Introduction

Metabolic acidosis is common and requires effective management in patients on dialysis [1]. Acetic acid is the most used acid buffer in standard bicarbonate-based dialysates worldwide to prevent calcium and bicarbonate precipitation. However, the use of acetic acid-based bicarbonate dialysate (acetate) can increase serum acetate concentration by up to 5–10 times its physiological concentrations after a hemodialysis session [1–4]. Such acetate accumulation has been associated with the incidence of headaches, nausea, and per-dialysis hypotension [3–7]. Alternatives, such as citric or hydrochloric acid, have been proposed for acetate-free dialysis.

Citric acid-based bicarbonate dialysate (citrate) has been in use in France since 2013. It is safe at low concentrations and provides clinical advantages, including anticoagulant effects and a better biocompatibility compared to acetic acid [8]. Citrate dialysate reduces complement pathway activation, oxidative stress, and free radical formation, while improving mitochondrial glutathione synthesis [9–11]. Benefits also include fewer episodes of intradialytic hypotension, improved dialysis efficiency and nutritional status, better metabolic acidosis control, and decreased propensity for calcification [6,7,12–14]. However, citrate dialysate can lower serum magnesium levels, cause muscle cramps, induce a negative calcium balance, stimulate PTH secretion, and was initially suspected to increase mortality risk, though this has not been confirmed [2,8,15–17].

Hydrochloric acid-based bicarbonate dialysate, which for the sake of brevity will be referred to as hydrochloric dialysate, has been proposed as an alternative to acetate or citrate [3]. A study from the REIN registry conducted in 2015 indicated that acetate-free dialysis with hydrochloric dialysate was associated with improved survival in patients aged 70 years and older, regardless of comorbidities and the use of hemodiafiltration (HDF) [18]. However, the high acidity of concentrate acidified by hydrochloric acid introduces technical challenges in manufacturing, packaging, and use, which increase production costs.

Parathyroid hormone (PTH) is a central regulator of calcium and phosphate homeostasis. In CKD, elevated PTH leads to secondary hyperparathyroidism (SHPT), whereas low PTH results in adynamic bone disease. The type of dialysate used in hemodialysis (HD) can influence serum calcium and phosphate levels, thereby impacting PTH secretion [19]. However, there is no specific guidelines recommending a particular type of dialysate for patients receiving dialysis, likewise, there are uncertainties surrounding the effects of citrate dialysate on mineral and bone metabolism.

Beyond bicarbonate delivery, dialysate buffer composition may influence intradialytic ionized calcium through calcium binding and related shifts in calcium balance, which can modulate CaSR activation on parathyroid cells and thereby affect PTH secretion. Citrate can chelate calcium and may transiently lower ionized calcium, potentially stimulating PTH unless counterbalanced by a higher calcium bath, whereas acetate and hydrochloric acid–based dialysates may have different profiles. Accordingly, we hypothesized that dialysate buffer type could be associated with different longitudinal pre-dialysis PTH trajectories. Because intradialytic ionized calcium was not routinely available in this real-world dataset, we evaluated repeated pre-dialysis PTH measurements over follow-up.

The aim of this study was to examine the association between three types of bicarbonate-based dialysates (acetate, citrate and hydrochloric acid) and longitudinal changes in serum PTH and other mineral and bone biomarkers in routine practice.

Methods

Results

All analyses were conducted on 20 multiply imputed datasets. The total percentage of missing data for each variable is presented in Table S1 in S1 File. Imputed values closely matched the observed marginal distributions and the algorithm converged rapidly, demonstrating fidelity and stability (see S1 File). Key results are presented in Tables 1–4 and Figs 1,2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Baseline main features of the whole study population.

https://doi.org/10.1371/journal.pone.0345776.t001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Pooled LMM results examining the association between dialysate buffer composition and serum PTH levels. Acetate serves as the reference group (i.e., comparisons show changes relative to Acetate). Percentage changes in PTH (with 95% confidence intervals) are shown. Overall Dialysate and Dialysate × Time interaction p‐values come from likelihood ratio tests (LRT); other p‐values derive from Wald tests.

https://doi.org/10.1371/journal.pone.0345776.t002

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Circulating mean bone-mineral biomarkers evolution for the three types of dialysates over time.

https://doi.org/10.1371/journal.pone.0345776.t003

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Annualized rate of adverse events per patient-year by dialysate type.

https://doi.org/10.1371/journal.pone.0345776.t004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Serum PTH levels by dialysate type over 24-month follow-up.

