Bone Alkaline Phosphatase and Mortality in Dialysis Patients

Discussion

The major finding of our study is that BAP is a strong risk factor for all-cause, cardiovascular, and noncardiovascular mortality in dialysis patients, particularly in the short term. Patients in the highest tertile of BAP levels had a markedly six-fold increased risk of death within 6 months compared with patients in the lowest tertile. The effects applied to both cardiovascular (HR, 5.0; 95% CI, 1.3 to 19.0) and noncardiovascular (HR, 6.2; 95% CI 1.3 to 29.2 mortality). Furthermore, high levels of BAP were associated with increased risks of cardiovascular and, by trend, with all-cause death also in the longer term. Compared with BAP, the effects of total AP were less pronounced. Patients in the highest tertile of AP levels had a two-fold increased risk of death within 6 months compared with patients in the lowest tertile. The associations were independent of common known risk factors because we adjusted for age, gender, dialysis modality, primary kidney disease, diabetes mellitus, cardiovascular disease, Khan comorbidity index, body mass index, systolic BP, levels of albumin, calcium, phosphate, PTH, and 25(OH)D.

These findings are in line with previous study results from Regidor et al. (8) examining the effects of total AP on mortality in a large American cohort of dialysis patients (DaVita study). The authors showed an HR of 1.25 (95% CI 1.21 to 1.29) for mortality within 3 years of follow-up in patients with AP levels >120 U/L. Although our study comprised fewer patients and did not examine repetitive laboratory measurements, our results expand the previous findings in several important aspects. We studied a well-characterized cohort of Caucasian patients, which also included a significant proportion of peritoneal dialysis patients. Importantly, the baseline data came from a very homogenous starting point regarding dialysis vintage (all patients 12 months after initiation of dialysis). In contrast, patients of the DaVita study showed a wide range of dialysis vintage (8). This might be of special importance, because our data indicate a striking short-term, versus no or only a moderate long-term, effect of AP and BAP on mortality.

In the outcome analyses, 25(OH)D was furthermore considered as potential confounder. It has been shown to impact on mortality in dialysis patients (15–17) and, moreover, might directly affect BAP or AP levels (18) in cases of renal bone disease with mineralization defects or overt osteomalacia. The most important extension of our study, however, is that the bone isoenzyme BAP was studied in parallel with total AP.

Low levels of both AP and BAP were associated with an improved survival, warranting potential explanations and mechanisms. An important question arises as to whether this association may allow the conclusion that low bone turnover is associated with better outcome or whether low BAP levels clearly reflect low bone turnover. The role of BAP as a biomarker of renal osteodystrophy has previously been investigated in several bone biopsy studies in patients with ESRD (altogether >400 patients) (10,11,19–22), but results are variable. Most patients underwent bone biopsy after tetracycline labeling, allowing the measurement of dynamic bone parameters such as bone formation rate. Bervoets et al. (10) and Urena et al. (11) found that BAP was a better predictor for bone metabolism than total AP. Depending on the laboratory methods used, applied cut-off levels, and the cohorts investigated, low BAP levels showed positive predictive values for adynamic bone disease between 0.89 to 1.0 (10,11,21). However, Youden indices for the diagnosis of renal osteodystrophy vary between 0.49 and 0.93 (10,19–21). Our own data with receiver operating characteristic analyses in 57 patients with advanced chronic kidney disease/ESRD indicated that the best BAP cut-off level for discriminating low turnover from high turnover bone disease was 15.8 U/L. However, sensitivity (0.75) and specificity (0.59) were moderate (V. Brandenburg and G. Lehmann, unpublished data). In summary, BAP at a single time point provides a hint, but not proof, of low turnover. The exact role of BAP in the assessment of bone turnover is currently addressed by an ongoing large bone biopsy trial performed by the Kidney Disease: Improving Global Outcomes (KDIGO) initiative.

