Abstract
Background
Chronic kidney disease (CKD)–mineral and bone disorder (MBD) contributes to fractures, cardiovascular events, and mortality in patients receiving maintenance hemodialysis (MHD). The longitudinal patterns of the CKD-MBD biomarkers parathyroid hormone (PTH) and alkaline phosphatase (ALP), as well as their prognostic value, remain uncertain.
Methods
We retrospectively identified patients initiating MHD between 2004 and 2012 using linked medical records and Taiwan National Health Insurance Research Database. Two-year serial ALP (n = 447) and PTH (n = 338) measurements were analyzed using group-based trajectory modeling to characterize longitudinal patterns. A landmark design was applied for outcome analysis, and multivariable Cox models were used to assess associations between trajectory group membership and subsequent clinical outcomes, including fractures, major adverse cardiovascular events, cardiovascular death, and all-cause mortality.
Results
Three distinct trajectories were identified for both ALP and PTH, and remained stable from MHD initiation. ALP trajectories were labeled as low-normal, high-normal, and high based on median levels relative to the reference range, while PTH trajectories were labeled as below-target, within-lower-target, and within-upper-target according to KDIGO-recommended target ranges for patients receiving MHD; these groups demonstrated statistically significant differences in median ALP and PTH levels across trajectories. A high-normal ALP trajectory was independently associated with cardiovascular death (adjusted HR, 8.44; 95% CI, 1.78–39.96; P = 0.0072) and all-cause mortality (adjusted HR, 1.54; 95% CI, 1.15–2.06; P = 0.0034). A below-target PTH trajectory was associated with a higher crude risk of all-cause mortality compared with other trajectory patterns.
Conclusions
ALP and PTH trajectories are identifiable early after HD initiation and reflect distinct risk profiles. These findings support the clinical value of longitudinal monitoring from the early dialysis period and highlight the importance of avoiding PTH oversuppression to improve risk stratification and management. Future large-scale prospective studies are needed to clarify subtle longitudinal changes in ALP and PTH and their associations with outcomes.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12882-026-05008-7.
Keywords: Alkaline phosphatase, Hemodialysis, Outcomes, Parathyroid hormone, Trajectory
Introduction
Chronic kidney disease (CKD)–mineral and bone disorder (MBD) remains a major source of fracture, morbidity, and mortality among patients receiving maintenance hemodialysis (MHD) [1, 2]. In this population, parathyroid hormone (PTH) and alkaline phosphatase (ALP) are commonly used as biochemical markers of bone turnover, along with cardiovascular risk and overall prognosis [3–5].
While a U-shaped association between PTH and mortality risk has been reported [4], its clinical utility is limited by high biological variability [6]. By contrast, ALP levels, which exhibit lower biological variability, may be a more reliable prognostic marker than PTH [3, 7]. In hemodialysis (HD) patients, elevated ALP levels are associated with increased fracture or cardiovascular risk and mortality [3, 8]; however, studies have also reported that elevated ALP levels are associated with improved nutrition and decreased mortality [9]. Thus, existing findings yield conflicting prognostic implications. Further research is required to clarify the role of ALP in predicting adverse outcomes among patients undergoing MHD.
The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines advocate serial assessment and trend interpretation of PTH rather than dependence on isolated values to inform clinical decisions [2]. However, the longitudinal patterns of PTH and ALP, and their clinical utility in patients receiving MHD, remain poorly understood. A few studies have evaluated the prognostic value of serial changes in PTH and ALP during the first 6 months of HD [10, 11]; however, this limited observation period may be insufficient to detect clinically meaningful patterns because bone turnover (reflected by both PTH and ALP) is a slow biological process, occurring over several years.
Trajectory analysis, a statistical approach, can be used to assess the longitudinal evolution of biomarkers [12]. We therefore hypothesized that distinct trajectories of PTH and ALP are associated with differential risks of adverse outcomes in patients receiving MHD. In this study, we characterized distinct trajectories of PTH and ALP in incident MHD patients and determined what affects these trajectories and how these trajectories are related to clinical outcomes.
Materials and methods
Data source
Baseline demographic and biochemical data were retrospectively extracted from electronic medical records and linked to the Ministry of Health and Welfare’s CKD and mortality databases, both derived from the Taiwan National Health Insurance Research Database (NHIRD) [13]. The dataset contained comprehensive information regarding prescriptions, diagnoses, and procedures, enabling an accurate integration of clinical, laboratory, and claims data. This study was performed in accordance with the STROBE guidelines.
Study population and study design
We screened 652 patients initiating MHD at a single HD center between January 1, 2001, and December 31, 2014. Exclusion criteria included subtotal parathyroidectomy ≤ 2 years before MHD initiation or during cohort follow-up (n = 12), presence of a hepatic disorder [n = 50; International Classification of Diseases (ICD) codes listed in Table S1], and death within the 2-year trajectory window (n = 77). Patients initiating MHD during 2001–2003 were excluded due to semiannual laboratory measurements (n = 18), and those initiating MHD in 2013–2014 were excluded to ensure adequate follow-up duration (n = 10). Therefore, the final cohort included patients initiating MHD between 2004 and 2012 (Fig. 1). After excluding 38 patients with missing ALP data, 447 patients were included in the ALP trajectory analysis. Among these, an additional 109 patients with missing PTH data were excluded, leaving 338 patients for the PTH trajectory analysis (Fig. 1).
Fig. 1.
Flow of exclusion process. *Patients were excluded because laboratory data were collected only semiannually (n = 18). †Patients were excluded to allow for adequate follow-up (n = 10). ‡Median (IQR) numbers of missing ALP and PTH measurements per patient over 24 months were 4.0 (4.0) and 2.0 (6.0), respectively. HD, hemodialysis; ALP, alkaline phosphatase; PTH, parathyroid hormone
Baseline demographic characteristics, biochemical data, and outcomes were collected between January 1, 2001, and December 31, 2017. A 2-year look-back period prior to dialysis initiation was used to ascertain comorbidities and medication history (Fig. S1). Biomarker trajectories were defined over the first 2 years following MHD initiation. To ensure temporal separation between trajectory classification and outcome assessment, a landmark design was adopted, with follow-up commencing at the end of the trajectory window and continuing until death or December 31, 2017, whichever occurred first. Outcomes were defined as the first event occurring after the trajectory period, and events occurring within the trajectory window were not included (Fig. S1).
Ethics approval and consent to participate
This study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Ping Tung Christian Hospital (IRB No. 724 A). The requirement for informed consent was waived due to the retrospective design of the study and the use of de-identified data.
Variables and outcomes
The independent variables of interest were the longitudinal trajectories of ALP and PTH, identified through group-based trajectory modeling (GBTM). During the study period, ALP levels were measured monthly using an automated colorimetric assay (Roche Diagnostics, Indianapolis, IN, USA), and PTH levels were measured every 3 months using a second-generation electrochemiluminescence immunoassay (Roche Diagnostics, Mannheim, Germany); the same assay platforms were used throughout the study. To harmonize measurement frequency, only ALP data obtained in the same months as the PTH level data were considered. For the GBTM analysis, only participants with complete data for all eight scheduled measurements of both biomarkers during the 2-year trajectory period were included.
