The Challenge of Sudden Death in Dialysis Patients

Case Presentation

A 60-yr-old patient with ESRD (82 kg/168 cm), hypertension for the previous 18 yr, and type 2 diabetes is found dead in bed by his wife at home on a Saturday morning. The patient's history is notable for having started dialysis 34 weeks previously. His last dialysis had been the previous day (on a Friday afternoon). Predialysis potassium had been 5.7 mmol/L, the dialysate K⁺ was 2.0 mmol/L, and Mg²⁺ was 0.5 mmol/L. Autopsy was not performed.

Before his morbid event, the patient had been doing well on dialysis. He had less than 6 episodes of hypotension (<100 mmHg) per month. His average predialytic weight gain was 4 kg, and, with fluctuations, his average predialysis BP was 155/65 mmHg. While continuing on gliquidone (an oral sulfonylurea hypoglycemic drug that does not accumulate in renal failure), he had no episodes of hypoglycemia, and his last glycosylated hemoglobin level was 7.2%. His medical history is of hypertension, type 2 diabetes, and left ventricular hypertrophy (LVH) with an ejection fraction of 55%. His electrocardiogram showed signs of LVH and flat T waves. He had no history of hypoglycemic episodes and no evidence of retinopathy. He reported no episodes of arrhythmia or precordial pain. A predialysis chest x-ray 2 wk before the terminal event had shown pulmonary congestion. His medications included 0.25 μg/d calcitriol, calcium carbonate, 10 mg of folate, 80 mg of verapamil, and no β blockers. In addition, he received varying doses of darbepoietin and iron gluconate as required. He did not receive a statin.

Dr. Eberhard Ritz: This case, unfortunately, is the usual presentation of patients on dialysis who succumb to sudden death: There are few, if any, specific premonitory symptoms or signs, cardiac standstill is usually the presenting symptom, and autopsy is not performed, so more specific information on underlying cardiac pathology, if any, is not available. The lack of pathoanatomic confirmation of the underlying pathology or pathologies–as in this case—continues to remain a problem that in the future should come onto the radar screen of nephrologists.

Nevertheless, this case presentation raises a number of unresolved issues and provides an occasion to discuss how the currently unsatisfactory results might be improved in the future. In this context, it is of interest to have a look at the recent 4D study. I have asked Dr. Christoph Wanner to discuss the issue of sudden death in the 4D study.

Dr. Christoph Wanner: One of the unexpected findings of the 4 D study was the failure of atorvastatin to cause a significant decrease of an adjudicated composite cardiovascular end point and specifically of sudden death in type 2 diabetic patients on maintenance hemodialysis (1). The latter finding is pertinent to the present case presentation but may even have some more general significance for understanding sudden death of nondiabetic dialysis patients as well.

The 4D study was a prospective, randomized, controlled study of dialyzed type 2 diabetic patients in 178 German dialysis centers with a follow-up of 3.96 yr. LDL cholesterol was lowered by 10 or 20 mg of atorvastatin by 42% compared with 1.3% on placebo. The primary composite end point (comprising death from cardiac causes, fatal stroke, nonfatal myocardial infarction, or nonfatal stroke) was reduced by 8% (95% confidence interval [CI] 0.77 to 1.10), which was not statistically significant (P = 0.37). These results have recently been confirmed in a small Swedish study where atorvastatin had no significant impact on cardiovascular outcome in dialysis patients (2).

The main post hoc explanation for the negative outcome of the 4D study is the fact that adjudicated coronary death accounted only for 9% of deaths, while other cardiac causes accounted for 35%; among these, sudden death was the most frequent, accounting for no less than 26% of total deaths. This finding is in stark contrast to all other statin intervention trials in nonrenal patients, but the negative outcome may provide a useful hint, reminiscent of “The Adventure of Silver Blaze” of A. Conan Doyle (3): (Inspector Gregory) “Is there anything you wish to draw my attention to?” (Sherlock Holmes) “Yes, the curious incident of the dog in the nighttime.” (Inspector Gregory) “The dog did nothing in the nighttime!” (Sherlock Holmes) “That was the curious incident.”

