methods Table 3 (from the original study protocol). Tacrolimus and glucocorticosteroid dosing schedule.
Supplemental Methods Table 4 (from the original study protocol). Follow-up schedule for study visits and study related-procedures.
References
- 1.Eide IA, Halden TA, Hartmann A, Åsberg A, Dahle DO, Reisaeter AV, et al.: Mortality risk in post-transplantation diabetes mellitus based on glucose and HbA1c diagnostic criteria. Transpl Int 29: 568–578, 2016 [DOI] [PubMed] [Google Scholar]
- 2.Seoane-Pillado MT, Pita-Fernández S, Valdés-Cañedo F, Seijo-Bestilleiro R, Pértega-Díaz S, Fernández-Rivera C, et al.: Incidence of cardiovascular events and associated risk factors in kidney transplant patients: A competing risks survival analysis. BMC Cardiovasc Disord 17: 72, 2017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kasiske BL, Snyder JJ, Gilbertson D, Matas AJ: Diabetes mellitus after kidney transplantation in the United States. Am J Transplant 3: 178–185, 2003 [DOI] [PubMed] [Google Scholar]
- 4.Cosio FG, Pesavento TE, Kim S, Osei K, Henry M, Ferguson RM: Patient survival after renal transplantation: IV. Impact of post-transplant diabetes. Kidney Int 62: 1440–1446, 2002 [DOI] [PubMed] [Google Scholar]
- 5.Revanur VK, Jardine AG, Kingsmore DB, Jaques BC, Hamilton DH, Jindal RM: Influence of diabetes mellitus on patient and graft survival in recipients of kidney transplantation. Clin Transplant 15: 89–94, 2001 [DOI] [PubMed] [Google Scholar]
- 6.Hjelmesaeth J, Hartmann A, Leivestad T, Holdaas H, Sagedal S, Olstad M, et al.: The impact of early-diagnosed new-onset post-transplantation diabetes mellitus on survival and major cardiac events. Kidney Int 69: 588–595, 2006 [DOI] [PubMed] [Google Scholar]
- 7.Cosio FG, Kudva Y, van der Velde M, Larson TS, Textor SC, Griffin MD, et al.: New onset hyperglycemia and diabetes are associated with increased cardiovascular risk after kidney transplantation. Kidney Int 67: 2415–2421, 2005 [DOI] [PubMed] [Google Scholar]
- 8.Valderhaug TG, Hjelmesæth J, Hartmann A, Røislien J, Bergrem HA, Leivestad T, et al.: The association of early post-transplant glucose levels with long-term mortality. Diabetologia 54: 1341–1349, 2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Moon JI, Barbeito R, Faradji RN, Gaynor JJ, Tzakis AG: Negative impact of new-onset diabetes mellitus on patient and graft survival after liver transplantation: Long-term follow up. Transplantation 82: 1625–1628, 2006 [DOI] [PubMed] [Google Scholar]
- 10.Liu FC, Lin JR, Chen HP, Tsai YF, Yu HP: Prevalence, predictive factors, and survival outcome of new-onset diabetes after liver transplantation: A population-based cohort study. Medicine (Baltimore) 95: e3829, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ling Q, Xu X, Xie H, Wang K, Xiang P, Zhuang R, et al.: New-onset diabetes after liver transplantation: A national report from China Liver Transplant Registry. Liver Int 36: 705–712, 2016 [DOI] [PubMed] [Google Scholar]
- 12.Roccaro GA, Goldberg DS, Hwang WT, Judy R, Thomasson A, Kimmel SE, et al.: Sustained posttransplantation diabetes is associated with long-term major cardiovascular events following liver transplantation. Am J Transplant 18: 207–215, 2018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hackman KL, Bailey MJ, Snell GI, Bach LA: Diabetes is a major risk factor for mortality after lung transplantation. Am J Transplant 14: 438–445, 2014 [DOI] [PubMed] [Google Scholar]