The covariates comprised demographic characteristics, comorbid conditions, and biochemical measures. Diabetes, hypertension, and hyperlipidemia were defined on the basis of ICD codes plus medication use; other comorbidities and outcomes were defined on the basis of ICD codes alone (Table S1). Major adverse cardiovascular events (MACEs) included hospitalization for any type of stroke, ischemic heart disease, or peripheral artery disease. Cardiovascular death was defined as in-hospital death with a cardiovascular diagnosis. Any-site fractures were defined as fractures at all skeletal locations excluding the skull and restricted to non–traffic accident–related events, whereas major-site fractures were defined as those occurring at the hip, vertebrae, or shoulder. In any event involving multiple fractures, we classified fractures according to the most proximal site and determined whether the event qualified as a major-site fracture.
Statistical analysis
All statistical analyses were performed using SAS (version 9.4; SAS Institute, Cary, NC, USA). Intergroup differences in continuous variables, expressed as medians (IQRs) due to their skewed distributions, were assessed using the Kruskal–Wallis test. Categorical variables, expressed as counts (percentages), were compared between groups using the chi-square or Fisher’s exact test, as appropriate. Because of their skewed distributions, ALP and PTH data were log-transformed.
GBTM was used to identify distinct longitudinal patterns of ALP and PTH [14]. Listwise exclusion was applied to retain only patients with complete data. The optimal model was selected on the basis of the highest Bayesian information criterion (BIC), the greatest posterior probability of group membership, and the smallest group size (proportion per class > 5%) [14]. Extremely small group sizes could result in only one or two outcome events; however, NHIRD privacy regulations prohibit the export of these data. Therefore, we avoided including excessively small groups whenever possible.
The associations of ALP and PTH trajectories with outcomes were evaluated using multivariable Cox proportional hazards models. Covariates included in the multivariable models were selected a priori based on clinical relevance and potential confounding. Specifically, we included (1) baseline demographic and clinical characteristics showing differences across trajectory groups, (2) variables known to be associated with ALP and PTH metabolism, and (3) established risk factors for adverse outcomes. The proportional hazards assumption was assessed using Schoenfeld residual–based tests, with no meaningful violations observed. Continuous variables were modeled as continuous linear terms in the Cox proportional hazards models. No categorization or nonlinear transformations, including restricted cubic splines, were applied. Given the potential mediating role of PTH in the relationship between ALP trajectories and outcomes, we performed sensitivity analyses to examine ALP trajectories with and without adjustment for PTH levels.
Results
Trajectory classification and model selection
According to GBTM methodological criteria [14], both 3- and 4-group models were considered potentially appropriate, as they demonstrated favorable model fit (higher BIC and posterior probabilities of assignment) and satisfied the minimum class size requirement (proportion per class > 5%; Supplementary Table S2). However, under the 3-group solution, the number of CV death events within certain ALP and PTH trajectory groups was already too low to permit data export from the database. A 4-group solution was therefore expected to further reduce the number of events per group, resulting in additional non-exportable data. Moreover, in the 4-group ALP model, one trajectory group comprised only 5.2% of the cohort, approaching the lower acceptable threshold. For PTH, the BIC values between the 3- and 4-group models were comparable, indicating no clear advantage of the more complex model. Accordingly, the trajectory groups for both ALP and PTH were specified as three (Fig. 2). Across the 2-year trajectory period, each trajectory group remained relatively stable from baseline. Moreover, no substantial deviations were observed in the longitudinal patterns.
Fig. 2.
(A) ALP and (B) PTH trajectory groups over 24 months in incident hemodialysis patients. Solid lines represent estimated trajectories, and shaded areas indicate 95% CIs. Groups were defined as described in Table 1. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP; PTH, parathyroid hormone; BT-PTH, below-target PTH; LT-PTH, within-lower-target PTH; UT-PTH, within-upper-target PTH
Baseline demographic and clinical characteristics by trajectory group
The three ALP trajectory subgroups were labeled as low-normal (LN-ALP), high-normal (HN-ALP), and high (H-ALP) according to their median ALP levels (56.8, 84.8, and 167.0 U/L, respectively) relative to the reference range (34–104 U/L) (Table 2). The KDIGO guidelines suggest maintaining PTH at 2–9 times the upper normal limit (approximately 130–585 pg/mL) for patients receiving MHD. Thus, PTH trajectory subgroups were labeled according to their median PTH levels as follows: below-target PTH (BT-PTH; 55.6 pg/mL), within-lower-target PTH (LT-PTH; 158.2 pg/mL), and within-upper-target PTH (UT-PTH; 431.4 pg/mL).
Table 2.
Clinical characteristics according to ALP and PTH trajectories defined on the basis of GBTM
| Variablea | ALP trajectories | PTH trajectories | ||||||
|---|---|---|---|---|---|---|---|---|
| LN-ALP (N = 168) |
HN-ALP (N = 223) |
H-ALP (N = 56) |
p valueb | BT-PTH (N = 93) |
LT-PTH (N = 148) |
UT-PTH (N = 97) |
p valueb | |
| ALP | 56.8 (47.7–61.6) | 84.8 (76.3–100.6) | 167.0 (147.8–217.6) | < 0.0001 | 68.1 (57.4–83.1) | 76.1 (60.3–90.5) | 95.8 (69.0-130.1) | < 0.0001 |
| PTH (pg/mL) | 122.6 (67.7–220.4) | 176.2 (95.9–301.5) | 330.8 (126.5–526.5) | < 0.0001 | 55.6 (38.9–68.0) | 158.2 (119.5-206.5) | 431.4 (329.0-596.8) | < 0.0001 |
| spKt/Vc | 1.7 (1.5–1.9) | 1.7 (1.5–1.9) | 1.7 (1.6–1.9) | 0.46 | 1.7 (1.6–1.9) | 1.7 (1.6–1.9) | 1.7 (1.5–1.8) | 0.0210 |
| nPCRd (g/kg/day) | 1.1 (1.0–1.3) | 1.1(1.0–1.3) | 1.2 (1.1–1.4) | 0.11 | 1.1 (0.9–1.3) | 1.1 (1.0-1.3) | 1.2 (1.0-1.3) | 0.0467 |
| Albumin (g/dL) | 3.9 (3.7–4.0) | 3.9 (3.7–4.1) | 3.9 (3.6–4.1) | 0.89 | 3.7 (3.6–3.9) | 3.9 (3.7–4.1) | 4.0 (3.9–4.2) | < 0.0001 |
| Ferritin (ng/mL) | 472.8 (369.0–619.4) | 477.9 (334.4–621.4) | 533.0 (404.8–636.6) | 0.70 | 529.0 (441.4-667.2) | 491.2 (373.3-606.7) | 398.9 (298.3-535.5) | 0.0003 |
