Routine laboratory measurements were extracted from the laboratory information system of the UMCG. Plasma PTH, calcium, phosphate, total alkaline phosphatase (ALP), creatinine, and albumin concentrations were measured at outpatient visits. Plasma calcium was corrected for albumin according to the following formula: corrected calcium (mg/dL) = measured calcium (md/dL)+0.025*(40–[albumin (g/dL)]). All routine measurements before March 2006 were performed on the Merck Mega Analyzer (Merck); measurements after March 2006 were performed on the Roche Modular (Roche Ltd.). Laboratory measurements prior to March 2006 were converted according to the equations [17] listed in Supplementary Table S1. The last PTH measurement prior to the kidney transplant procedure was used for analyses. Reference values for plasma-corrected calcium were 8.8–10.4 mg/dL (2.20–2.60 mmol/L) and for plasma phosphate 2.17–4.64 mg/dL (0.70–1.50 mmol/L) [10]. At each individual measurement, patients were classified as having hypo-, normo-, or hypercalcemia and hypo-, normo-, or hyperphosphatemia according to these definitions. Creatinine-based eGFR was calculated according to the CKD Epidemiology Collaboration Equation (EPI) equation [18, 19]. Primary cytomegalovirus (CMV) infection was defined as CMV viremia demonstrated by PCR in the absence of CMV-specific IgG antibodies. All other measurements were performed using standard laboratory techniques.

All patients who received a kidney transplant underwent a standardized follow-up regime. Patients received a standardized immunosuppression protocol, comprising triple therapy with tacrolimus or cyclosporine, in combination with mycophenolate mofetil and corticosteroids, as previously reported [20]. Shortly after transplantation, patients visit the outpatient department weekly. The frequency of visits is tapered to every 4–6 weeks during the first year after transplantation, and at least four times a year after the first year. End of follow-up was December 2020. Donor and recipient characteristics were collected as part of the TransplantLines registry [21]. The primary cause of kidney failure was categorized according to the European Renal Association Registry Coding System [22]. Acute rejection was defined according to the Banff criteria. There was no loss to follow-up.

The three co-primary outcomes were DGF, defined as the need for dialysis within the first 7 days posttransplant, DCGF, defined as return to dialysis or re-transplantation, censored for death, and all-cause mortality. Up-to-date follow-up was warranted through the continuous surveillance system of the outpatient clinic.

Statistical analyses were performed using IBM SPSS version 23.0 (SPSS Inc., Chicago, IL). In all analyses p < 0.05 was considered significant. Variable distribution was evaluated by Kolmogorov Smirnov test. Categorical variables are presented as n (%), normally distributed variables as mean ± standard deviation (SD) and non-normally distributed variables as median with interquartile range (IQR). Skewed variables were log-transformed where appropriate. We handled remaining missing data for key variables using multiple imputation of variables with less than 10% missing data. Data of the following variables were imputed using multiple imputation by chained equations with five imputations: total alkaline phosphatase, calcium, phosphate, cold ischemia time, warm ischemia time, number of human leucocyte antigen (HLA) mismatches, body mass index (BMI), donor status, presence of diabetes, use of cinacalcet, and use of vitamin D. Results from analyses on each imputed data set were then pooled according to Rubin’s rules [23].

To analyze whether pre-transplant plasma PTH, ALP, calcium and phosphate were independently associated with DGF, DCGF and all-cause mortality, we performed logistic and Cox-proportional regression analyses, respectively. Plasma levels were analyzed as categorical variables and as (log-transformed) continuous variables. The proportional hazard assumption was tested using statistical tests and graphical diagnostics based on the scaled Schoenfeld residuals. For PTH, patients were clustered in three groups according to the KDIGO guidelines-recommended thresholds: ≤2 times (≤150 pg/mL), >2 and <9 times (>150 and <771 pg/mL), or ≥9 times (≥771 pg/mL) of the upper limit of normal for the assay [24], using the middle range as reference. For ALP, calcium and phosphate, patients were analyzed in quartiles. Since most patients had normal or elevated levels of ALP at time of transplantation, we defined the lowest quartile as reference category for this parameter. For calcium and phosphate, we used the second quartile as reference category, since both very high and very low levels may occur at transplantation and both might be associated with adverse outcomes. We performed multivariate Cox regression analyses, cumulatively adjusted for age and sex (Model 1), and further variables previously associated with outcomes after kidney transplantation and bone mineral metabolism such as primary cause of kidney failure, primary CMV infection, acute allograft rejection, dialysis vintage, preemptive transplant, number of HLA mismatches, donor age and sex, living donor status, cold and warm ischemia time, history of diabetes, body mass index, serum calcium, phosphate and albumin at transplantation time, cinacalcet, vitamin D use, history of parathyroidectomy and decade of transplantation (Model 2; Supplementary Figure S2). The associations of pre-transplant PTH levels with DGF, DCGF and mortality were further investigated using restricted cubic splines using fully adjusted models. We also evaluated whether pre-transplant plasma PTH and ALP contributed to prediction of mortality risk using ROC-curve analysis with determination of the area under the ROC-curve (AUC). Finally, we studied the relationship between PTH at 1 year post-transplantation and DCGF or mortality, both in continuous analysis and using quartiles (fully adjusted models as described above). In these analyses, patients who developed graft loss or died within the first year post-transplant were excluded.

