Introduction

The first case of calciphylaxis was described in 1898 [1]. The term "calciphylaxis" was coined by Hans Selye in a book of his lifetime work published in 1962. He described an animal model in which he was able to induce calcification of tissues with a sensitizer [vitamin D, parathyroid hormone (PTH), diet high in calcium and phosphorus, renal failure], followed by a "critical time" interval before a challenger (iron salts, steroids, trauma, egg albumin). While this extra-skeletal calcification model was generated by many of the clinical and mineral metabolism risk factors identified in dialysis patients with extra-skeletal calcification, the pathological lesions were calcification of the dermis, not of the arterioles [2]. In patients with chronic kidney disease (CKD), calciphylaxis lesions are usually located in two distinct patterns: distal, with lesions on the lower extremities, or proximal, with lesions on the abdomen, inner thighs, and buttocks [3]. Calciphylaxis is characterized by lesions that present as painful red nodules, often with livedo reticularis that progress to ulcers with necrotic centers and violaceous borders. Histologically, there is vascular calcification of small arterioles (usually 40–50 μm) in the medial layer with resultant ischemic epidermolysis. In general, the lesions are not associated with an inflammatory cell infiltrate, and only rarely is there thrombosis and intimal layer changes. The mineral that is deposited is composed of hydroxyapatite (calcium phosphate) [4]. Thus, the calcification is in arterioles, not dermis, in contrast to the calcification model of Selye. As a result of these pathological differences, Coates et al. [4] (and we concur) recommended using the term calcific uremic arteriolopathy (CUA) to describe the clinical disease observed in dialysis patients. Recently, there has been an increase in the number of publications about CUA, but whether this represents an increased incidence or renewed interest is unknown.

Unfortunately, most of the literature on CUA consists of case reports or case series. The differential diagnosis of CUA includes calcinosis cutis (calcification in skin, not vessels, truly Selye's calciphylaxis), panniculits, often associated with pancreatitis, subcutaneous fat necrosis of the newborn, pseudoxanthoma elasticum, scleroderma, and porphyria cutanea tarda [5]. It is important to note that dialysis is not a prerequisite, nor is renal failure, as there have been isolated case reports in patients with a functioning renal allograft [3, 6] and in primary hyperparathyroidism [7]. Several early case series identified elevated parathyroid hormone as a key risk factor, leading to the recommendation that patients with CUA undergo a parathyroidectomy [3, 6, 8]. In particular, Hafner et al. [8] in a large review of the literature reported that 38 of 58 patients that underwent parathyroidectomy survived, whereas only 13 of 37 without parathyroidectomy survived (P<0.01). This same paper reported that the survival was 75% for those patients with the distal form of the disease and only 26% for those with the proximal form of the disease [8]. Another review of the literature noted that hyperparathyroidism was found in 82% of cases compared with hyperphosphatemia in 68%, hypercalcemia in 20%, and elevated calcium×phosphorus product greater than 70 mg2/dl2 in 33% [9]. However, Mawad et al. [10] recently reported 7 cases from 100 dialysis patients; 5 of these 7 patients had biopsy-proven adynamic bone disease, clearly demonstrating that elevated PTH is not required. Several medications have been identified as potential culprits in the etiology of CUA, including Coumadin (sodium warfarin) [11], insulin injections [12], calcium-based phosphate binders [13, 14], and intravenous iron dextran [15].

Recently, there have been four case-control series that have helped further clarify these risk factors. The first, by Bleyer et al. [16] examined 9 cases of pathologically proven proximal CUA over 6 years (excluding 1 case of distal CUA), and compared these patients with their remaining dialysis patient population of 347 patients. We performed a similar case-control study examining 10 cases of pathologically proven CUA from a dermatopathology database over 5 years (both proximal and distal lesions), and compared these patients with our dialysis patient population of 235 patients [17]. Mazhar et al. [18] identified 19 cases of calciphylaxis, 16 of which had pathologically proven disease. One patient had a functioning transplant; the others were undergoing hemodialysis. This study compared the CUA cases with controls (3 per case) from their hemodialysis center, matched for the date of initiation of hemodialysis [18]. Lastly, Fine and Zacharias [19] reported 36 new cases of CUA in 7 years, the majority of which were early, non-ulcerating lesions, noting an increasing incidence rate of 4.5/1,000 patient-years in the years 1998–2001. They compared these "plaque only" cases with 2 controls per case, controlled for dialysis modality, duration of dialysis, and gender [20]. The results of these four case-control studies [16, 17, 18, 19] demonstrate female gender, Caucasian race, and obesity as clear risk factors. In contrast, diabetes was a risk factor in two of the four series, and Coumadin was not significantly associated with CUA, although case-control studies do have limitations. The mortality ranged from 45% to 65%. Analysis of the biochemical abnormalities in these case-control studies demonstrates that hyperphosphatemia and hypoalbuminemia are associated risk factors, whereas hyperparathyroidism and hypercalcemia are not associated with CUA (Fig. 1).