| Cae (mg/dL) | 9.2 (9.0–9.5) | 9.2 (8.9–9.5) | 9.0 (8.6–9.6) | 0.07 | 9.3 (9.2–9.6) | 9.1 (8.9–9.3) | 9.1 (8.7–9.5) | < 0.0001 |
| Phosphate (mg/dL) | 5.1 (4.4–5.8) | 5.0 (4.3–5.7) | 5.1 (4.1–5.8) | 0.22 | 4.6 (4.2–5.3) | 4.9 (4.4–5.3) | 5.7 (5.1–6.2) | < 0.0001 |
| Hemoglobin (g/dL) | 10.3 (10.0–10.6) | 10.5 (9.9–10.8) | 10.2 (9.8–10.6) | 0.06 | 10.3 (10.1–10.7) | 10.4 (10.0-10.7) | 10.4 (10.0-10.8) | 0.91 |
| AST (U/L) | 18.9 (16.5–23.1) | 20.0 (17.5–24.3) | 24.4 (20.6–29.5) | < 0.0001 | 21.4 (17.0-26.4) | 19.9 (17.4–24.5) | 19.0 (15.7–22.4) | 0.0423 |
| ALT (U/L) | 15.5 (12.9–18.5) | 16.7 (13.8–21.6) | 18.3 (15.3–27.8) | < 0.0001 | 17.1 (13.3–23.1) | 16.2 (13.4–20.7) | 15.8 (13.6–18.9) | 0.62 |
| Total Bilirubin (mg/dL) | 0.6 (0.6–0.7) | 0.6 (0.6–0.7) | 0.7 (0.6–0.8) | 0.0006 | 0.6 (0.6–0.7) | 0.6 (0.6–0.7) | 0.6 (0.6–0.7) | 0.23 |
| Cholesterol (mg/dL) | 174.9 (155.6–187.9) | 172.2 (153.3–189.0) | 173.1 (151.0–194.5) | 0.89 | 171.0 (150.8-185.7) | 171.8 (153.7-182.2) | 182.4 (154.6-194.5) | 0.0480 |
| Triglyceride (mg/dL) | 141.9 (98.5–182.2) | 121.6 (94.2–173.4) | 126.1 (95.7–163.3) | 0.37 | 133.0 (90.8-187.4) | 134.2 (96.8-175.8) | 133.9 (103.9-174.4) | 0.75 |
| Creatinine (mg/dL) | 10.0 (8.7–11.8) | 9.9 (8.3–11.5) | 9.6 (7.8–11.9) | 0.22 | 8.7 (7.4–9.9) | 9.6 (8.4–11.3) | 11.2 (9.9–12.9) | < 0.0001 |
| Uric acid (mg/dL) | 7.0 (6.5–7.6) | 6.9 (6.3–7.8) | 7.4 (6.5–7.9) | 0.18 | 6.8 (6.0-7.5) | 6.9 (6.3–7.6) | 7.5 (7.0-8.3) | < 0.0001 |
| Na (mEq/L) | 136.4 (134.9–137.9) | 136.2 (134.8–137.6) | 136.4 (134.3–138.3) | 0.42 | 135.6 (134.3-136.7) | 136.1 (134.8-137.6) | 137.3 (136.1-138.4) | < 0.0001 |
| Medicines, n (%) | ||||||||
| Ca-based binders | 83 (49.4) | 89 (39.9) | 31 (55.4) | 0.0489 | 38 (40.9) | 53 (35.8) | 48 (49.5) | 0.10 |
| Calcitriol | 56 (33.3) | 95 (42.6) | 33 (58.9) | 0.0028 | 18 (19.4) | 59 (39.9) | 71 (73.2) | < 0.0001 |
| Al-based binders | 27 (16.1) | 36 (16.1) | 14 (25.0) | 0.26 | 14 (15.1) | 13 (8.8) | 19 (19.6) | 0.0487 |
Serum biochemistry parameters represent median values across the 2-year trajectory window, whereas medication use was recorded during this period
Groups were defined as described in Table 1. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP; PTH, parathyroid hormone; BT-PTH, below-target PTH; LT-PTH, within-lower-target PTH; UT-PTH, within-upper-target PTH
ap values were calculated separately across ALP- and PTH-based trajectory groups
bspKt/V; Kt/V (single pool) calculated according to the Daugirdas II formula
cnPCR was calculated from urea kinetic modeling according to standard methods
dCorrected Ca, albumin-adjusted Ca (mg/dL) = total Ca (mg/dL) + 0.8 × [4.0 − albumin (g/dL)]
Abbreviations: Al-based binders, aluminum-based phosphate binders; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Ca, calcium; Ca-based binders, calcium-based phosphate binders; GBTM, group-based trajectory modeling; Na, sodium; nPCR, normalized protein catabolic rate; spKt/V, single-pool Kt/V
The H-ALP group had the lowest prevalence of hypertension and dyslipidemia (Table 1). During the 2-year trajectory period, the H-ALP group was characterized by markedly elevated PTH levels and liver function indexes, as well as the lowest hemoglobin levels (Table 2). Notably, the H-ALP group also had the highest proportion of patients receiving calcium-containing phosphate binders and calcitriol, indicating relatively intensive management of secondary hyperparathyroidism (SHPT).
Table 1.
Baseline demographic according to ALP and PTH trajectories defined on the basis of GBTM
| Variable | ALP trajectories | PTH trajectories | ||||||
|---|---|---|---|---|---|---|---|---|
| LN-ALP (N = 168) |
HN-ALP (N = 223) |
H-ALP (N = 56) |
p valuea | BT-PTH (N = 93) |
LT-PTH (N = 148) |
UT-PTH (N = 97) |
p valuea | |
| Female | 95 (56.6) | 115 (51.6) | 36 (64.3) | 0.20 | 55 (59.14) | 78 (52.70) | 54 (55.67) | 0.62 |
| Age (y)b | 63 (52–71) | 63 (52–72) | 61 (52–67) | 0.32 | 67 (57–74) | 64 (55–72) | 54 (47–63) | < 0.0001 |
| ≧ 65 (n, %) | 71 (42.3) | 98 (44.0) | 20 (35.7) | 0.54 | 52 (55.91) | 69 (46.62) | 22 (22.68) | < 0.0001 |
| Follow-up (y) | 4.8 (3.2–7.1) | 4.0 (2.2–7.0) | 4.8 (1.8–7.3) | 0.18 | 3.9 (1.6-6.0) | 4.4 (2.9–6.4) | 5.2 (3.7–7.6) | 0.0005 |
| Comorbidities, n (%) | ||||||||
| Diabetes | 88 (52.4) | 131 (58.7) | 25 (44.6) | 0.13 | 68 (73.1) | 81 (54.7) | 36 (37.1) | < 0.0001 |
| Hypertension | 112 (66.7) | 145 (65.0) | 24 (42.9) | 0.0039 | 59 (63.4) | 101 (68.2) | 51 (52.6) | 0.0453 |
| Hyperlipidemia | 44 (26.2) | 66 (29.6) | 7 (12.5) | 0.0339 | 26 (28.0) | 44 (29.7) | 22 (22.7) | 0.47 |
| ASCVDc | 40 (23.8) | 43 (19.3) | 11 (19.6) | 0.53 | 24 (25.8) | 40 (27.0) | 10 (10.3) | 0.0047 |
| HF | 33 (19.6) | 38 (17.0) | 10 (17.9) | 0.80 | 17 (18.3) | 32 (21.6) | 11 (11.3) | 0.12 |
| Prior fractured | 10 (6.0) | 16 (7.2) | 3 (5.4) | 0.83 | 10 (10.8) | 9 (6.1) | 3 (3.1) | 0.10 |
| Prior Oral corticosteroidse | 82 (48.8) | 121 (54.3) | 25 (44.6) | 0.34 | 58 (62.4) | 70 (47.3) | 44 (45.4) | 0.0325 |
Group definitions: Study subjects were categorized according to ALP and PTH trajectories. ALP trajectory groups were named based on their median values across the 2-year trajectory window relative to the reference range, whereas PTH trajectory groups were defined according to their median values in relation to the KDIGO-recommended target range. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP; PTH, parathyroid hormone; BT-PTH, below-target PTH; LT-PTH, within-lower-target PTH; UT-PTH, within-upper-target PTH
ap values were calculated separately across ALP- and PTH-based trajectory groups
bAge at MHD initiation
cASCVD: any type of stroke, coronary artery disease, or peripheral artery disease
dBone fracture history, all sites, before HD initiation
eOral corticosteroid use before HD initiation
Abbreviations: ASCVD, atherosclerotic cardiovascular disease; GBTM, group-based trajectory modeling; HF, heart failure
The BT-PTH group had the oldest age and the highest prevalence of diabetes (Table 1). Biochemically, this group had the lowest ALP levels and the highest Kt/V (Table 2). The BT-PTH group also had higher serum ferritin and albumin-adjusted calcium levels but lower nutritional markers, including serum albumin, as well as lower phosphate and uric acid levels, than the other groups. (Table 2). Regarding medication use, the BT-PTH group had the lowest calcitriol use but the highest exposure to acid-suppressing medications during the trajectory period.
