|
Patients with end-stage renal disease (ESRD) whose life
prolongation is dependent on maintenance dialysis experience much
poorer outcome as compared to the general population [1-3].
Numerous reports indicate that in contrast to the general
population, where markers of over-nutrition are associated
with increased risk of cardiovascular disease, decreased nutritional
measures, such as a low body mass index (BMI) or weight-for-height
[4]
or a reduced serum cholesterol or creatinine concentration [5],
are strongly correlated with increased morbidity and mortality,
including a higher risk of cardiovascular events and death in
dialysis patients (Fig.
1). Similar findings have been reported concerning blood
pressure in that low blood pressure, and not hypertension, appears
to be more strongly related to poor outcome in dialysis patients (Table
1). These paradoxic observations have been referred to as
"reverse epidemiology"[6,
7] or "risk factor paradox"[8-10].
Such terminologies may not necessarily mean that the principles of
vascular pathophysiology are different in ESRD patients but may
indicate that there are other superimposed and more dominant factors
that cause apparent reversal of the relationships between risk
factors and outcome. The phenomenon of an established risk factor in
the general population having a markedly different and indeed
opposite predictive pattern in ESRD may not be unique to the
dialysis population. Elderly individuals in nursing homes [11],
hospitalized patients [12],
patients with malignancy [13],
and possibly other subpopulations may have similar epidemiology.
Hence, a better understanding of the causes of reverse epidemiology
in ESRD may help improve the poor outcome in this and other similar
but distinct populations.
Weight and body mass: Is obesity
protective?
The mean body weight in the general population in the United
States is on the rise [14].
An appreciable number of epidemiologic studies have shown a strong
relationship between obesity and increased risk of cardiovascular
disease and mortality in the general population [15-17].
BMI, also known as the Quetelet index [i.e., ratio of weight (kg) to
height squared (m2)][18],
and other adjusted measures of weight (such as for height) [4]
are the commonly used parameters to quantify changes in body mass
adjusted for height, and the association between body mass and
outcome. In some studies of normal adults, a J or U curve effect has
been observed in which those individuals with a low BMI also
demonstrated an increased mortality, although not as high as obese
individuals [16,
17, 19].
Hemodialysis patients have lower BMI values as compared with age-
and sex-matched controls from the general population [1,
3]. A lower BMI is consistently found to be a strong predictor
of an elevated mortality risk [20-22]
(Fig.
1). In contrast, a higher BMI, either mild to moderate
overweight or obesity, has generally not been associated with any
increase in mortality risk, except in Asian Americans [22].
BMI predicts mortality risk in dialysis patients independently of
serum albumin and recorded clinical assessment of nutritional
status. The Diaphane Collaborative Study Group in France was one of
the first to report that overall mortality risk decreased with
increasing BMI [23].
Notably, however, this was reported for a cohort of younger, mostly
nondiabetic, French patients treated with hemodialysis during the
1970s, and adjustment was only made for age and sex. Leavy et al [20]
described the predictive value for mortality over 5 years of
follow-up of a number of risk factors, recorded at baseline, in a
national sample of 3607 hemodialysis patients in the 1990s. In
hazard regression models, low BMI was independently and
significantly predictive of increased mortality; its independent
predictive value of mortality risk persisted even 5 years later. No
evidence of increasing mortality risk was found for higher values of
BMI. Fleischmann et al [21]
analyzed the BMI and its relation to 1-year mortality and hospital
stay in 1346 predominately African American hemodialysis patients.
The 1-year survival rate was significantly higher in the overweight
patients and lower in the underweight patients. With a one-unit
increase in BMI over 27.5 kg/m2, the relative risk for
dying was reduced by 30% (P< 0.04), and with a one-unit
decrease in BMI below 20, the relative risk was increased by
1.6-fold (P< 0.01). The longer the patients were on
dialysis, the lower were the BMI values. The greater duration of
dialysis vintage with lower BMI might be a cause of the increased
mortality, but it is unlikely that slightly higher dialysis vintage
with lower BMI could have accounted for such a substantial impact on
mortality [21].