https://doi.org/10.1371/journal.pone.0345776.g001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Estimated PTH trajectories by dialysate for total patients.

https://doi.org/10.1371/journal.pone.0345776.g002

Discussion

The main findings of this study are as follows: First, and for the first time in real-world setting, serum PTH levels significantly decreased over two years in patients exposed to bicarbonate-based hydrochloric acid dialysate, reaching a decline of −10.9% at two years. In contrast, PTH declined only modestly and non-significantly by −2.9% and −2.3% in patients using acetate or citrate dialysates, respectively. Second, patients treated with citrate buffer did not display a rise in PTH but rather showed a small, non-significant downward trend over time, despite the interference of this buffer with divalent cations. Third, the three groups of patients differed markedly in baseline characteristics and treatment patterns, a case-mix that inevitably influenced the observed outcomes. Therefore, our findings should be interpreted as real-world associations between buffer type and PTH trajectories under routine co-prescription of different calcium baths and modalities, rather than as isolating the causal effect of buffer composition alone.

PTH is essentially regulated by circulating ionized calcium concentration, which binds and activates the calcium-sensing receptor, a G protein-coupled receptor present in parathyroid cells, to inhibit PTH production [21]. Acid–base status may also influence the calcium–PTH set-point and potentially modify responsiveness to calcimimetics. Although predialysis bicarbonate was available and described, intradialytic pH/ionized calcium dynamics were not captured, so residual confounding related to acid–base physiology cannot be excluded. Consequently, selecting an appropriate dialysate calcium concentration is crucial for maintaining neutral calcium balance and keeping PTH within the KDIGO-recommended values (2–9 times the upper normal threshold). The 2025 KDIGO controversy highlights persistent uncertainty over the optimal PTH target, as well as the absence of high-quality data informing the optimal calcium-dialysate concentration. Nevertheless, they suggest, based on observational studies, that using a dialysate calcium of 1.25 to 1.50 mmol/L would be the most appropriate concentrations in order to prevent CKD-MBD and vascular calcifications [22]. If the calcium balance becomes negative with these dialysates, this can be countered by prescribing a high-calcium diet, oral calcium salts, calcium-based phosphate binders, and/or active vitamin D analogues. However, these recommendations have never taken into consideration the type of acid component of the buffer used for the dialysate fabrication, which may have a direct impact on the dialysate free calcium concentration and its per-dialytic transfer to the patient. Our results support this assertion by demonstrating a significant dialysate-type × time interaction in the LMM (p = 0.048), supporting that dialysate buffer type is associated with different PTH trajectories over two years after multivariable adjustment.

One corroborating finding of our study is that PTH tended to decrease toward the recommended values in the three groups of patients. Although 25-OH-D increased modestly in the hydrochloric group, we cannot infer a causal or synergistic effect on PTH; the hydrochloric dialysate–associated decline in PTH persisted after adjustment for 25-OH-D and other covariates in the mixed-effects model.

Several mechanisms could underlie the greater decrease of PTH with hydrochloric dialysate. Differences in dialysate calcium prescription may partly contribute [23]; however, associations persisted after adjustment for dialysate calcium, suggesting buffer composition could play a role. The LMM nevertheless confirmed that dialysate-calcium concentration was an independent predictor of PTH: compared with ≤ 1.50 mmol/L, a bath of 1.75 mmol/L lowered PTH by 12.2% (95% CI −15.4% to −8.7%; p < 0.001), whereas 1.65 mmol/L produced a borderline 5.8% reduction (p = 0.074). Alternatively, hydrochloric dialysate could favor greater calcium transfer to patients, resulting in a more positive calcium balance; unfortunately, total calcium balance could not be assessed in this study. Of note, although the pre-dialysis serum total-calcium levels did not differ between the three groups, they accounted for 5.9% of the PTH variation in the hydrochloric group according to the LMM. The second possibility could be better PTH control because the baseline PTH was already within the recommended KDIGO range in 55% of these patients, which is a well-known determinant of PTH evolution [24]. The third possibility is that patients on hydrochloric dialysate were older, and had more comorbidities, including diabetes, conditions known to be associated with low PTH [25].