Influence of ALP and PTH trajectories on outcomes
ALP trajectories
During the follow-up period after the 2-year trajectory period, the incidence of all-cause mortality was 11.2, 13.2, and 10.6 per 100 patient-years in the LN-ALP, HN-ALP, and H-ALP groups, respectively (Table 3). Table 3 summarizes the incidence of other clinical outcomes across ALP and PTH trajectory groups.
Table 3.
Incidence rates of clinical outcomes across ALP and PTH trajectories
| ALP | p value | PTH | p value | |||||
|---|---|---|---|---|---|---|---|---|
| LN-ALP (N = 168) |
HN-ALP (N = 223) |
H-ALP (N = 56) |
BT-PTH (N = 93) |
LT-PTH (N = 148) |
UT-PTH (N = 97) |
|||
| Fracture, all sites | ||||||||
| Event | 35 | 38 | 12 | 0.57 | 18 | 29 | 20 | 0.97 |
| Follow-up years, mean | 4.4 | 4.1 | 4.2 | 3.6 | 4.0 | 5.1 | ||
| Follow-up period, patient-years | 735.3 | 923.6 | 237.1 | 333.4 | 594.5 | 490.3 | ||
| Incidence, per 100 patient-years | 4.8 | 4.1 | 5.1 | 5.4 | 4.9 | 4.1 | ||
| Fracture, major sites | ||||||||
| Event | 22 | 25 | 9 | 0.59 | 13 | 17 | 13 | 0.83 |
| Follow-up years, mean | 4.7 | 4.3 | 4.5 | 3.8 | 4.4 | 5.3 | ||
| Follow-up period, patient-years | 793.7 | 959.0 | 251.6 | 352.1 | 648.6 | 511.9 | ||
| Incidence, per 100 patient-years | 2.8 | 2.6 | 3.6 | 3.7 | 2.6 | 2.5 | ||
| MACE | ||||||||
| Event | 76 | 111 | 24 | 0.53 | 47 | 77 | 47 | 0.86 |
| Follow-up years, mean | 3.9 | 3.5 | 4.0 | 2.8 | 3.6 | 4.5 | ||
| Follow-up period, patient-years | 656.8 | 785.9 | 223.5 | 262.0 | 539.8 | 437.7 | ||
| Incidence, per 100 patient-years | 11.6 | 14.1 | 10.7 | 17.9 | 14.3 | 10.7 | ||
| CV death | ||||||||
| Event | 18* | 0.13 | 13* | 0.39 | ||||
| Follow-up years, mean | 5.2 | 4.7 | 4.7 | 4.2 | 4.7 | 5.7 | ||
| Follow-up period, patient-years | 866.4 | 1042.8 | 264.8 | 389.2 | 698.7 | 554.4 | ||
| Incidence, per 100 patient-years | NA | NA | ||||||
| All-cause mortality | ||||||||
| Event | 97 | 138 | 28 | 0.25 | 63 | 93 | 36 | < 0.0001 |
| Follow-up years, mean | 5.2 | 4.7 | 4.7 | 4.2 | 4.7 | 5.7 | ||
| Follow-up period, patient-years | 866.4 | 1042.8 | 264.8 | 389.2 | 698.7 | 554.4 | ||
| Incidence, per 100 patient-years | 11.2 | 13.2 | 10.6 | 16.2 | 13.3 | 6.5 | ||
Groups were defined as described in Table 1. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP; PTH, parathyroid hormone; BT-PTH, below-target PTH; LT-PTH, within-lower-target PTH; UT-PTH, within-upper-target PTH
The PTH trajectory analysis was conducted in a subset of the ALP cohort after excluding patients with insufficient PTH data; therefore, the sample sizes and event numbers differ between the two analyses
*In accordance with NHIRD privacy regulations, export data for any group with one or two outcome events is prohibited; only aggregated totals are permitted for export
Abbreviations: CV, cardiovascular; MACE, major adverse cardiovascular event
The crude Kaplan–Meier curves for all outcomes indicated no significant differences between the ALP trajectory groups (Fig. 3). In a univariate Cox regression analysis, the HN-ALP group exhibited a significantly higher cardiovascular death risk than the other trajectory groups (Table 4). After multivariable adjustment, HN-ALP remained independently associated with both cardiovascular death [adjusted hazard ratio (HR), 8.44; 95% confidence interval (CI), 1.78–39.96; P = 0.0072] and all-cause mortality (HR, 1.54; 95% CI, 1.15–2.06; P = 0.0034; Table 4). These results were consistent with those of the sensitivity analysis (excluding PTH from the multivariable models), in which the HN-ALP group continued to demonstrate significantly higher cardiovascular death and all-cause mortality risks than the other groups (Table 5).
Fig. 3.
Kaplan–Meier curves for clinical outcomes stratified by ALP (A–E) and PTH (F–J) trajectories. Groups were defined as described in Table 1. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP; PTH, parathyroid hormone; BT-PTH, below-target PTH; LT-PTH, within-lower-target PTH; UT-PTH, within-upper-target PTH
Table 4.