Considering potential negative effects that oversuppression of bone metabolism has on bone and vascular health (23), the finding of low BAP levels being associated with improved outcome appears surprising. Potential explanations need to be acknowledged. First, low BAP levels as assessed at one single time point after initiation of dialysis may indicate that serious attention was paid to the development and correction of secondary hyperparathyroidism during predialysis care by the attending physicians. Low BAP may thus additionally represent a surrogate parameter of the general quality of care before and during ESRD. In particular, low BAP levels may partially reflect the amount of active vitamin D administered, because the application of active vitamin D or vitamin D receptor activators can lower BAP by more than one third within several weeks of treatment and uncouple serum BAP levels from bone turnover (24,25). In this context, low BAP as a potential sign for good vitamin supplementation also had a beneficial impact on noncardiovascular mortality. Our strong association between low BAP levels and survival may potentially differ when long-term dialysis patients develop oversuppression of bone turnover. Unfortunately our data do not allow us to look into BAP developments over time that might reflect therapeutic interventions and changes in bone turnover. Future studies are needed to address this topic in further detail, investigating whether BAP reflects more than bone turnover status. Such an association of BAP with conditions other than bone disease is supported by further hints. First, we found a striking association between high BAP and noncardiovascular short-term mortality. Second, our combined BAP + PTH survival analyses indicated that low BAP levels may help overcome the overall reduced survival associated with low PTH levels as currently seen with the Analysing Data Recognising Excellence Optimising Outcomes in CKD (ARO) study (26). Interestingly, the combined analyses showed that, compared with the group with low BAP and low PTH (reference), the group of patients with high BAP and low PTH had an adjusted 2.8-fold increased risk of death within 6 months, which was highly significant (Table 4). To further identify these patients, we evaluated the clinical characteristics and found that this patient group may have the poorest nutrition. Compared with the other patient groups, the individuals with high BAP and low PTH showed lower levels of phosphate, creatinine, total cholesterol, and albumin, as well as body mass index, potentially representative of wasting or malnourishment. Furthermore, residual GFR was lower and the percentage of diabetes higher in this group, potentially additionally supportive of the notion that wasting may be present in these patients (data not shown).

Further explanations of the protective effects of low BAP and AP levels include the mechanisms via pyrophosphate. Experimental data link high APs to the development of coronary artery calcification (27) with their ability to hydrolyze inorganic pyrophosphate (28). Pyrophosphate is a potent inhibitor of vascular calcification, and its biologic action is reduced by phosphatases.

In the context of BAP relating to clinical outcomes, our study impressively showed that high levels of BAP were associated with mortality and specific fatal events, particularly in the short term. Considering that BAP levels were determined only once in single measurements, potential changes in BAP status over time may contribute to explain the time differentiating effects. In comparison with total AP, the effect sizes related to clinical outcomes were much higher for BAP. Apart from the pathophysiologic characteristics differentiating both enzymes, BAP may furthermore provide additional value for the risk stratification of dialysis patients. In this context, measuring BAP may help identify people at risk for all-cause and cardiovascular mortality. In additional analyses, BAP increased the C-statistic of the full model from 0.75 to 0.76 in long-term analyses and from 0.78 to 0.79 in 1-year mortality analyses. C-statistics of short-term models were not possible, however, because of the low number of events and may be promising targets for further studies predominantly designed for prediction modeling. An intriguing perspective emerges from the fact that high BAP levels were also associated with a high mortality in patients with low PTH. This makes us hypothesize about the role of BAP in identifying those patients in whom vitamin D therapy with consecutive lowering of BAP might decrease mortality despite low PTH levels.

Potential limitations need to be acknowledged. Our data indicate a strong association for both AP and BAP with all-cause and cardiovascular mortality in stable ESRD patients; however, we cannot extrapolate our findings toward ESRD patients starting on dialysis. Furthermore, the PTH measurements were performed using various first-generation immunometric PTH assays, dependent on the different participating centers. We are aware that these methods may yield different results, and we cannot report about potential changes in assay use over time. Another limitation of our study is the missing data on active vitamin D treatment. Finally, potential influences by prolonged storage on BAP measurements cannot completely be ruled out. We investigated and compared the median levels of BAP in older and younger samples, which were not meaningfully different. Therefore, potential influences caused by storage time are likely to be small, and the samples used for BAP measurements did not undergo repetitive freeze-thaw cycles.

In summary, high levels of BAP were strongly associated with short-term mortality in dialysis patients, suggesting BAP as an important biomarker of bone and mineral disorder in dialysis patients. Longitudinal assessments of BAP may be useful for the treatment monitoring in clinical practice in patients with ESRD.