Wolfe et al [24]
investigated the role of body size on 2-year mortality risk
associated with dialysis dose in 9165 hemodialysis patients and
found that body size markers, including body weight and BMI, were
independently and inversely related to mortality when adjusted for
age and diabetes as well as for Kt/V. Another recent study by Port
et al [25],
based on data from 45,967 incident hemodialysis patients after
adjustment for demographics and 18 comorbid conditions, found that
of the three body-size groups, the lowest BMI group had a 42% higher
mortality risk than the highest BMI tertile. Kopple et al [3]
evaluated approximately 13,000 hemodialysis patients and showed that
those hemodialysis patients who had greater weight-for-height
percentiles had lower mortality rates.
The prospective Dialysis Outcomes and Practice Patterns Study
(DOPPS) [26,
27] has allowed comparison of BMI-mortality relationships in the
United States and Europe and among a variety of "healthier," as
compared with "sicker" hemodialysis patient subgroups, such as
younger patients, never-smokers, and those with less chronic
illnesses. The DOPPS study provided baseline demographic,
comorbidity, and BMI data on 9714 hemodialysis patients in the
United States and Europe from 1996 to 2000. Relative mortality risk
decreased with increasing BMI. A BMI <20 kg/m2 was
consistently associated with the highest relative mortality risk.
Overall, a lower relative risk (RR) of mortality as compared with a
BMI of 23 to 24.9 kg/m2, was found for overweight (BMI 25
to 29.9 kg/m2; RR 0.84; P = 0.008), for mild
obesity (BMI 30 to 34.9 kg/m2; RR 0.73; P =
0.0003), and for moderate obesity (BMI 35 to 39.9 kg/m2;
RR 0.76; P = 0.02) (Fig.
2A). Most intriguingly, subanalysis of categories of patients
defined by overall health status at the beginning of the study and
indicated by severity of illness tertiles were performed and
resulted in similar associations (Fig.
2B), contrary to the initial hypothesis of the investigators
that reverse epidemiology may not exist in healthier or younger ESRD
patients. There was a survival benefit for healthy overweight
patients (BMI 25 to 29.9 kg/m2) that was even greater for
the obese patients (BMI  30 kg/m2), and this was observed
for the healthier as well as the sicker groups of
hemodialysis-treated patients. Even within a cohort of patients
younger than 45 years old with low comorbidity, overweight/obesity
was not associated with decreased survival [26,
27].
Finally, a similar inverse weight-mortality relationship has been
reported among chronic peritoneal dialysis patients. In the CANUSA
study, a 1% difference in percent lean body mass was associated with
a 3% change in the RR of death [28,
29]. McCusker et al [30]
found a significantly lower survival rate in patients with lower
lean body mass. For those with an initial lean body mass>73%, 63%
to 73%, and <63%, the 2-year survival probabilities were 88.3%,
81.2%, and 65.2%, respectively. Johnson et al [31]
studied the BMI in a limited number of peritoneal dialysis patients
and found that obesity conferred a significant survival advantage in
the peritoneal dialysis population. Chung, Lindholm, and Lee [32]
performed a similar study in Korean peritoneal dialysis patients and
found that lean body mass was an independent predictor of death in
peritoneal dialysis patients.
There are very few studies that have shown a deleterious effect
of a high BMI in ESRD patients. These are usually based on small
sample sizes or different subgroups of dialysis patients. For
instance, Kaizu et al [33]
studied 116 nondiabetic hemodialysis patients and used Kaplan-Meier
survival analysis and proportional hazard models to calculate the RR
of mortality in BMI quintiles. They showed that patients with BMI of
less than 16.9 kg/m2 and higher than 23.0
kg/m2 showed lowered survival relative to the patients
with BMI of 17.0 to 18.9 kg/m2. However, even in this
study, patients with a BMI <16.9 kg/m2 were shown to
have the highest risk of mortality independent of other factors.
Although high BMI has been emphasized as a strong risk factor for
morbidity and mortality in the general population, a recent study
suggests that a higher BMI in certain age groups in normal or
nonuremic individuals may not necessarily be associated with higher
morbidity and mortality [11].
Similar to ESRD patients, Grabowski and Ellis [34]
showed that a high BMI does not predict mortality in older people.
In their Longitudinal Study of Aging that examined 7527 participants
70 years old and older, they demonstrated reduced mortality in obese
older people and showed that thin older people remained more likely
to die than normal older people. A similar finding from Italy showed
that a low BMI was a significant and independent predictor of
shortened survival in hospitalized patients [12].
Blood pressure: Is hypertension protective?
The role of hypertension as a risk factor for increased
cardiovascular or cerebrovascular events in the general population
is indisputable [35-37].