- 14.Hackman KL, Snell GI, Bach LA: Poor glycemic control is associated with decreased survival in lung transplant recipients. Transplantation 101: 2200–2206, 2017 [DOI] [PubMed] [Google Scholar]
- 15.Kim HJ, Jung SH, Kim JJ, Yun TJ, Kim JB, Choo SJ, et al.: New-onset diabetes mellitus after heart transplantation: Incidence, risk factors and impact on clinical outcome. Circ J 81: 806–814, 2017 [DOI] [PubMed] [Google Scholar]
- 16.American Diabetes Association: 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2018. Diabetes Care 41: S13–S27, 2018 [DOI] [PubMed] [Google Scholar]
- 17.Hecking M, Kainz A, Werzowa J, Haidinger M, Döller D, Tura A, et al.: Glucose metabolism after renal transplantation. Diabetes Care 36: 2763–2771, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Muller MM, Schwaiger E, Kurnikowski A, Haidinger M, Ristl R, Tura A, et al.: Glucose metabolism after kidney transplantation: Insulin release and sensitivity with tacrolimus- versus belatacept-based immunosuppression. Am J Kidney Dis 77: 462–4642020 [DOI] [PubMed] [Google Scholar]
- 19.Jenssen T, Hartmann A: Post-transplant diabetes mellitus in patients with solid organ transplants. Nat Rev Endocrinol 15: 172–188, 2019 [DOI] [PubMed] [Google Scholar]
- 20.Hecking M, Sharif A, Eller K, Jenssen T: Management of post-transplant diabetes: Immunosuppression, early prevention, and novel antidiabetics. Transpl Int 34: 27–48, 2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hecking M, Werzowa J, Haidinger M, Hörl WH, Pascual J, Budde K, et al.; European-New-Onset Diabetes After Transplantation Working Group: Novel views on new-onset diabetes after transplantation: Development, prevention and treatment. Nephrol Dial Transplant 28: 550–566, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Chakkera HA, Weil EJ, Castro J, Heilman RL, Reddy KS, Mazur MJ, et al.: Hyperglycemia during the immediate period after kidney transplantation. Clin J Am Soc Nephrol 4: 853–859, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hecking M, Haidinger M, Döller D, Werzowa J, Tura A, Zhang J, et al. Early basal insulin therapy decreases new-onset diabetes after renal transplantation. J Am Soc Nephrol 23: 739–749, 2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Yates CJ, Fourlanos S, Colman PG, Cohney SJ: Divided dosing reduces prednisolone-induced hyperglycaemia and glycaemic variability: A randomized trial after kidney transplantation. Nephrol Dial Transplant 29: 698–705, 2014 [DOI] [PubMed] [Google Scholar]
- 25.Davidson J, Wilkinson A, Dantal J, Dotta F, Haller H, Hernández D, et al.; International Expert Panel: New-onset diabetes after transplantation: 2003 International consensus guidelines. Proceedings of an international expert panel meeting. Barcelona, Spain, 19 February 2003. Transplantation 75[Suppl]: SS3–SS24, 2003 [DOI] [PubMed] [Google Scholar]
- 26.International Expert Committee: International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care 32: 1327–1334, 2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Expert Committee on the Diagnosis and Classification of Diabetes Mellitus: Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes Care 26[Suppl 1]: S5–S20, 2003 [DOI] [PubMed] [Google Scholar]
- 28.Sharif A, Hecking M, de Vries AP, Porrini E, Hornum M, Rasoul-Rockenschaub S, et al.: Proceedings from an international consensus meeting on posttransplantation diabetes mellitus: recommendations and future directions. Am J Transplant 14: 1992–2000, 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Finfer S, Chittock DR, Su SY, Blair D, Foster D, Dhingra V, Bellomo R, Cook D, Dodek P, Henderson WR, Hebert PC, Heritier S, Heyland DK, McArthur C, McDonald E, Mitchell I, Myburgh JA, Norton R, Potter J, Robinson BG, Ronco JJ; Investigators N-SS: Intensive versus conventional glucose control in critically ill patients. N Engl J Med 360: 1283–1297, 2009 [DOI] [PubMed] [Google Scholar]