Potential effect modification for the association between pre-transplant PTH and outcomes was explored using multiplicative interaction terms followed by prespecified subgroup analyses according to age, sex, use of vitamin D, use of cinacalcet, preemptive kidney transplant, dialysis vintage, donor status (living or deceased), serum calcium, phosphate and ALP. The p-values of interaction terms were considered significant when <0.05.

Finally, we performed sensitivity analyses for the DGF analysis, restricted to patients who received from a postmortal donor (as DGF is much more common after postmortal donor kidney transplantation), and for the mortality analyses after exclusion of individuals who died within the first year post-transplant, and restricted to patients transplanted after 2010.

A total of 1,576 kidney transplant recipients (KTRs) (age 51.6 ± 14.0 years, 57.3% male) were included in the primary analyses. Baseline patient and transplant characteristics are presented in Table 1. In brief, donor age was 51.2 ± 13.5 years, 785 (49.8%) patients received a graft from a living donor, median (IQR) dialysis vintage was 18.0 (0–40.0) months, and 535 (33.9%) patients underwent a pre-emptive transplantation. Median (IQR) pre-transplant plasma PTH concentration was 231 (122–425) pg/mL, with 121 (7.7%) of patients presenting with plasma PTH ≥771 pg/mL (median 942 [838–1,236]pg/mL). Calcimimetics were used at time of transplant by 198 (12.6%) patients, and vitamin D analogs by 58% of patients. There were few missing data points (Supplementary Table S2).

As shown in Table 2, 278 (17.6%) patients developed DGF. Upon fully adjusted logistic regression analysis with pre-transplant PTH as a continuous variable, no significant association was found with DGF. When analyzing patients in three groups (≤2 times, >2 and <9 times, ≥9 times the upper limit of normal for the assay of plasma PTH, corresponding with ≤150 pg/mL, >150 and <771 pg/mL, and ≥771 pg/mL, respectively), patients with pre-transplant PTH >771 pg/mL had a risk of DGF that was comparable to the reference group (Table 2). Interaction analysis revealed significant effect modification by sex (P-interaction = 0.03), as shown in Figure 1. The incidence of DGF was similar among men (18.6%) and women (16.3%, p = 0.15 vs. men). The fully adjusted association between plasma PTH and DGF was significant among women (HR 1.37 [95% CI 1.05–1.78], p = 0.02), but not among men. In a sensitivity analysis restricted to women who received a graft from a postmortal donor, the association between low PTH levels and DGF did not persist (HR 1.14 [95% CI 0.89–1.46], p = 0.30).

During median follow-up of 5.0 (range 0.2–29.5) years, 150 (9.9%) patients developed DCGF. In fully adjusted Cox regression analyses, pre-transplant plasma PTH levels were not associated with DCGF (Figure 2A). As shown in Table 2, patients with pre-transplant PTH ≤150 pg/mL or PTH >771 pg/mL had a risk that was comparable to the reference group (HR 0.98 [95% CI 0.87–1.67], p = 0.85; HR 1.01 [95% CI 0.45–2.25], p = 0.59, respectively). No significant effect modification was observed (Figure 3). Sensitivity analyses after exclusion of 39 individuals who developed DCGF within the first year after transplantation (N = 111, fully adjusted HR 0.94 [95% CI 0.83–1.08], p = 0.45) yielded similar results. At 1 year post-transplantation, the median PTH level was 110.5 (72.4–168.3) pg/mL. A higher PTH level at 1 year after transplantation was associated with an increased risk of DCGF (fully adjusted HR 1.31 [95% CI 1.03–1.69], p = 0.03) in continuous analysis (Supplementary Table S3). Patients in the highest quartile of PTH levels at 1 year post-transplant also had an increased risk of DCGF when compared with the lowest quartile (HR 2.64 [95% CI 1.21–5.80], p = 0.02).