Even high normal blood pressure confers a significantly greater risk
of cumulative incidence of cardiovascular events [37].
Similar to the general population, several studies have also
suggested that hypertension in dialysis patients is associated with
increased mortality risk [38,
39], including the French Tassin data on long-duration
hemodialysis [40,
41]. This has led several advisors to focus on predialysis blood
pressure (e.g., systolic blood pressure>160 mm Hg) as an
indicator of quality of care by hemodialysis units [39].
However, although hypertension in ESRD traditionally has been poorly
controlled [42,
43], elevated blood pressure may not represent the primary risk
for overall survival in dialysis patients [44-46].
Tassin data showing improved dialysis survival with blood pressure
control [40,
41] may be confounded by other survival advantages since the
patients treated with 6-hour, three times a week dialysis are
reported not to be volume expanded; hence, it can be argued that low
blood pressure in these individuals is less likely to indicate
cardiac pump failure and more likely to reflect a desirable
homodynamic state. Indeed, several recent studies, including those
based on large sample sizes, failed to find that high blood pressure
is an independent mortality risk factor [47,
48]. Findings presented by Lowrie et al [48],
Duranti, Imperiali, and Sasdelli [44],
Zager et al [47],
and Iseki et al [49]
from studies of large hemodialysis populations indicated
surprisingly little or no risk associated with systolic
hypertension. In a cross-sectional study of 936 hemodialysis
patients in the HEMO Study, Cheung et al [50]
reported no significant linear or nonlinear association between
predialysis systolic blood pressure and any of the cardiovascular
disease end points.
While contemporary thinking reiterates the link between
hypertension and cardiovascular morbidity and mortality in the
general population [35-37],
recent studies of ESRD patients point to a link between low blood
pressure and poor survival. Even though the adverse consequences of
low blood pressure is not a new concept, its impact on dialysis
outcome and on cardiovascular mortality has been underappreciated.
Iseki et al [49]
showed a strong association between low diastolic blood pressure and
risk of death in a cohort of 1243 hemodialysis patients who were
followed up for 5 years. In this study, the death rate was inversely
correlated with diastolic blood pressure, which per se showed a
significant positive correlation with serum albumin and a negative
correlation with age. Zager et al [47],
in a study of more than 5000 hemodialysis patients followed up for a
mean of 2.9 years, noted that the relative death rate for patients
with predialysis or postdialysis hypotension (systolic blood
pressure <110 mm Hg) increased to four times normal or greater
than 2.5 times normal, respectively. Port et al [51]
analyzed data from 4839 hemodialysis patients in the Case Mix
Adequacy Study of the United States Renal Data System (USRDS) and
found that when predialysis systolic blood pressure decreased below
the reference group (120 to 149 mm Hg), the relative mortality risk
increased and reached statistically significant values at pressure
measurements less than 110 mm Hg (Fig.
3). Fleischmann, Bower, and Salahudeen [8]
found that the cumulative hazard for dying was significantly
influenced by the predialysis mean arterial pressure tertiles, with
the lowest survival in the lowest third and the best survival in the
upper third. A correlation of low predialysis systolic blood
presssure with increased mortality has also been reported by Lowrie
et al [48]
and Duranti, Imperiali, and Sasdelli [44].
The concept of a U-shaped curve linking increased mortality to
low blood pressure was previously reported in a non-ESRD population
by Cruickshank, Thorpe, and Zacharias [52]
when they examined outcomes in the general population undergoing
antihypertensive treatment. Historically, hypertension is associated
with concentric left ventricular hypertrophy (LVH), ventricular
dilatation, ischemic heart disease, and congestive heart failure.
However, after the development of cardiac failure, low blood
pressure or hypotension predicts mortality and is a potential marker
for severity of cardiac disease [38,
53]. In addition, patients presenting with baseline predialysis
hypotension may possess such subclinically significant risk factors
or comorbidities such as hypoalbuminemia, heart failure, and
ischemic cardiomyopathy [49,
54]. Low blood pressure may be a reflection of autonomic
neuropathy that, in turn, is a marker for more severe uremic
complications. Because antihypertensive drug therapy may contribute
to low predialysis systolic blood pressure, one may hypothesize that
either overmedication (perhaps specific antihypertensive agents) or
cardiac pump failure are the major contributors to the observed
associations. To date, no study has examined the potential role of
medication as a cause of low blood pressure and increased mortality
in dialysis patients. Ongoing and future studies might address these
issues.