Univariate and multivariable Cox regression of ALP
| Univariate analysis | p value | Multivariable analysis | p value | |||||
|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | HR | 95% CI | |||||
| Fracture, all sites | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 0.85 | 0.54 | 1.35 | 0.50 | 0.93 | 0.56 | 1.53 | 0.76 |
| H-ALP | 1.05 | 0.55 | 2.03 | 0.88 | 1.44 | 0.65 | 3.20 | 0.38 |
| Fracture, major sites | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 0.94 | 0.53 | 1.66 | 0.82 | 1.06 | 0.56 | 2.00 | 0.87 |
| H-ALP | 1.29 | 0.59 | 2.80 | 0.52 | 2.49 | 0.94 | 6.59 | 0.07 |
| MACE | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 1.18 | 0.88 | 1.58 | 0.26 | 1.26 | 0.92 | 1.74 | 0.15 |
| H-ALP | 0.93 | 0.59 | 1.47 | 0.74 | 1.10 | 0.65 | 1.88 | 0.72 |
| CV death | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 3.62 | 1.03 | 12.69 | 0.0449 | 8.44 | 1.78 | 39.96 | 0.0072 |
| H-ALP | 2.25 | 0.38 | 13.49 | 0.37 | 3.79 | 0.40 | 36.17 | 0.25 |
| All-cause mortality | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 1.19 | 0.92 | 1.55 | 0.19 | 1.54 | 1.15 | 2.06 | 0.0034 |
| H-ALP | 0.95 | 0.62 | 1.44 | 0.79 | 1.34 | 0.82 | 2.18 | 0.24 |
Groups were defined as described in Table 1. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP
Model was adjusted for sex, age, diabetes, hypertension, hyperlipidemia, prior ASCVD, heart failure, Kt/V (single pool), nPCR, serum albumin, albumin-adjusted calcium, phosphate, PTH, hemoglobin, GOT, GPT, total bilirubin, creatinine, uric acid, and sodium. Covariates were selected a priori based on clinical relevance, including variables differing across trajectory groups, factors related to ALP/PTH metabolism, and established predictors of study outcomes
Abbreviations: CV, cardiovascular; MACE, major adverse cardiovascular event
Table 5.
Univariate and multivariable Cox regression of ALP without adjustments for PTH
| Univariate analysis | p value | Multivariable analysis | p value | |||||
|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | HR | 95% CI | |||||
| Fracture, all sites | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 0.85 | 0.54 | 1.35 | 0.4971 | 0.82 | 0.51 | 1.33 | 0.4187 |
| H-ALP | 1.05 | 0.55 | 2.03 | 0.8765 | 1.11 | 0.54 | 2.29 | 0.7849 |
| Fracture, major sites | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 0.94 | 0.53 | 1.66 | 0.8217 | 0.86 | 0.47 | 1.58 | 0.6182 |
| H-ALP | 1.29 | 0.59 | 2.80 | 0.5243 | 1.66 | 0.69 | 3.99 | 0.2549 |
| MACE | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 1.18 | 0.88 | 1.58 | 0.2612 | 1.25 | 0.92 | 1.70 | 0.1610 |
| H-ALP | 0.93 | 0.59 | 1.47 | 0.7448 | 1.06 | 0.65 | 1.74 | 0.8067 |
| CV death | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 3.62 | 1.03 | 12.69 | 0.0449 | 8.97 | 1.94 | 41.48 | 0.0050 |
| H-ALP | 2.25 | 0.38 | 13.49 | 0.3734 | 4.53 | 0.50 | 40.78 | 0.1779 |
| All-cause mortality | ||||||||
| LN-ALP | Ref. | Ref. | ||||||
| HN-ALP | 1.19 | 0.92 | 1.55 | 0.1863 | 1.41 | 1.07 | 1.86 | 0.0155 |
| H-ALP | 0.95 | 0.62 | 1.44 | 0.7933 | 1.14 | 0.72 | 1.80 | 0.5868 |
Groups were defined as described in Table 1. ALP, alkaline phosphatase; LN-ALP, low-normal ALP; HN-ALP, high-normal ALP; H-ALP, high ALP; PTH, parathyroid hormone
Model was adjusted for sex, age, diabetes, hypertension, hyperlipidemia, prior ASCVD, heart failure, Kt/V (single pool), nPCR, serum albumin, albumin-adjusted calcium, phosphate, PTH, hemoglobin, GOT, GPT, total bilirubin, creatinine, uric acid, and sodium. Covariates were selected a priori as described in Table 4, based on clinical relevance
Abbreviations: CV, cardiovascular; MACE, major adverse cardiovascular event
PTH trajectories
The incidence of all-cause mortality was 16.2, 13.3, and 6.5 per 100 patient-years in the BT-PTH, LT-PTH, and UT-PTH groups, respectively (Table 3). Kaplan–Meier curves indicated that the BT-PTH group had the highest all-cause mortality (log-rank P < 0.0001; Fig. 3J). In univariate Cox regression, the BT-PTH group was associated with increased risks of MACE and all-cause mortality; however, these associations were attenuated after multivariable adjustment (Table 6).
Table 6.
Univariate and multivariable Cox regression of PTH
| Univariate analysis | p value | Multivariable analysis | p value | |||||
|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | HR | 95% CI | |||||
| Fracture, all sites | ||||||||
| UT-PTH | Ref. | Ref. | ||||||
| LT-PTH | 1.18 | 0.67 | 2.09 | 0.57 | 0.62 | 0.25 | 1.50 | 0.28 |
| BT-PTH | 1.26 | 0.67 | 2.39 | 0.48 | 0.49 | 0.15 | 1.61 | 0.24 |
| Fracture, major sites | ||||||||
| UT-PTH | Ref. | Ref. | ||||||
| LT-PTH | 1.00 | 0.49 | 2.07 | 0.99 | 0.49 | 0.15 | 1.58 | 0.23 |
| BT-PTH | 1.40 | 0.65 | 3.03 | 0.39 | 0.43 | 0.09 | 2.01 | 0.28 |
| MACE | ||||||||
| UT-PTH | Ref. | Ref. | ||||||
| LT-PTH | 1.30 | 0.90 | 1.86 | 0.16 | 0.93 | 0.54 | 1.62 | 0.81 |
| BT-PTH | 1.60 | 1.07 | 2.40 | 0.02 | 0.88 | 0.43 | 1.78 | 0.71 |
| CV death | ||||||||
| UT-PTH | Ref. | Ref. | ||||||
| LT-PTH | 2.98 | 0.63 | 14.06 | 0.17 | 0.44 | 0.02 | 11.35 | 0.62 |
| BT-PTH | 2.08 | 0.35 | 12.47 | 0.42 | 0.10 | 0.00 | 7.98 | 0.30 |
| All-cause mortality | ||||||||
| UT-PTH | Ref. | Ref. | ||||||
| LT-PTH | 2.07 | 1.41 | 3.04 | 0.0002 | 1.05 | 0.57 | 1.95 | 0.87 |
| BT-PTH | 2.53 | 1.68 | 3.81 | < 0.0001 | 1.02 | 0.47 | 2.24 | 0.95 |
Groups were defined as described in Table 1. PTH, parathyroid hormone; BT-PTH, below-target PTH; LT-PTH, within-lower-target PTH; UT-PTH, within-upper-target PTH
Model was adjusted for sex, age, diabetes, hypertension, hyperlipidemia, prior ASCVD, heart failure, Kt/V (single pool), nPCR, serum albumin, calcium, phosphate, PTH, hemoglobin, GOT, GPT, total bilirubin, creatinine, uric acid, and sodium. Covariates were selected a priori as described in Table 4, based on clinical relevance
Abbreviations: CV, cardiovascular; MACE, major adverse cardiovascular event
Discussion
Previous studies examining longitudinal changes in ALP and PTH, as well as their impact on clinical outcomes, are limited. In this cohort of incident MHD patients, distinct ALP and PTH trajectories were already evident at the time of MHD initiation. The BT-PTH trajectory was characterized by older age, higher diabetes prevalence, and biochemical malnutrition indicators. The HN-ALP trajectory was independently associated with increased mortality, whereas the BT-PTH group exhibited higher crude mortality.
