Bone Dysregulation in Acute Kidney Injury

Acute kidney injury (AKI) is a highly prevalent condition with dire acute and chronic consequences [1]. AKI is conventionally defined by an increase in serum creatinine ≥0.3 mg/dL within 48 h or ≥1.5 times from baseline within 7 days and/or a decrease in urine output [2]. The high acute mortality is staggering from multiple complications from AKI as well as from the underlying conditions that predispose to AKI. Survivors of AKI are at risk of AKI-to-chronic kidney disease (CKD) transition, which imposes a colossal burden on overall wellness and health systems [3]. Regardless of whether AKI survivors succumb to CKD, one should be cognizant of long-term complications of AKI. An important question is whether there are bone sequelae from AKI that are independent of CKD (Fig. 1). By that, one is referring to bone disease prior to the onset, or in the absence of CKD – a form of “post-AKI osteopathy.” It is therefore relevant to determine if AKI turns out to be an independent risk factor for bone disease independent of CKD, and how post-AKI bone disease affects kidney and overall health. This review summarizes a fraction of the existing data to provide some guidance in future research efforts.

There is a host of metabolic and hormonal changes during and post-AKI. The requisites for inclusion in this review are (1) the parameter is altered in AKI and post-AKI at least in rodents, and preferably in humans and (2) the disturbance has the potential to affect bone health. Given the limited data in AKI and bone health, this review does not elaborate on the complex interplay and feedback loops among the various highlighted endocrine-metabolic parameters.

Klotho was discovered serendipitously when the accidental disruption of its genetic locus led to premature multi-organ failure [4]. Klotho is a cell maintenance and cytoprotective protein that is released from the kidney to the circulation to serve distant organs [5]. Klotho levels in blood, urine, and the kidney all drop precipitously in many forms of experimental ischemic, nephrotoxic, septic, and obstructive AKI indicating the universal nature of Klotho deficiency in AKI regardless of the underlying etiology [6]. Low urinary Klotho has also been described in human AKI [7, 8]. The bone lesion in chronic Klotho deficiency is well documented although the findings are not uniform. Low bone turnover, independent impairment of osteoblasts and osteoclast, patchy mineralization defect, and premature mineralization of cartilage have been described in Klotho-deleted mice or Klotho-hypomorphs [9, 10], whereas osteocyte-specific Klotho deletion showed increased osteoblast and increased bone volume [11]. It is conceivable that even a transient state of Klotho deficiency at the critical window around AKI can have long-term impact as in the case of AKI-to-CKD progression [12].

Klotho has many potential clinical utilities [13]. Further work will determine whether it is a predictor of bone complications post-AKI, and most importantly a therapeutic target (i.e., repletion of Klotho) to prevent post-AKI bone complications.

Fibroblast growth factor-23 (FGF23) is a bone-derived hormone belonging to a 22-member FGF family, which along with FGF19 and FGF21 constitutes the endocrine subfamily of FGF’s [5, 14]. FGF23 originates mainly from bone. FGF23 has a wide spectrum of functions including phosphate, calcium, and iron homeostasis, as well as modulation of inflammation and immune functions [15]. Cumulative data in humans and rodents indicate that FGF23 increases rapidly (<1 h) and massively (>15-fold) after onset of AKI [16]. One murine study using bilateral ischemia-reperfusion injury showed increase in kidney tissue and circulating glycerol-3-phosphate levels, which stimulated bone FGF23 production via local lysophosphatidic acid synthesis and binding to lysophosphatidic acid receptor 1. Glycerol-3-phosphate levels were also elevated in human AKI, highlighting a potential target to modulate bone production of FGF23 during AKI [17]. The effect of FGF23 on bone is well described in the congenital and acquired conditions of chronic high FGF23 consisting of rickets, osteomalacia, and dental abscesses which can be largely but not exclusively accounted for by renal phosphate wasting [18]. In contrast, the extremely high FGF23 in CKD is associated with a myriad of bone lesions not typical of osteomalacia that result from the complex uremic syndrome [19]. To date, the bone lesion in CKD has yet to be proven causally related to the high FGF23. The unanswered question is whether the transient albeit extremely high FGF23 in AKI can lead to lasting post-AKI bone defects.

Some human studies showed acute elevation of parathyroid hormone (PTH) with AKI that subsequently returned to normal with resolution of AKI [20‒22]; while others did not find statistically significant differences in AKI [23, 24]. There is likely a highly variable and only transient increase in PTH in human AKI. The classic work of Massry implicated peripheral PTH resistance [25] but change in parathyroid gland set-point due to acute inflammation has also be postulated in AKI [26]. The skeletal complications of chronic PTH elevation are well described as osteitis fibrosa cystica [27], but there is no data describing transient PTH elevation having short-term or long-term effects on bone health in animals or humans.