During median follow-up of 5.2 (range 0.2–29.5) years, 432 (28.6%) patients died. In fully adjusted Cox regression analyses, pre-transplant plasma PTH levels were not associated with all-cause mortality (Table 2; Figure 2B). There was no significant effect modification in interaction analyses (Supplementary Figure S3). Sensitivity analyses after exclusion of 30 individuals who died within the first year post-transplant (N = 402, fully adjusted HR 1.03 [95% CI 0.95–1.12], p = 0.50), restricted to patients who were transplanted after 2010 (N = 197, HR 1.11 [95% CI 0.97–1.27], p = 0.12), or after exclusion of preemptive transplantation (N = 94, HR 0.99 [95% CI 0.82–1.19], p = 0.91) yielded similar results. An additional analysis in a limited number of patients with available PTH data from 6 to 12 months prior to transplantation (N = 187) showed no significant association with all-cause mortality (HR 0.78 [95% CI 0.55–1.16], p = 0.23). Finally, PTH at 1 year after transplantation showed a trend towards association with all-cause mortality (fully adjusted HR 1.14 [95% CI 0.99–1.32], p = 0.05) upon continuous analysis; patients in the highest PTH quartile (at 1 year post-transplant) had a higher risk of all-cause mortality than those in the lowest quartile (HR 2.64 [95% CI 2.20–5.81], p = 0.02).

Subsequently, we analyzed the associations of pre-transplant serum ALP levels with risk of DGF, DCGF and all-cause mortality. There were no significant associations between serum ALP and risk of DGF, or DCGF (Supplementary Table S4). In age- and sex-adjusted analyses, higher serum ALP was associated with an increased risk of all-cause mortality, but this association lost significance upon multivariable adjustment. Compared with patients in the first quartile of pre-transplant serum ALP (52.0 [46.0–57.0]U/L), patients in fourth quartile with median pre-transplant serum ALP of 126.0 [110.8–158.0]U/L had a higher risk of all-cause mortality in fully adjusted model (HR 1.27 [95% CI 1.00–1.87]). There was no effect modification by pre-transplant dialysis status (preemptive or not) in interaction analysis (HR 1.30 [95% CI 0.72–2.32], p = 0.39). ALP, either alone or combined with PTH, did not change risk prediction for mortality when added to a model with established risk factors (Table 3). Plasma calcium was not significantly associated with the risk of DGF, DCGF, or all-cause mortality (Supplementary Table S5). Compared with patients in the second quartile of pre-transplant serum phosphate (4.2 [3.8–4.7]mg/dL), those in the fourth quartile with median pre-transplant serum phosphate of 6.5 [5.7–12.9]mg/dL had an increased risk of DGF in the fully adjusted model 2.21 [95% CI 1.21–4.03, p < 0.01] (Supplementary Table S6). Interaction analysis revealed significant effect modification by donor status (P-interaction<0.001). The fully adjusted association between plasma phosphate and DGF was significant among recipients of post-mortal draft (HR 1.5 [95% CI 1.03–2.33], p = 0.03), but not among recipients from a living donor. Plasma phosphate at transplantation was not associated with DCGF or mortality.

From 121 patients with severe hyperparathyroidism before transplantation, only 12 (10%) remained with elevated PTH (≥771 pg/mL) at 1 year post-transplant (Figure 4A). At baseline, almost one-third of patients had a calcium value outside the reference (13% hypocalcemia and 12% hypercalcemia). At 1 year after transplantation, 4.2% of patients presented with hypocalcemia, while 14.4% presented with hypercalcemia (Figure 4B). At time of transplantation more than half of patients had hyperphosphatemia (873 [55.6%] patients). At 1 year after transplantation, the prevalence of hyperphosphatemia had decreased to 3%. On the other hand, hypophosphatemia had a prevalence of 2% at baseline and of 10% at 1 year after transplantation (Figure 4C).