In the present study, ALP and PTH trajectories were already discernible at MHD initiation, and the trajectory classes exhibited differing clinical characteristics (e.g., comorbidities and nutritional indicators) during this early phase of dialysis, suggesting that these patterns may originate in earlier stages of CKD. A previous study demonstrated that pre-ESRD ALP levels, averaged over the 6 months prior to dialysis initiation, were predictive of post-dialysis mortality [15]. In addition, renal osteodystrophy develops in the early stages of CKD [16] and has been associated with increased mortality [17, 18]. Collectively, these findings support the importance of longitudinal monitoring of bone biomarkers, including ALP and PTH, from earlier CKD stages.
In the current study, both ALP and PTH trajectories remained stable over the two-year period, suggesting that they primarily represent persistent exposure states rather than rapidly changing or crossing patterns. Because ALP reflects bone turnover, which evolves over years, the observed stability of ALP trajectories in our cohort is biologically plausible. PTH trajectories were likewise relatively stable. This pattern may, in part, reflect population-specific characteristics. Our cohort consisted of Taiwanese patients, representing an Asian population, and prior study has shown that Japanese patients tend to maintain stable PTH levels over time, whereas increases are more commonly observed in European and North American dialysis populations [19], suggesting that Asian patients may exhibit less fluctuation in PTH. In addition, features of the study design may have contributed to this stability. Specifically, the study period (2004–2012) preceded the widespread use of calcimimetics, and patients who underwent parathyroidectomy were excluded, both of which would limit major therapeutic or surgical perturbations of PTH levels. Finally, the modest sample size may have reduced the ability to detect more subtle dynamic changes. The selection of a two-year observation window was based on prior prognostic studies [4]; however, a longer observation period may be required to capture the full extent of long-term ALP dynamics. At the same time, short-term fluctuations in ALP also occur and have been shown to predict mortality during CKD progression [20]. Similarly, PTH is highly dynamic and responsive to calcium, phosphate, and vitamin D therapies. Therefore, future studies with larger sample sizes are needed, incorporating multiple trajectory window lengths (e.g., long-term ≥ 3 years and shorter-term < 2 years), together with measures of biomarker variability. Such approaches may better capture nuanced temporal changes and more precisely characterize the evolution of ALP and PTH trajectories in patients undergoing HD.
The observation that the trajectories were largely parallel but at different levels, together with underlying differences in baseline demographic and clinical characteristics over the 2-year period, suggests that these groups likely represent distinct exposure profiles. Importantly, the observed differences in clinical outcomes across trajectory groups further support their relevance as clinically meaningful and distinct longitudinal phenotypes. These findings support the use of GBTM to characterize the longitudinal dynamics of biomarkers in HD patients and to further elucidate their relationships with clinical outcomes.
Studies examining baseline determinants of ALP and PTH trajectories have been scarce. Consistent with a prior study [21], we observed that high ALP levels were accompanied by elevated PTH concentrations, likely compatible with SHPT-related high-turnover bone disease and increased bone-specific ALP (bs-ALP) levels [22]. Because bs-ALP is linked to adverse outcomes [23], bs-ALP measurement may provide additional value in patients with persistently high ALP. Regarding PTH, the BT-PTH trajectory was associated with the lowest ALP levels, advanced age, highest Kt/V, diabetes, elevated ferritin and calcium, and reduced nutritional markers, consistent with those reported previously [21]. Future studies should clarify the determinants of these trajectories and assess whether early intervention can modify trajectory patterns and improve outcomes.
Elevated ALP levels have been associated with increased mortality in patients with CKD and those receiving dialysis [11, 24, 25], and this association appears to be dose-dependent [4]. Longitudinal changes in ALP may provide additional prognostic information beyond baseline values. For example, elevated ALP during the first 6 months after dialysis initiation has been associated with an increased risk of death during subsequent follow-up [5]. Greater ALP variability has also been associated with increased all-cause mortality [20]. The association between elevated ALP and mortality has been linked to vascular calcification and SHPT [26, 27]. In the present study, the HN-ALP trajectory was independently associated with higher cardiovascular and all-cause mortality, although the 95% CI for the hazard ratio of cardiovascular mortality was wide because of the small number of cardiovascular death events. One possible explanation for why the prognostic signal appeared stronger in the HN-ALP trajectory than in the H-ALP trajectory is that, compared with the H-ALP group, the HN-ALP group had a higher burden of comorbidities, such as hypertension and dyslipidemia, both of which are known to be associated with poor prognosis. The absence of statistically significant associations in the H-ALP trajectory group may be explained by two factors. First, because patients who died during the trajectory window were excluded and only those with complete biomarker measurements were included, high-risk patients may have been systematically excluded from the trajectory groups, thereby attenuating the observed associations between adverse ALP trajectories and subsequent clinical outcomes. Second, the small sample size in this subgroup may have limited statistical power. Therefore, this null finding should be interpreted as a probable type II error rather than evidence of a true lack of biological effect. Taken together, our findings, along with the established links between elevated ALP and mortality, suggest that the unfavorable ALP trajectories contributing to adverse outcomes may be broad. Given the multiple origins of ALP (skeletal versus extraskeletal), the underlying mechanisms linking ALP to clinical outcomes are likely complex. Future studies should more carefully examine subtle changes in ALP trajectories to better clarify the mechanisms underlying ALP-related mortality.
PTH levels have been associated with increased mortality, with a U-shaped relationship showing increased risk at very low (< 150 pg/mL) and very high (> 600 pg/mL) levels [4, 28–34]. Longitudinal studies have further shown that decreasing PTH levels are associated with higher mortality, even when baseline values fall within guideline-recommended ranges [11, 35]. A large cohort of incident HD patients followed for a median of 3 years demonstrated that if iPTH was below the lowest risk range at baseline (< 239 ng/L), a subsequent increase in levels was associated with improved survival [36]. In our study, the BT-PTH trajectory showed crude associations with both MACE and all-cause mortality. Therefore, sustained or progressive suppression of PTH below the KDIGO-recommended range should be avoided in patients undergoing maintenance HD. The BT-PTH group was characterized by older age, a higher burden of comorbidities, and poorer nutritional status; notably, malnutrition is one of the proposed mechanisms linking decreased PTH levels to adverse outcomes [32]. Future studies should investigate whether nutritional interventions can modify unfavorable PTH trajectories and improve outcomes in CKD. The LT-PTH trajectory was also significantly associated with all-cause mortality. This may be because the median PTH level in the LT-PTH group (158.2 pg/mL) lies closer to the lower boundary of the KDIGO-recommended target range (130–585 pg/mL), whereas the UT-PTH group (431.4 pg/mL) remains relatively distant from boundary. Larger trajectory-based studies are needed to more precisely define the optimal PTH trajectory associated with improved survival. Furthermore, given that our cohort demonstrated associations of poor outcomes with HN-ALP and BT-PTH, considering that elevated ALP is often accompanied by elevated PTH, future studies incorporating joint ALP–PTH trajectory modeling may better elucidate their combined effects on clinical outcomes.