Serum cholesterol: Is hypercholesterolemia
protective?
In the general population, hypercholesterolemia is a known risk
factor for cardiovascular morbidity and mortality [55-57].
Among lipid components, increased serum levels of low-density
liproprotein (LDL) and non-high-density lipoprotein (HDL)
cholesterol appear to have the strongest predictive value for poor
cardiovascular outcome, whereas HDL-cholesterol is generally
considered protective in the general population [57].
In the Diaphane Collaborative Study Group, Degoulet et al [23]
reported that a lower plasma total cholesterol level was associated
with a significantly higher risk of death from cardiovascular
disease in a cohort of 1453 French hemodialysis patients. Lowrie and
Lew [5]
analyzed a group of more than 12,000 prevalent hemodialysis patients
and found that serum cholesterol level was inversely correlated with
the risk for death in a multivariate logistic regression adjusting
for several case-mix factors and biochemical parameters (Fig.
4). Avram et al [58]
showed that serum cholesterol concentration was elevated in the
long-term dialysis survivors. Fleischmann, Bower, and Salahudeen [8]
studied 453 hemodialysis patients prospectively and showed that
among lipids, the second and third tertiles of total cholesterol as
well as LDL and HDL cholesterol were associated with better
survival, although this was only a trend that was not statistically
significant. More recently, Iseki et al [59]
reported similar findings based on a sample of 1167 hemodialysis
patients who were followed for up to 10 years. We showed similar
findings in a small cohort of hemodialysis patients who were
followed for up to 12 months [60].
Of note, some less recent studies in ESRD patients indicated an
opposite direction of the serum cholesterol-mortality relationship,
consistent with that seen in the general population. For example, a
retrospective study of a cohort of 190 peritoneal patients with an
average follow-up of 12 months and based on Cox model showed that an
elevated serum cholesterol level was associated with increased
mortality [61].
Nevertheless, this same study showed that a low serum creatinine
concentration also predicted mortality (see below).
Similar findings regarding the association of low serum
cholesterol and poor outcome have been reported for elderly persons
who do not have ESRD [62-64].
For instance, Volpato et al [63]
investigated the relationship between low cholesterol and mortality
in older persons in a prospective cohort consisting of 4128 elderly
patients with a mean age of 78.7 years and a median follow-up period
of 4.9 years. They found that in those with a low serum cholesterol
level, defined as  160 mg/dL, all-cause mortality was
significantly higher than those with normal or high serum
cholesterol concentrations. Krumholz et al [64]
failed to show any significant association between
hypercholesterolemia or low HDL-cholesterol and all-cause mortality,
coronary heart disease mortality, or hospitalization for myocardial
infarction or unstable angina in a cohort of persons older than 70
years.
Serum creatinine: Is high better than low?
In the general population, a slight or moderate increase in serum
creatinine has been shown to be an independent risk factor of
cardiovascular disease [65-67].
A secondary analysis of the HOPE Study showed that in patients who
had either preexisting vascular disease or diabetes mellitus and an
additional cardiovascular risk factor, the presence of mild renal
insufficiency significantly increased the risk for subsequent
cardiovascular events [65].
The HOT Study showed that a baseline elevation in serum creatinine
is powerful predictors of cardiovascular events and death [66].
In dialysis patients, the serum creatinine, a reflection of
muscle mass or meat ingestion and/or the degree of dialysis
efficiency, has also been shown to be a predictor of mortality but
in the opposite direction (i.e., those dialysis patients with a
higher serum creatinine live longer). This is contrary to the notion
that well-dialyzed dialysis patients with higher Kt/V should have
lower serum creatinine concentrations compared to those in whom
dialysis treatment is not adequate. Lowrie and Lew [5]
found that the serum creatinine level was inversely correlated with
the risk for death. Avram et al [58]
found that the enrollment serum creatinine was significantly higher
among long- and very long-term hemodialysis and peritoneal
survivors. In the study by Tattersall, Greenwood, and Farrington [68],
there was no significant change in the predialysis serum creatinine
values for hemodialysis patients from the time they initiated
dialysis as compared with their values 6 months later. These results
suggest that the factors that link lower predialysis serum
creatinine values to increased mortality in dialysis patients may be
determined when the patients commence renal replacement therapy. In
the 1992 USRDS Annual Report, the investigators analyzed
hemodialysis patients initiating dialysis in 1986 and 1987 and found
that increasing serum creatinine level was associated with
decreasing mortality [69].