Alterations of vitamin D metabolism contribute to elevation of PTH, inflammation, cardiovascular disease, and mineral bone disease in CKD, but its impact on AKI is less well established [28]. While observational studies have found consistent lower 1,25 (OH)₂ vitamin D and less consistent lower 25 (OH) vitamin D in patients with AKI versus without AKI in acute settings, the relationship between these measurements and relevant clinical outcomes such as AKI progression, need of dialysis or death has not been confirmed [26, 29]. While one should still acknowledge the possibility that certain patients with AKI or CKD could benefit from vitamin D supplementation, identification of more specific clinical sub-phenotypes and mechanisms of action is needed to assist in the design of future clinical trials. This is highlighted by several negative clinical trials evaluating vitamin D supplementation in the general population, including persons with CKD [30].

Adiponectin is an important circulating hormone, produced mainly by adipocytes. Adiponectin regulates glucose and lipid metabolism and protects from atherosclerosis [31] but also has complex effects on mineral metabolism [32]. Overexpression of adiponectin protects against podocyte injury and albuminuria in mice, whereas global adiponectin deletion promotes podocyte injury and renal fibrosis [33‒35]. However, human studies in CKD have shown that higher rather than lower adiponectin levels associate with increased mortality and progressive CKD [33, 36]. Similarly, urinary adiponectin excretion has been shown to increase in patients with diabetic nephropathy, although the mechanisms of these changes are not fully known [37]. The data on AKI are less available than CKD but one study showed that adiponectin-deleted mice have lower serum creatinine and less histologic damage in response to ischemia-reperfusion injury [38]. While the preclinical data are highly intriguing, it remains to be determined whether acute changes in adiponectin in AKI have long-lasting bone effects.

To date, there are no preclinical data examining post-AKI bone health. However, there are some studies in experimental sepsis most likely with some degree of AKI, but there is little information on the subset of animals who suffered AKI in these sepsis models. One study in rats randomized to cecal ligation and puncture versus sham surgery showed that femoral trabecular bone strength was functionally reduced as early as 24 h after sepsis and was associated with a reduction in collagen and mineral elastic modulus determined by multifrequency scanning probe microscopy [39]. There were no bone architectural or bone mineral density differences by microCT or histologic bone changes between the groups [39]. Another cecal ligation and puncture study revealed ablation of osteoblasts and consequent reduction of common lymphoid progenitors in septic mice, while the pharmacological activation of osteoblasts improved sepsis-induced lymphopenia [40]. This study highlighted the relationship between sepsis-induced inflammation and the immune and bone systems, which underpin a novel bone-centric therapeutic target in sepsis that warrants further investigation. In murine models of arthritis, resolution of inflammation has also been shown to induce changes in the expression of regulators of the Wnt/β-catenin pathway, which is critical for osteoblast differentiation and function [41].

There is paucity of data examining bone health in humans that survived sepsis or AKI. One single-center retrospective cohort study in Japan evaluated bone mineral density using abdominal CT in 52 adult patients admitted with sepsis by Sepsis-3 criteria in whom CT was done for routine care on admission and during follow-up (median of 9 days). Patients were classified based on the presence or absence of osteoporosis (average vertebral body Hounsfield units [HU] <110) on admission. The study showed statistically significant decrements in bone mineral density from admission to follow-up, particularly in the non-osteoporosis group. Only 2 patients in the study had bone fracture; both in the osteoporosis group [42]. Limitations of this study include the small sample size and the selection/indication bias of the study design given the non-protocolized assessment of bone mineral density using CT done for routine clinical care.

A large retrospective study using data from Taiwan’s National Health Insurance Research Database evaluated 13,178 patients diagnosed with sepsis (2000–2012) and a propensity score-matched cohort of 13,178 patients without sepsis for the occurrence of osteoporosis during follow-up until 2013. The overall incidence of osteoporosis was 10.2 and 10.7 (per 1,000 person-years) in the sepsis and non-sepsis groups, respectively. Sepsis significantly increased the risk of osteoporosis (adjusted HR 1.17, 95% CI: 1.04–1.31), particularly in adults younger than 65 years [43]. Limitations of the study are the residual confounding effects of unmeasured parameters such as medication exposures, other risk factors for osteoporosis, and acute illness characteristics during the hospitalization with sepsis, as well as the relative short study follow-up.