Limitations
This study has several limitations. First, unmeasured confounders, including inflammatory status and vitamin D levels, may have influenced the observed associations. Second, our dataset predates calcimimetic use, which may limit generalizability to current practice. Third, the modest sample size may have reduced sensitivity in detecting subtle trajectory patterns and precluded robust subgroup analyses. Fourth, excluding patients who died during the trajectory window may have introduced survivorship bias, potentially underestimating the prognostic value of early unfavorable trajectories. Furthermore, the requirement for complete longitudinal ALP and PTH data for trajectory modeling may have resulted in selection bias, potentially limiting the representativeness of the study population. Future studies using shorter observation periods (e.g., ≤ 1 year) and applying statistical approaches to handle missing data may help mitigate these biases. Fifth, bs-ALP was not measured, which may have reduced specificity for bone turnover assessment. Moreover, the absence of key regulators (e.g., fibroblast growth factor 23 and 1,25-dihydroxyvitamin D) may have confounded interpretations of ALP and PTH dynamics and their prognostic implications. Sixth, treatment changes during the trajectory period (e.g., phosphate binders, calcitriol dose, dialysis prescription) were not accounted for, which may have influenced biomarker evolution. Seventh, we did not comprehensively assess potential ALP–PTH interactions or apply competing-risk models for fracture outcomes. Eighth, the number of covariates included in the multivariable models was relatively large in relation to the number of events for certain outcomes. This may introduce the potential for overadjustment, overfitting, and limited estimate stability. Ninth, this study is based on a retrospective observational design, which inherently limits causal inference. Finally, the NHIRD is an administrative claims–based database in which lifestyle factors are unavailable and diagnostic accuracy depends on coding quality. To partially mitigate these limitations, NHIRD claims data were linked to hospital electronic medical records, complete laboratory data were obtained, and major comorbidities were validated using both diagnostic codes and medication records.
Conclusions
In this cohort, ALP and PTH trajectories are identifiable early after dialysis initiation and represent clinically meaningful exposure patterns. The HN-ALP trajectory was independently associated with adverse outcomes, while the BT-PTH trajectory was also linked to an increased risk of mortality. These findings underscore the importance of early longitudinal monitoring and avoiding excessive PTH suppression, supporting trajectory-based approaches to improve risk stratification and guide management in HD patients. Future large-scale prospective studies are warranted to clarify subtle longitudinal changes in ALP and PTH and their outcome associations.
Electronic Supplementary Material
Below is the link to the electronic supplementary material.
Acknowledgements
We thank the Center for Medical Informatics and Statistics of Kaohsiung Medical University for providing statistical and administrative support.
Abbreviations
- ALP
Alkaline phosphatase
- ALT
Alanine aminotransferase
- AST
Aspartate aminotransferase
- ASCVD
Atherosclerotic cardiovascular disease
- BT-PTH
Below-target parathyroid hormone
- CKD
Chronic kidney disease
- CKD-MBD
Chronic kidney disease–mineral and bone disorder
- CV
Cardiovascular
- GBTM
Group-based trajectory modeling
- HD
Hemodialysis
- HF
Heart failure
- HN-ALP
High-normal alkaline phosphatase
- H-ALP
High alkaline phosphatase
- ICD
International Classification of Diseases
- IQR
Interquartile range
- KDIGO
Kidney Disease: Improving Global Outcomes
- LN-ALP
Low-normal alkaline phosphatase
- LT-PTH
Within-lower-target parathyroid hormone
- MACE
Major adverse cardiovascular event
- MHD
Maintenance hemodialysis
- NHIRD
National Health Insurance Research Database
- nPCR
Normalized protein catabolic rate
- PTH
Parathyroid hormone
- spKt/V
Single-pool Kt/V
- UT-PTH
Within-upper-target parathyroid hormone
Author contributions
Conceptualization: H. C. W. and L. C. C. Methodology: H. C. W., L. C. C., and W. S. F. Investigation, validation, and data curation: H. C. W. and W. S. F. Writing—original draft preparation and funding acquisition: H. C. W. Writing—review and editing: W. P. H., L. C. C., L. H. H., and H. S. Y. Supervision: C. C. H. and H. M. L. Project administration: H. C. W. All authors have read and agreed to the published version of the manuscript.
Funding
This research, including the APC, was funded by Ping Tung Christian Hospital (grant no. PS112009).
Data availability
The clinical data used in this study are available from Ping Tung Christian Hospital; however, some restrictions apply to their availability. Access to these data requires approval from the Institutional Review Board of Ping Tung Christian Hospital and cannot be shared publicly. The NHIRD data used in the current study were obtained under license and are not publicly available due to governmental policy restrictions.
Declarations
Ethics approval and consent to participate
This study was approved by the Institutional Review Board of Ping Tung Christian Hospital (IRB No. 724 A) and conducted in accordance with the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective design and use of de-identified data.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Chung-Hwan Chen, Email: hwan@kmu.edu.tw.
Mei-Ling Ho, Email: homelin@kmu.edu.tw.
References
- 1.Naylor KL, McArthur E, Leslie WD, et al. The three-year incidence of fracture in chronic kidney disease. Kidney Int. 2014;86(4):810–8. [DOI] [PubMed] [Google Scholar]
- 2.CKDMBDUWG, Kidney Disease: Improving Global Outcomes. KDIGO 2017 Clinical Practice Guideline Update for the Diagnosis, Evaluation, Prevention, and Treatment of Chronic Kidney Disease-Mineral and Bone Disorder (CKD-MBD). Kidney Int Suppl (2011). 2017;7(1):1–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Yamamoto S, Jorgensen HS, Zhao J, et al. Alkaline Phosphatase and Parathyroid Hormone Levels: International Variation and Associations With Clinical Outcomes in the DOPPS. Kidney Int Rep. 2024;9(4):863–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kalantar-Zadeh K, Kuwae N, Regidor DL, et al. Survival predictability of time-varying indicators of bone disease in maintenance hemodialysis patients. Kidney Int. 2006;70(4):771–80. [DOI] [PubMed] [Google Scholar]
- 5.Regidor DL, Kovesdy CP, Mehrotra R, et al. Serum alkaline phosphatase predicts mortality among maintenance hemodialysis patients. J Am Soc Nephrol. 2008;19(11):2193–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gardham C, Stevens PE, Delaney MP, LeRoux M, Coleman A, Lamb EJ. Variability of parathyroid hormone and other markers of bone mineral metabolism in patients receiving hemodialysis. Clin J Am Soc Nephrol. 2010;5(7):1261–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sardiwal S, Gardham C, Coleman AE, Stevens PE, Delaney MP, Lamb EJ. Bone-specific alkaline phosphatase concentrations are less variable than those of parathyroid hormone in stable hemodialysis patients. Kidney Int. 2012;82(1):100–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chang CH, Liou HH, Wu CK. Moderate-severe aortic arch calcification and high serum alkaline phosphatase co-modify the risk of cardiovascular events and mortality among chronic hemodialysis patients. Ren Fail. 2025;47(1):2449572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Beberashvili I, Samogalska O, Azar A, Stav K, Efrati S. Nutritional Status and Mortality Predictability for Time-Varying Serum Alkaline Phosphatase in Hemodialysis Patients: A Longitudinal Study. J Ren Nutr. 2020;30(5):452–61. [DOI] [PubMed] [Google Scholar]