Fink et al [70]
studied 5388 incident hemodialysis patients followed up for almost 2
years and found that serum creatinine level was inversely correlated
with mortality risk. Among studies of peritoneal dialysis patients,
a retrospective study of a cohort of 190 peritoneal patients with an
average follow-up of 12 months and based on the Cox model found that
a low serum creatinine was an independent variable significantly
associated with increased risk of death [71].
Plasma homocysteine: Is high better than
low?
Plasma total homocysteine (tHcys) has been established as a new
risk factor for increased cardiovascular morbidity and mortality in
the general population [72,
73]. Estimated survival among non-ESRD patients with coronary
artery disease can be stratified according to plasma total
homocysteine levels. Higher homocysteine levels are associated with
a worse outcome [72].
A similar association as in the general population was found in some
studies of dialysis patients (i.e., a higher plasma tHcys was found
to be associated with increased risk of cardiovascular disease) [74-79].
However, conflicting findings have been reported regarding the
association between the plasma tHcys level and the prevalence of
cardiovascular disease in chronic renal failure patients. Bostom et
al found no relationship between the fasting plasma tHcys level and
the prevalence of cardiovascular disese in dialysis patients, using
crude or multiple logistic regression analyses adjusted for other
risk factors [80].
More recently Suliman et al [81,
82] found that among 117 hemodialyis patients, those with
cardiovascular disease (60%) had significantly lower tHcys levels
than those without cardiovascular disease, although both groups
still had higher plasma tHcys levels than the general population.
Wrone et al[83]
performed a cross-sectional analysis of 459 dialysis patients, and
showed that those with a history of cardiovascular disease had lower
levels of predialysis plasma tHcys than those without a history of
cardiovascular disease. This negative relationship persisted in
multivariate analyses when controlling for predictors of
cardiovascular disease, such as age, gender, and BMI. Mean plasma
tHcys was higher in patients without a history of cardiovascular
disease than in those with cardiovascular disease [83].
Of note, both of the above-mentioned studies found a statistically
significant, positive correlation between predialysis plasma tHcys
and albumin concentration, which is a known, strong predictor of
mortality in dialysis patients. We have recently found a similar
positive correlation in a sample of 368 hemodialysis patients, which
may indicate that plasma tHcys is a nutritional and/or inflammatory
marker in dialysis patients (abstract; Kalantar-Zadeh
et al, Am Soc Nephrol, 13:222A, 2002). It is noteworthy that
studies showing the inverse relationship in dialysis patients are
all of recent origin. One reason why similar results had not been
reported in the past may be related to publication bias that will be
discussed later in this manuscript (see below).
However, these observations do not necessarily exclude the role
of hyperhomocysteinemia as a risk factor for poor ESRD outcome,
since other factors might be present that confound the relationship
between tHcys levels and cardiovascular disease. Of particular
importance is the fact that even studies that found a reverse
epidemiology confirmed that tHcys was higher in dialysis patients as
compared to the general population [82,
83]. Assuming that an increase in tHcys level carries an
independent risk of cardiovascular, this may imply that almost all
dialysis patients are already exposed to this risk. Thus, the
apparently paradoxic association between tHcys and cardiovascular
disease does not strictly conflict with the thesis that
hyperhomocysteinemia is a risk factor for cardiovascular disease.
However, within this population, those with the highest tHcys level
may have some survival advantages due to better nutrition.
Other possible examples of reverse
epidemiology
Excess parathyroid hormone (PTH) has long been considered
detrimental to the health of patients, including those with ESRD [84].
PTH has been implicated as a multisystem uremic toxin, and
hyperparathyroidism can be a debilitating complication in dialysis
patients [84].
Hyperphosphatemia that is closely related to hyperparathyroidism is
associated with increased mortality in dialysis patients [85].
Avram et al [86]
studied prospectively the relationship between the enrollment serum
intact PTH and all-cause mortality in 345 hemodialysis and 277
peritoneal dialysis patients for 14 years and found that lower than
expected levels of PTH in uremic patients are associated with
increased mortality. Moreover, Guh et al [87]
recently reported similar findings that low levels of serum PTH at
entry and lower time-dependent PTH levels predict mortality in
hemodialysis patients. Avram et al [86]
hypothesized that inadequate protein intake, phosphorus intake or
both result in impaired development of the expected secondary
hyperparathyroidism and in the excess mortality risk inherent with
malnutrition. However, to date epidemiologic studies have shown a
positive association between a high serum phosphorus and poor
outcome among ESRD patients [85,
86]. Hence, the association between serum PTH and outcome in
dialysis patients may be unrelated to serum phosphorus and may
reflect other aspects of nutritional status.