In the 4D study, the dog that did not bark was the failure of sudden death to respond to a statin. Such negative outcome may suggest the following hypothesis: In nonrenal patients, coronary heart disease and cardiac ischemia from classical coronary atherosclerosis with plaque rupture is the main cause of sudden death. In contrast, in dialysis patients, classical coronary heart disease is either not responsible for sudden death or is not responsive to statins. We are not able to solve this issue, but in the following we try to go beyond simplistic explanations and to provoke discussion, reflection, and planning of future studies.

Dr. Eberhard Ritz: Let us first discuss the epidemiologic magnitude of the problem of sudden death on dialysis. In the general population, Myerburg and colleagues (4–6) found a rate of “sudden death” of 0.36 to 1.28‰ per year, and this has been confirmed in more recent series. With respect to the elderly dialysis population, a key observation is that, in the background population, the rate in males age 60 to 69 yr is as high as 8/1000/yr (7). In the general population, sudden cardiac death accounts for approximately 50% of cardiovascular deaths, and the majority, based on first recorded rhythms, is ventricular fibrillation in 75 to 80%; bradyarrhythmia accounts for most of the rest (5). Most cases are accounted for by ischemic heart disease and myocardial infarction (MI). In one autopsy study on coronary patients dying from sudden death, 59 of 113 patients had acute coronary thrombosis and 54 of 113 patients had severe narrowing of the coronary by a stable atherosclerotic plaque without acute thrombosis (8). Two further autopsy studies found ischemic heart disease in 59 and 62% respectively; nonischemic heart disease accounted for 7.5% and other pathologies (e.g., pulmonary embolism, rupture of the aorta, intracerebral hemorrhage for 27.6% [9,10]). Other possibilities to consider—for which epidemiologic information is less reliable—include suicide, seizures, primary hypoventilation, and massive pulmonary embolism (which was extremely rare in the distant past, but this may have changed with the use of erythropoietin).

In contrast, in the dialysis population, Herzog (11) found a 100-fold higher rate (93‰ per year) accounting for 60% of cardiac deaths: 47 presented as “cardiac arrest” (cause unknown) and 13% as “arrhythmia.” It has long been known that sudden death accounts for the majority of deaths in dialysis patients, particularly in diabetics, and sudden death was also identified as the most frequent cause of death in two national random samples of hemodialysis patients (12,13). It is of note that the frequency of cardiac arrest on dialysis increases with time on dialysis, almost doubling within 4 yr. It is also significantly higher, by a factor of 1.6, in diabetic compared with nondiabetic patients (11). In view of this excessive rate in dialyzed patients, it is quite surprising that a recent task force report on sudden death does not mention dialysis as a predisposing factor, illustrating the need for more interaction between specialties (5). Much of the information on dialysis patients is difficult to interpret: In nonrenal patients, the link between “sudden death” and structural cardiac disease or malfunction of ion channels (“channelopathy”) is usually straightforward; in dialysis patients, however, many confounding factors—uncommon in the general population but unrelated to heart disease—may also cause death fulfilling the criteria of sudden death (e.g., hyperkalemia, suicide, air embolism [5,6,14]). The causes of sudden death in dialysis patients may also vary between countries: One autopsy controlled study in Japan found stroke as the most frequent cause of sudden death (25.8%), followed by cardiac disease (19.4%), infectious disease (17.2%), and others (15).

The European and worldwide experience suggests that the frequency of cardiac death in the dialysis population varies in parallel with its frequency in the background population: 27% of the variation in the frequency of cardiac death in the dialysis population is explained by the mortality in the general population (16,17). This observation would suggest that the dialyzed patients largely die as the result of the same cardiovascular causes as individuals in the general population. This conclusion, however, conflicts with some of the above data and would also not explain the failure of atorvastatin in the 4D study to significantly affect sudden death (1). One cannot exclude the possibility, however, that advanced and calcific coronary heart disease of dialysis patients responds less well to statin treatment but is still causally related to sudden cardiac death. The outcome of (presumably observed) cardiac arrest in dialysis patients is abysmal: 30 d survival is 32%, and 1 yr survival is 15% (11).