- 10.Streja E, Wang HY, Lau WL, et al. Mortality of combined serum phosphorus and parathyroid hormone concentrations and their changes over time in hemodialysis patients. Bone. 2014;61:201–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Soohoo M, Feng M, Obi Y, et al. Changes in Markers of Mineral and Bone Disorders and Mortality in Incident Hemodialysis Patients. Am J Nephrol. 2016;43(2):85–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol. 2010;6:109–38. [DOI] [PubMed] [Google Scholar]
- 13.Lin LY, Warren-Gash C, Smeeth L, Chen PC. Data resource profile: the National Health Insurance Research Database (NHIRD). Epidemiol Health. 2018;40:e2018062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Jones BL, Nagin DS. Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociol methods Res. 2007;35(4):542–71. [Google Scholar]
- 15.Sumida K, Molnar MZ, Potukuchi PK, et al. Prognostic significance of pre-end-stage renal disease serum alkaline phosphatase for post-end-stage renal disease mortality in late-stage chronic kidney disease patients transitioning to dialysis. Nephrol Dial Transpl. 2018;33(2):264–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.El-Husseini A, Abdalbary M, Lima F, et al. Low Turnover Renal Osteodystrophy With Abnormal Bone Quality and Vascular Calcification in Patients With Mild-to-Moderate CKD. Kidney Int Rep. 2022;7(5):1016–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bhuriya R, Li S, Chen SC, McCullough PA, Bakris GL. Plasma parathyroid hormone level and prevalent cardiovascular disease in CKD stages 3 and 4: an analysis from the Kidney Early Evaluation Program (KEEP). Am J Kidney Dis. 2009;53(4 Suppl 4):S3–10. [DOI] [PubMed] [Google Scholar]
- 18.Borrelli S, Chiodini P, De Nicola L, et al. Prognosis and determinants of serum PTH changes over time in 1–5 CKD stage patients followed in tertiary care. PLoS ONE. 2018;13(8):e0202417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yamamoto S, Karaboyas A, Komaba H, et al. Mineral and bone disorder management in hemodialysis patients: comparing PTH control practices in Japan with Europe and North America: the Dialysis Outcomes and Practice Patterns Study (DOPPS). BMC Nephrol. 2018;19(1):253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Kim Y, Wu PH, Kwon S, et al. Intra-individual fluctuations in alkaline phosphatase predict mortality and CKD progression: insights from a large CKD cohort. Clin Kidney J. 2026;19(3):sfag022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Li J, Molnar MZ, Zaritsky JJ, et al. Correlates of parathyroid hormone concentration in hemodialysis patients. Nephrol Dial Transpl. 2013;28(6):1516–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sardiwal S, Magnusson P, Goldsmith DJ, Lamb EJ. Bone alkaline phosphatase in CKD-mineral bone disorder. Am J Kidney Dis. 2013;62(4):810–22. [DOI] [PubMed] [Google Scholar]
- 23.Drechsler C, Verduijn M, Pilz S, et al. Bone alkaline phosphatase and mortality in dialysis patients. Clin J Am Soc Nephrol. 2011;6(7):1752–9. [DOI] [PubMed] [Google Scholar]
- 24.Fan Y, Jin X, Jiang M, Fang N. Elevated serum alkaline phosphatase and cardiovascular or all-cause mortality risk in dialysis patients: A meta-analysis. Sci Rep. 2017;7(1):13224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Taliercio JJ, Schold JD, Simon JF, et al. Prognostic importance of serum alkaline phosphatase in CKD stages 3–4 in a clinical population. Am J Kidney Dis. 2013;62(4):703–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kim DW, Hwang SY, Nam YJ, Kim D, Shin SJ, Yoon HE. The combined prognostic significance of alkaline phosphatase and vascular calcification in patients with end-stage kidney disease. Nutr Metab Cardiovasc Dis. 2020;30(9):1476–83. [DOI] [PubMed] [Google Scholar]
- 27.Bozic M, Diaz-Tocados JM, Bermudez-Lopez M, et al. Independent effects of secondary hyperparathyroidism and hyperphosphataemia on chronic kidney disease progression and cardiovascular events: an analysis from the NEFRONA cohort. Nephrol Dial Transpl. 2022;37(4):663–72. [DOI] [PubMed] [Google Scholar]
- 28.Kovesdy CP, Ahmadzadeh S, Anderson JE, Kalantar-Zadeh K. Secondary hyperparathyroidism is associated with higher mortality in men with moderate to severe chronic kidney disease. Kidney Int. 2008;73(11):1296–302. [DOI] [PubMed] [Google Scholar]
- 29.Tentori F, Wang M, Bieber BA, et al. Recent changes in therapeutic approaches and association with outcomes among patients with secondary hyperparathyroidism on chronic hemodialysis: the DOPPS study. Clin J Am Soc Nephrol. 2015;10(1):98–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Lee SA, Lee MJ, Ryu GW, et al. Low serum intact parathyroid hormone level is an independent risk factor for overall mortality and major adverse cardiac and cerebrovascular events in incident dialysis patients. Osteoporos Int. 2016;27(9):2717–26. [DOI] [PubMed] [Google Scholar]
- 31.Wang M, Obi Y, Streja E, et al. Association of Parameters of Mineral Bone Disorder with Mortality in Patients on Hemodialysis according to Level of Residual Kidney Function. Clin J Am Soc Nephrol. 2017;12(7):1118–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Avram MM, Mittman N, Myint MM, Fein P. Importance of low serum intact parathyroid hormone as a predictor of mortality in hemodialysis and peritoneal dialysis patients: 14 years of prospective observation. Am J Kidney Dis. 2001;38(6):1351–7. [DOI] [PubMed] [Google Scholar]
- 33.Fernandez-Martin JL, Martinez-Camblor P, Dionisi MP, et al. Improvement of mineral and bone metabolism markers is associated with better survival in haemodialysis patients: the COSMOS study. Nephrol Dial Transpl. 2015;30(9):1542–51. [DOI] [PubMed] [Google Scholar]
- 34.Naves-Diaz M, Passlick-Deetjen J, Guinsburg A, et al. Calcium, phosphorus, PTH and death rates in a large sample of dialysis patients from Latin America. The CORES Study. Nephrol Dial Transpl. 2011;26(6):1938–47. [DOI] [PubMed] [Google Scholar]
- 35.Villa-Bellosta R, Rodriguez-Osorio L, Mas S, et al. A decrease in intact parathyroid hormone (iPTH) levels is associated with higher mortality in prevalent hemodialysis patients. PLoS ONE. 2017;12(3):e0173831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Lamina C, Kronenberg F, Stenvinkel P, et al. Association of changes in bone mineral parameters with mortality in haemodialysis patients: insights from the ARO cohort. Nephrol Dial Transpl. 2020;35(3):478–87. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The clinical data used in this study are available from Ping Tung Christian Hospital; however, some restrictions apply to their availability. Access to these data requires approval from the Institutional Review Board of Ping Tung Christian Hospital and cannot be shared publicly. The NHIRD data used in the current study were obtained under license and are not publicly available due to governmental policy restrictions.