Fig. 1.
figure 1

Role of mineral metabolism in the pathogenesis of calcific uremic arteriolopathy (CUA, calciphylaxis). Data from four case-control studies [16, 17, 18, 20] are presented with results for serum levels of phosphorus (top panel), calcium (middle panel), and parathyroid hormone (lower panel). Serum phosphorus levels were higher in CUA patients compared with control patients in all four studies. In contrast, the serum levels of calcium were lower in CUA than control patients only in the study by Bleyer et al. [16]. Similarly, there was no difference in the intact parathyroid hormone levels in the CUA patients and control patients in any of the studies. (*P<0.05 comparing CUA patients with control patients within a given study)

Full size image

Vascular calcification is not a new phenomenon in patients with CKD. In 1979 Ibels et al. [21] demonstrated that patients undergoing a renal transplant had increased intimal thickness and calcification, and increased medial thickness and calcification, of internal iliac and renal arteries compared with transplant donors and autopsy subjects. Furthermore, the magnitude of calcification quantified by biochemical techniques correlated with duration of dialysis [21]. In 2000, Schwarz et al. [22] demonstrated increased calcification of atherosclerotic plaques in dialysis patients at autopsy compared with non-dialysis patients, but the volume of plaque was not different. In addition, the medial layer was markedly thickened compared with patients not on dialysis. New imaging techniques, such as electron beam computed tomographic scan (EBCT), have demonstrated markedly increased coronary artery calcification in dialysis patients compared with non-uremic patients with angiographically proven coronary artery disease [23]. This process also occurs in children and young adults [24]. In 2002, we demonstrated medial calcification in the inferior epigastric artery of many asymptomatic patients undergoing a renal transplant in association with expression of bone-specific factors [25, 26]. Thus, calcification of both intimal/atherosclerotic lesions and the medial layer of arteries are increased in patients with CKD.

Is CUA simply another manifestation of vascular calcification?

There are case reports of CUA, defined as medial calcification of small arterioles, leading to gastrointestinal hemorrhage [4, 27, 28] and penile gangrene [29]. Pathologically, the arteries involved in these unusual clinical sequelae of CUA have medial calcification indistinguishable from skin lesions of CUA and from the medial calcification we observed in the inferior epigastric artery [25]. Furthermore, this medial calcification is also pathologically the same as that in distal peripheral arteries, the so called Mönckeberg calcification seen in individuals with diabetes, which is associated with increased cardiovascular mortality [30]. Even calcification of small arteries in the breasts of individuals without renal failure correlates with cardiovascular mortality [31, 32]. Clearly, calcification of arteries, whether arterioles of the skin, small arteries in the abdomen or breast, medium arteries in the peripheral extremities, the aorta, or coronary arteries is pathological, and is a marker for disease. The precise clinical manifestation may be due to the location of the abnormal artery and the tissue it perfuses. In addition, there is likely some sort of systemic process or disease that leads some individuals to calcify their arteries and others not. However, in CUA the lesions appear more rapidly progressive and may cause greater mortality than in many other forms of vascular calcification. This may be because small arterioles are the final point of blood supply and therefore collateral blood flow is not an alternative, or there is a systemic process that rapidly depletes the body of naturally occurring inhibitors of mineralization.

Although arterial calcification has been recognized for more than 150 years [33], the basic pathogenesis of this disorder is still not completely understood. The work of Stary et al. [34] in mammals and in human autopsy studies demonstrated that calcification in atherosclerotic (intimal) disease appears to be a late process, occurring well after the intimal plaque. However, in medial calcification, the calcification is the first obvious sign of disease, although medial wall thickening may precede the mineral deposition. So how can intimal/atherosclerotic calcification and medial calcification processes be related? Until recently, the actual calcification of the vascular tissue had been regarded as a passive process associated with atherosclerosis or normal aging, resulting from a non-specific response to tissue injury or necrosis. Now there is strong evidence that vascular calcification is an active and regulated process that shares features with normal embryonic bone formation and bone repair [17, 25, 35, 36, 37, 38, 39, 40, 41, 42]. Bones form through endochondral ossification with a cartilage intermediate, or by direct intra-membranous ossification from condensations of mesenchymal cells. Osteoblasts differentiate from a pluripotent mesenchymal stem cell to a mature osteoblast with different morphology, and gene and matrix protein production in each phase of differentiation, bone formation, and mineralization. The latter is felt to occur through blebs from osteoblasts called matrix vesicles that contain hydroxyapatite, which are guided to areas by collagen and non-collagenous matrix proteins [43].