As citrate is a well-known calcium binder, its use as a dialysate buffer, even at relatively low concentrations (0.8–1.5 mmol/L), has been associated with a negative calcium balance and increased PTH [6,7,12–14]. In our study, we did not observe any significant increase of PTH; on the contrary, PTH decreased slightly by 2.3% after two years of follow-up. The proportion of patients with PTH in the high PTH category (>600 pg/mL) decreased from 33 to 23%, and there was a 13% increase in the proportion of patients with PTH within target at two years. This effect was probably related to the use of a calcium dialysate concentration of 1.65 mmol/L in 72.6% of these patients. We found a higher incidence of hypomagnesaemia (0.54 events patient-year) in patients using citrate compared to hydrochloric acid (0.30) and acetate (0.48), with no difference in the incidence of muscle cramps, probably because of the high dialysate-calcium concentration. Citrate dialysate may offer other clinical advantages, including anticoagulation effects and better biocompatibility compared to acetic acid [8]. It reduces complement pathway activation, oxidative stress, and free-radical formation, while improving mitochondrial glutathione synthesis [9–11]. It is also associated with fewer episodes of intradialytic hypotension, improved dialysis efficiency, better metabolic-acidosis control, decreased propensity for calcification, and improved nutritional status [6,7,12–14]. Altogether, these data suggest that citrate dialysate, which has been in use in France since 2013, has not been associated with increased mortality in prior observational reports [15–17].

KDIGO guidelines recommend a calcium-dialysate of 1.25 or 1.50 mmol/L to avoid positive calcium balance and vascular calcifications [22]. Our data show that 1.50 mmol/L was prescribed in 50.2% of patients, whereas 1.25 mmol/L was used in only 1.1%. A strict restriction to a single calcium bath across all three dialysate types was not feasible due to local prescribing patterns (notably, 1.50 mmol/L was not used in the citrate group, and 1.65 mmol/L was exclusive to citrate), so calcium-bath imbalances may still contribute despite multivariable adjustment. Therefore, we cannot robustly address the effect of a 1.25 mmol/L bath on PTH dynamics. This point is clinically important, because large observational programs (e.g., DOPPS) have reported temporal increases in PTH following implementation of lower dialysate calcium recommendations [26]. The decline in PTH was observed in our study in the group using hydrochloric dialysate at 1.50 mmol/L calcium (77.0% of patients).

The observation of a significantly higher PTH at baseline in patients on the acetate dialysate compared to hydrochloric and citrate dialysates was not surprising. This was likely related to the nephrologists’ background and prescription habits, in the absence of specific internal, national, and international guidelines. Acetate prescription was predominant in out-patients and self-care HD units treating younger patients with fewer comorbidities and less fear of the potential deleterious effects of acetate. Even at the relatively low concentration of acetate in the dialysate (3.0 mmol/L), acetate may still perturb intermediary metabolism and generate several metabolites, including acetoacetate and β-hydroxybutyrate, which might potentially induce several adverse effects such as headaches, nausea, per-dialysis hypotension, and ultimately stimulating interleukin-1 production and inflammation [3,5–7]. Notably, recent experimental studies have suggested that microbiota-dependent butyrate plays an important role regulating the skeletal action of PTH [27].

Several limitations should be acknowledged. First, several covariates had substantial missingness (up to 50%). Although we used multiple imputation by chained equations under a missing-at-random assumption and diagnostics (Figs S1–S2 in S1 File) supported convergence and distributional similarity, bias cannot be excluded if data were missing not at random. Second, this was an observational study and dialysate type was not randomly assigned; residual confounding, including confounding by indication, may persist despite adjustment. Third, a 1.25 mmol/L dialysate calcium bath was rarely prescribed (<1%), precluding assessment of its impact on PTH trajectories. Fourth, the prevalent nature of the cohort may introduce survivor bias and limit generalizability to incident patients. Fifth, we did not assess hard clinical outcomes, and medication exposure was captured as quarterly indicators rather than dosages, preventing evaluation of treatment de-escalation. Finally, intradialytic or post-dialysis ionized calcium measurements and total calcium balance were unavailable; FGF23 was not measured; and post-dialysis bicarbonate was not routinely recorded.

In this real-world cohort, hydrochloric acid–based bicarbonate dialysate was associated with lower PTH levels over 24 months compared with acetate after adjustment for baseline PTH, dialysate calcium, medications and other covariates. Citrate dialysate was not associated with a higher PTH trajectory or higher cramp rates under the calcium-bath strategies used in routine practice. Given the observational design, baseline case-mix differences and the lack of intradialytic ionized calcium/pH measurements, prospective controlled studies are needed to clarify the independent role of buffer composition and optimal buffer–calcium bath combinations for CKD-MBD management.

Supporting information