Thus far, the only human study evaluating bone health specifically post-AKI also comes from the Taiwan’s National Health Insurance Research Database [44] (Fig. 2). Adult survivors of dialysis-requiring AKI (AKI-D) who were alive and no longer requiring dialysis by 90 days post-discharge (n = 448, admitted between 2000 and 2008) were compared to 1,792 hospitalized patients without AKI, dialysis, or bone fracture history for the risk of bone fractures during follow-up. The interaction between AKI and post-AKI end-stage kidney disease with the bone fracture outcome was addressed in propensity score-adjusted Cox hazard models. The incidence of bone fracture was 320 per 10,000 person-years in the AKI group and 93 per 10,000 person‐years in the non‐AKI group. In adjusted models, the hazard ratio for developing bone fracture in the AKI group, relative to the non-AKI group, was 6.59 (95% CI: 2.45–17.73). The occurrence of bone fractures post-AKI increased the risk of mortality (adjusted HR 1.43, 95% CI: 1.19–1.71) [44]. While this study highlights important observations, one should recognize some key limitations such as the absence of data on bone mineral density and other intermediate phenotypes, residual confounding effects of unmeasured parameters such as medication exposures, other risk factors for bone fracture such as body mass index, and acute illness characteristics during the hospitalization with AKI-D, including the duration of exposure to dialysis. Finally, this study included only patients with AKI-D, which hinders reproducibility to patients with milder forms of AKI and did not include younger patients.

Bone complications from AKI have not received much attention by practitioners and researchers. It is quite understandable that bone complications during the acute phase of AKI are considered low priority and are very far from the minds of the physicians caring for patients in the hospital. Given that AKI is a heterogeneous condition with overlapping etiologies, the identification of specific risk factors for acute bone involvement during AKI is challenging. Further, the lack of validated and recorded short-term biomarkers of bone injury that associate with subacute or chronic bone and cardiovascular health precludes the identification of high-risk clinical phenotypes. In addition, post-AKI care is also very heterogeneous and the routine examination of bone health during post-AKI outpatient care is seldom or never performed. To date, there is just one retrospective analysis by Wang et al. [44] demonstrating unequivocally the increased risk of bone fracture post-AKI. Given the limited database and faced with a potentially morbid condition, how investigators should plan future work evaluating bone health in AKI? The goal is to conclude with a high level of evidence whether the condition of post-AKI osteopathy independent of CKD exists, or not. Either outcome will be informative and will change clinical practice. Enabling this conclusion requires the gathering of evidence on multiple levels.

An instant approach is to mine existing data. The drawback with this retrospective method is that data on bone health-related parameters is either extremely sparse or outright unavailable in AKI databases. The interrogation of bone parameters in biospecimens or imaging has not been part of the regimens due to lack of interest in or cognizance of bone involvement in AKI. Undoubtedly, the best and most conclusive study will be a prospective longitudinal observational examination of subjects with and without AKI who have long-term survival after the acute illness. Novel methods of clinical AKI sub-phenotyping and biological endo-phenotyping could be applied, as well as risk-classification tools using electronic health records, to enrich the study population according to the study hypothesis [45]. A suggested design is shown in Figure 3. Utilizing academic or community-based large nephrology practices with post-AKI follow-up established protocols will be a good starting substrate.

Securing animal data on AKI and bone is highly informative for several reasons. (1) Controlled AKI with different and specific etiologies and varying severity can be imposed with precise timing. (2) Prospective longitudinal studies can be performed with contracted time spans. (3) Tissue sampling is possible at various stages. (4) Individual parameters can be manipulated to test association versus causality. (5) Molecular and cellular pathophysiology can be dissected especially when coupled with in vitro cultured cells. (6) Genetically altered strains, biologic or small molecule interventions can be applied to test pathogenesis and open pipelines for future therapy. (7) Bone can be harvested and banked from the large number of studies performed everyday on many aspects of AKI, albeit bones are seldom saved. (8) Once a pathophysiologic intermediate is unambiguously defined in animals, translation to human AKI can be much more focused and specific.

One retrospective analysis showed increased risk of fracture post-AKI independent of CKD development in humans. While there are theoretical reasons for late disturbances of bone health after AKI, no definitive data are available to date. While preclinical studies examining bone health after acute stressors have focused mostly on sepsis models, multiple experimental AKI models are readily available for longitudinal bone health interrogation. Collectively, future research should be tailored to define whether AKI is a risk factor, independent of CKD, for bone disease and if present, the time course and type of bone disease, and finally risk-modification through novel therapeutic interventions.

J.A.N. is supported by grants from NIH (R01DK128208, R01DK133539, U01DK12998, and P30 DK079337). O.W.M. is supported by NIH (P30 DK-079328, T32-DK007257, R01DK081423, R01 DK115703, R01 DK091392, and R01 DK092461) and the Pak Foundation.

Not applicable.

The authors have no conflicts of interest to disclose.

None.

Javier A. Neyra and Orson W. Moe contributed equally to the conceptualization and writing of this article.

This work was conducted at each of the affiliated institutions.

Not applicable.