These same mesenchymal stem cells form vascular smooth muscle cells (VSMC). Work in the last decade has demonstrated that VSMC and/or vascular pericytes are capable of producing bone-like proteins in cell culture [44, 45]. Additional experiments in both human and bovine VSMC have demonstrated that phosphorus, in the form of β-glycerophosphate (which is cleaved by alkaline phosphatase to form free phosphate), can induce calcification similar to that observed in osteoblast cultures [46, 47]. To examine the pathophysiology of vascular calcification observed in dialysis patients, we examined arteries histologically [17, 25]. We found expression of osteopontin at the base of the calcium spicules in skin arterioles with calcification from patients with CUA, but no expression of osteopontin in arterioles without calcification in the same section [17]. We also found evidence of other bone proteins, including osteonectin and bone sialoprotein (unpublished observation), and evidence for matrix vesicles in CUA specimens [17]. We then prospectively evaluated inferior epigastric arteries from patients with CKD stage 5 undergoing renal transplantation [25]. Many of these arteries demonstrated calcification in the form of medial calcinosis in association with the expression of the "bone" matrix proteins osteopontin, alkaline phosphatase, type I collagen, and bone sialoprotein. Positive immunostaining for these bone proteins was found more frequently than was overt calcification, which suggests that the deposition of these proteins precedes calcification [25]. These findings suggest that the initial changes that occur in the vessels of dialysis patients are the deposition of these bone matrix proteins into the medial layer, followed by calcification. These results confirm a cell-mediated process in vascular calcification in patients with end-stage renal disease (ESRD). This is similar to findings in vessels of non-dialysis patients with both calcified atherosclerotic coronary arteries [35, 36, 48, 49, 50] and medial calcinosis in small distal vessels [38]. The presence of bone proteins in all forms of vascular calcification was also recently confirmed by Canfield et al. [51] who found osteopontin and cartilage oligomeric matrix protein in arteries from patients with CUA, atherosclerosis, and chronic vascular rejection in renal allografts, but no evidence for staining in normal arteries.

What makes vascular calcification so prominent in dialysis patients?

Dialysis patients are known to have many vascular risk factors, such as a history of cardiac disease, diabetes, elevated levels of homocysteine and oxidized lipids, and hypertension. In addition to these traditional vascular risk factors, there is increasing evidence that elevated serum phosphorus, serum calcium×phosphorus product, and/or calcium load in the form of calcium-containing phosphate binders are associated with various vascular end-points including CUA [17], coronary artery calcification by EBCT [24], carotid and aortic calcification [40, 52], and hemodynamic abnormalities [53]. These studies suggest a relationship between these laboratory values and positive calcium and phosphorus balance with vascular disease. Chertow et al. [54] recently demonstrated in a prospective randomized trial that the non-calcium phosphate binder sevelamer can arrest coronary artery and aorta vascular calcification in ESRD patients, whereas calcium-based phosphate binders increased calcification in both coronary arteries and aorta. We and others have also demonstrated a healing of CUA lesions in patients with removal of calcium-containing phosphate binders [14, 55]. Jono et al. [56] demonstrated that phosphorus-induced calcification was dependent on the sodium-phosphate (Na/Pi) co-transporter. Furthermore, exogenous phosphate added to human VSMC culture upregulated core-binding factor α-1 (Cbfa1) expression [56], a transcription factor critical for osteoblast differentiation and the expression of the bone matrix proteins osteopontin, osteocalcin, and type I collagen [57]. Cbfa1 knock-out mice fail to form mineralized bone, proving that Cbfa1 is critical for the initial differentiation of osteoblasts [58], thus indicating that VSMC de-differentiate to osteoblast-like cells. In addition, arteries from the matrix gla protein knock-out mice lose smooth muscle markers and gain expression of Cbfa1 as they progressively mineralize their arteries [59]. Thus, the in vivo data confirm a role for abnormal mineral metabolism in vascular calcification in ESRD patients, and in vitro data in VSMC indicate that phosphorus can lead to calcification, and that phosphorus can induce Cbfa1 and the expression of bone matrix proteins.