Associated Factors and Pathogenesis

It has recently been stated that (in nonrenal) patients, virtually all sudden death from cardiac causes is secondary to underlying coronary disease (18). This statement is certainly correct, but while coronary heart disease is a potential cause in uremic patients, cardiac abnormalities unrelated to coronary disease are also likely to be causal, or at least aggravating, factors in the genesis of sudden death (5,6,19). Unfortunately, only scant autopsy information is available, and the few available studies fail to clarify the issue of which proportion of dialysis patients dying from sudden death have coronary disease as a causal factor.

Potential Factors Increasing the Risk for Sudden Death in Dialysis Patients (Table 1)

Therapeutic Approaches

Remaining Challenges

In the 4D Study, the frequency of sudden death was not significantly affected by treatment with statins. This observation suggests but does not prove that pathomechanisms unrelated to the growth of coronary plaques play a more important role in the genesis of sudden death of dialyzed patients compared with the general population. Adjudicated death from coronary heart disease constituted only a relatively minor proportion of cardiac deaths (9% of overall death), however; the majority of patients died from sudden death (26%) and heart failure (6%). In the absence of autopsy data, it is impossible to decide whether sudden death was associated with coronary heart disease that had possibly advanced to a stage unresponsive to statins because of calcification or inflammation. Alternatively, the noncoronary mechanisms discussed may have become the major perpetrators. This alternative is further supported by the recent observation that in patients with advanced CKD 4 or 5, intensified multiple risk factor intervention for 2 yr compared with conventional care failed to modify significantly vascular abnormalities (carotid intima media thickness and endothelial function as evaluated by brachial artery reactivity); it also failed to affect cardiovascular events (80). It remains sensible, however, to start intervention with statins in early stages of CKD, when they are still effective (81).

Let us come back to the above case: What is the best guess for what caused his sudden death? As a type 2 diabetic, he was certainly at high risk for an MI, and this possibility is by no means excluded by the fact that before the event he was completely asymptomatic. This is the rule, rather than the exception, in diabetics with uremia and advanced autonomic neuropathy. A further good possibility is hyperkalemia, usually the consequence of dietary indiscretion with or, in this case, without insufficient dialysis.

A more theoretical possibility is a potential adverse role of his treatment with a sulfonylurea compound. Sulfonylureas target ATP-dependent K⁺ channels in the pancreatic β cells, but an isoform of these receptors is also present in the heart sarcolemma and mitochondria. Inhibitors of the opening of these cardiac K_ATP channels interfere with ischemic preconditioning of the heart and have been claimed to increase cardiac risk, although recent observational data appear reassuring (82,83).

Possibly the absence of β blocker therapy might have been a predisposing factor, although in nonrenal patients, verapamil (which the patient received) was equivalent to β blockers in diabetic patients with coronary heart disease patients (84). So I cannot see what else in the management of this case could have prevented this all-too-common catastrophe.

Question and Answers

1. Dr. Ajay K. Singh (Renal Division, Brigham and Women's Hospital, Associate Professor of Medicine, Harvard Medical School): Professor Ritz, thank you for your case discussion and the comprehensive review on the syndrome of sudden death. I would like to ask you to elaborate on the potential role that automated defibrillator devices play in preventing sudden death. What do we know about the benefits? What type of patients should receive these devices?

   Prof. Ritz: Certainly the results with ICD in patients with CKD are worse than in nonrenal patients. I agree with the strong but so far unsuccessful plea of Charles Herzog, however, that to be sure about the real value in dialysis patients, one has to conduct a controlled prospective trial—and, I add, accompanied by optimal conservative treatment in the intervention and control arms including β blockers. In the absence of definitive evidence, the cardiologist/nephrologist team must make decisions with clinical common sense.

2. Dr. Colm Magee (Renal Division, Brigham and Women's Hospital, Assistant Professor of Medicine, Harvard Medical School): Professor Ritz, my impression is that an ST segment elevation MI (or, to put it another way, acute coronary artery occlusion due to plaque rupture and thrombosis) is a relatively uncommon phenomenon in dialysis patients. Is this true?