- 30.Heisel O, Heisel R, Balshaw R, Keown P: New onset diabetes mellitus in patients receiving calcineurin inhibitors: A systematic review and meta-analysis. Am J Transplant 4: 583–595, 2004 [DOI] [PubMed] [Google Scholar]
- 31.Luan FL, Stuckey LJ, Ojo AO: Abnormal glucose metabolism and metabolic syndrome in non-diabetic kidney transplant recipients early after transplantation. Transplantation 89: 1034–1039, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Torres A, Hernández D, Moreso F, Serón D, Burgos MD, Pallardó LM, et al.: Randomized controlled trial assessing the impact of tacrolimus versus cyclosporine on the incidence of posttransplant diabetes mellitus. Kidney Int Rep 3: 1304–1315, 2018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Romagnoli J, Citterio F, Violi P, Cadeddu F, Nanni G, Castagneto M: Post-transplant diabetes mellitus: A case-control analysis of the risk factors. Transpl Int 18: 309–312, 2005 [DOI] [PubMed] [Google Scholar]
- 34.Shah T, Kasravi A, Huang E, Hayashi R, Young B, Cho YW, et al.: Risk factors for development of new-onset diabetes mellitus after kidney transplantation. Transplantation 82: 1673–1676, 2006 [DOI] [PubMed] [Google Scholar]
- 35.Taking sex into account in medicine. Lancet 378: 1826, 2011 [DOI] [PubMed] [Google Scholar]
- 36.de Mattos AM, Olyaei AJ, Prather JC, Golconda MS, Barry JM, Norman DJ: Autosomal-dominant polycystic kidney disease as a risk factor for diabetes mellitus following renal transplantation. Kidney Int 67: 714–720, 2005 [DOI] [PubMed] [Google Scholar]
- 37.Phillips LS, Branch WT, Cook CB, Doyle JP, El-Kebbi IM, Gallina DL, et al.: Clinical inertia. Ann Intern Med 135: 825–834, 2001 [DOI] [PubMed] [Google Scholar]
- 38.Giugliano D, Maiorino MI, Bellastella G, Esposito K: Clinical inertia, reverse clinical inertia, and medication non-adherence in type 2 diabetes. J Endocrinol Invest 42: 495–503, 2019 [DOI] [PubMed] [Google Scholar]
- 39.Porrini EL, Díaz JM, Moreso F, Delgado Mallén PI, Silva Torres I, Ibernon M, et al.: Clinical evolution of post-transplant diabetes mellitus. Nephrol Dial Transplant 31: 495–505, 2016 [DOI] [PubMed] [Google Scholar]
- 40.Sharif A, Baboolal K: Diagnostic application of the A(1c) assay in renal disease. J Am Soc Nephrol 21: 383–385, 2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Elliott L, Fidler C, Ditchfield A, Stissing T: Hypoglycemia event rates: A comparison between real-world data and randomized controlled trial populations in insulin-treated diabetes. Diabetes Ther 7: 45–60, 2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Chakkera HA, Weil EJ, Pham PT, Pomeroy J, Knowler WC: Can new-onset diabetes after kidney transplant be prevented? Diabetes Care 36: 1406–1412, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Hecking M, Sharif A, Port FK, Säemann MD: Comment on: Chakkera et al. Can new-onset diabetes after kidney transplant be prevented? Diabetes Care 36: e182, 2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Mauvais-Jarvis F, Bairey Merz N, Barnes PJ, Brinton RD, Carrero JJ, DeMeo DL, et al.: Sex and gender: Modifiers of health, disease, and medicine. Lancet 396: 565–582, 2020 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.