In order to further understand the mechanism by which kidney disease or the uremic state can induce vascular calcification, we incubated bovine VSMC in the presence of pooled normal human serum versus pooled human serum from patients on hemodialysis for at least 2 years (to eliminate residual renal function) [60]. Using these pooled sera in vitro, we demonstrated that uremic serum led to increased and accelerated calcification of VSMC. Furthermore, the uremic serum upregulated the expression of osteopontin in VSMC compared with normal serum. Unlike the exogenous addition of phosphorus in the form of β-glycerophosphate, the uremic serum-induced osteopontin was only partially dependent on both alkaline phosphatase and Na/Pi co-transport [60]. Of importance, the final media concentration of phosphorus was similar in the VSMC cultures with 10% normal and those with 10% uremic serum (~0.5 mM) [60], well below levels known to induce calcification in the study of Jono et al. (2.0 mM) [56]. We also found that uremic serum, compared with control human serum, induced the expression of Cbfa1 in bovine VSMC in a time-dependent, non-phosphorus-mediated mechanism. Furthermore, we have found ex vivo evidence of the expression of Cbfa1 in both medial and intimal VSMC in calcified inferior epigastric arteries obtained at the time of kidney transplant [26]. These results suggest that Cbfa1 may be a key regulatory factor in vascular calcification observed in dialysis patients. However, expression of Cbfa1 was also simultaneously observed in calcification of atherosclerotic plaques from patients without CKD [61]. These results suggest that multiple factors may lead to Cbfa1 expression, and that this expression is associated with calcification. Whether Cbfa1 is a marker of VSMC differentiation or a cause remains to be determined.

Thus, we hypothesize that vascular calcification in the course of dialysis may be a four-step process (Table 1). First, VSMC are stimulated by injury or uremic toxins [26], including phosphorus [56], oxidized low-density lipoproteins [62], and perhaps ischemic injury, to transform into osteoblast-like cells via upregulation of Cbfa1. In CUA, this injury may be ischemic due to altered blood flow in the pannus of obese women or injection sites. In intimal disease this injury may represent the diverse factors that lead to plaque formation. These cells then lay down a bone matrix of type I collagen and non-collagenous proteins. Some of the cells may also undergo apoptosis [63]. In step 2, this framework of matrix proteins with and without apoptotic bodies may act as a nidus for mineralization, similar to the recent observations in kidney stone formers [64]. In step 3, extension of the mineralization of this nidus/matrix occurs, in part through a physiochemical process due to supersaturation of serum and in part through a process "guided" by the matrix proteins and osteoblast-like cells. This step is likely to be accelerated in the presence of elevated calcium×phosphorus product in the serum, as well as positive calcium balance due to calcium-containing phosphate binders [24, 40, 54]. In addition, a lack of naturally occurring inhibitors likely plays a role. This includes decreased matrix gla protein, which is decreased in animals on Coumadin [65, 66], and fetuin-A, which is decreased in ESRD patients with acute inflammation, and in patients with CUA [67, 68]. Thus, the mineralization step is a balance of the pro-mineralization and anti-mineralization forces at the cellular level. The fourth stage is the resultant ischemia of the tissue perfused by the calcified artery. In the case of CUA, the damage is skin necrosis due to calcification of the skin arterioles. In the case of coronary artery disease, the damage is an acute myocardial infarction with occlusion of the coronary artery or sudden death when the calcified distal vessels fail to supply extra oxygen needed during stress. We believe these are all manifestations of vascular calcification, a continuum of extra-skeletal osteogenesis exacerbated by injury and aggravated by the uremic milieu.

Table 1. Proposed pathogenesis of vascular calcification in chronic kidney disease (VSMC vascular smooth muscle cells, Cbfa1 core-binding factor α-1, LDL low-density lipoprotein, CUA calcific uremic arteriolopathy)
Full size table

Unfortunately, definitive therapies for these manifestations of vascular calcification, including CUA, are lacking. At present, prevention appears to be the only hope, as no studies have demonstrated an ability to modulate, or remove, the mineral once it is deposited. Unfortunately, therapy at present is limited (Table 1), but should include (1) reducing traditional and non-traditional vascular disease risk factors, (2) minimizing mineral overload by aggressive control of serum phosphate with non-calcium binders or increased dialysis frequency and decreasing vitamin D use, and (3) aggressive wound care. In addition, optimizing naturally occurring inhibitors such as fetuin-A by decreasing inflammation and bisphosphonate therapy (at least in animal models [69]), and stopping Coumadin therapy to increase matrix gla protein remain theoretical at present. Nevertheless, while we strive to understand the pathophysiology of this complicated process, we owe it to our patients to minimize cardiovascular risk factors that may lead to initial injury and avoid positive calcium and phosphorus balance that add further insult to injury.