   Prof. Ritz: You are right, and you are wrong: The relative proportion of dialyzed patients dying from ST segment elevation is small relative to the patients dying without clinical signs of coronary disease from sudden death or heart failure. To be absolutely sure, one would of course wish to have solid autopsy data. On the other hand, if you calculate the absolute rates per 1000 patient-years, then you will certainly find that it is higher than in the background population.

3. Dr. Joseph V. Bonventre (Director, Renal Division, Brigham and Women's Hospital, Robert Ebert Professor of Medicine, Harvard Medical School): Coronary disease is common in patients with CKD. Could you elaborate on the contribution of myocardial ischemia to the syndrome of sudden death in patients with CKD?

   Prof. Ritz: In the absence of solid clinical data, the best approach is to analyze the available experimental data. In the subtotally nephrectomized rat, we found with Dr. Amann less capillaries in the heart, thicker intramyocardial arteries (both causing presumably less coronary perfusion reserve in response to increased demand, as in syndrome X), and finally with Dikow we found that ischemia tolerance was reduced and infarct size bigger. Information on the hypoxia sensor HIF1α might clarify the underlying pathomechanisms, but as of today we don't have the smoking gun to identify the culprit for diminished ischemia tolerance. We also don't know the relative importance of vascular versus metabolic factors reducing ischemia tolerance. The problem has, however, extreme clinical relevance, since interventions must be targeted to relevant culprit mechanisms (currently ongoing experiments in MI concerning organoprotection use, for instance, erythropoietin, but also glucose-insulin infusion, etc.).

4. Dr. J. Kevin Tucker (Renal Division, Brigham and Women's Hospital, Assistant Professor of Medicine, Harvard Medical School): Do you have information on the syndrome of sudden death among patients on daily or nocturnal dialysis?

   Prof. Ritz: I don't have information. Given the small numbers of patients on this procedure and given the fact that they are not representative for the current dialysis population, even if we had data, they would be difficult to interpret. The ongoing trial might in the future give at least a hint. Intuitively, I would say the chances are excellent that the patients on these forms of dialysis do better with respect to sudden death: They have lower BP, less electrolyte fluctuations, less sympathetic overactivity, etc.

5. Dr. Huong Tran (Renal Division, Brigham and Women's Hospital, Research Fellow, International Society of Nephrology Fellow): Dr. Ritz, could you please discuss the role of hypertension as a risk factor for sudden death in dialysis patients?

   Prof. Ritz: We know that LVH and cardiac interstitial fibrosis predispose to sudden death, so one can be sure that hypertension, which not only causes LVH but also promotes progression of LVH and fibrosis in experimental models such as the subtotally nephrectomized rat, will presumably increase the risk for sudden death.

6. Dr. Stefan G. Tullius (Chief, Division of Transplant Surgery, Brigham and Women's Hospital, Associate Professor of Medicine, Harvard Medical School): Could you please address the influence of secondary and tertiary hyperparathyroidism as a risk factor for the syndrome of sudden death? Is the rate of sudden death reduced following subtotal parathyroidectomies?

   Prof. Ritz: I don't know. One reason why this is an appropriate and important question is the finding by Block which I mentioned that hyperphosphatemia is associated with increased sudden death. Is it phosphate per se? Is it parathyroid hormone, or is it some third culprit, like FGF23? We don't know.

7. Dr. Julie Lin (Renal Division, Brigham and Women's Hospital, Assistant Professor of Medicine, Harvard Medical School): I am interested in whether β blockers modulate the risk for developing sudden death in this population. More specific, are dialysis patients who are not taking β blockers at higher risk for sudden death?

   Prof. Ritz: There is only one relatively small trial available from Naples concerning dialysis patients with cardiomyopathy (60). Carvedilol caused in this study a dramatic reduction of death with a suggestion of less sudden death (which was not statistically significant, however). Whether this finding in cardiomyopathic dialysis patients can be generalized is uncertain. It is equally uncertain, if not unlikely, that all β blockers are created equal. Given the extreme sympathetic overactivity in uremia, I assume that β blockers will affect sudden death.

What observational data (a study of ours and apparently never fully published URREA data [85]) suggest is that the risk for cardiovascular death and overall death, respectively, is less in dialysis patients on β blockers (58,59). There are, to the best of my knowledge, no specific data available on sudden death.

Disclosures

None.