Another example is serum ferritin and its association with
anemia. In the general population, a low serum ferritin is a marker
of iron deficiency and anemia that may not respond to erythropoietin
unless iron is repleted [88].
However, in ESRD patients, a high and not a low serum ferritin is
associated with a more severe and refractory anemia [89,
90]. This may be due to hyporesponsiveness to erythropoietin,
which can occur in the setting of the malnutrition-inflammation
complex syndrome (MICS) in dialysis patients [60],
especially since ferritin is an acute phase reactant [89,
90].
Finally, another possible example of risk factor reversal is the
association between the amount of energy (calorie) or food intake
and mortality. In the general population, an increased energy intake
may be associated with increased BMI and hypercholesterolemia and,
hence, with a higher mortality rate and risk of cardiovascular
disease [91,
92]. In contrast, in dialysis patients increased protein intake,
estimated from the urea generation and urea kinetics, may lead to an
improved nutritional status, reflected by an increased albumin
level, and a better survival [93-95].
However, data in dialysis patients are mostly based on indirect
measurements of protein intake and studies concerning the possible
association between energy (calorie) intake and dialysis outcomes
are yet to be developed.
The concept of reverse epidemiology appears at first to be
confusing, especially because hypertension, obesity, and high levels
of serum cholesterol, creatinine, and homocysteine are established
risk factors for ischemic heart disease in the general population.
The paradox becomes even more paramount when it is recognized that
it is not a question of the existence or lack of an association
between these risk factors and the clinical outcomes, but often the
complete reversal and indeed the opposite direction of this
relationship. Hence, there must be prevailing conditions that are
characteristically present in dialysis patients that render them
more susceptible to a poor outcome when low body mass, low blood
pressure, or decreased serum values of cholesterol, creatinine, and
homocysteine are present. Several suggested explanations are offered
for this inverse association. Survival bias, time discrepancies
among competing risk factors, the MICS, and several other hypotheses
are presented as possible causes.
Survival bias
Since dialysis patients have undergone specific processes of
selection and survival, their characteristics may not be similar to
the general population. Therefore, the relationship between the risk
factors and outcomes may have been modified. According to an
analysis based on the National Health and Nutrition Examination
Survey [96],
in the United States there are over 10 to 20 million patients with
chronic kidney disease (CKD) and an elevated serum creatinine due to
a chronic, irreversible and probably progressive damage to the
kidney. However, according to the USRDS data [1],
the number of people with ESRD is approximately 400,000 at this time
and is projected to increase to almost 650,000 by the year 2010.
This constitutes only a small fraction (less than 5%) of the large
pool of CKD patients in the United States. Current thinking suggests
that a large population of CKD patients will not live long enough to
develop ESRD and commence maintenance dialysis. One explanation for
this phenomenon is that CKD patients have a high mortality rate
since many of them have severe and complex comorbid conditions, such
as diabetes mellitus, hypertension, and atherosclerotic vascular
disease. Moreover, renal disease with proteinuria or an increased
serum creatinine by itself is an independent risk factor for greater
morbidity and mortality, particularly from atherosclerotic heart and
cerebrovascular diseases [65,
66]. Thus, many CKD patients do not reach ESRD due to their high
mortality. This can explain why only a small proportion of CKD
patients develop ESRD. It is not clear what specific characteristics
of this relatively small percentage of CKD patients give them a
greater survival chance to reach ESRD status. An alternative
explanation, however, is that those CKD patients who develop ESRD
simply have a more accelerated rate of progression of their chronic
renal failure. But whatever the survival features are, these
"unfortunately lucky" individuals may be considered as "specifically
selected" people, who are not necessarily genetically or
phenotypically similar to their CKD predecessors and may not have
the survival characteristics and epidemiologic features of their
progenitors. Some of those who have survived to comprise the ESRD
population might be "exceptional individuals" who successfully
survived the conventional (traditional) risk factors, which are
often present strongly in CKD patients. Hence, the assumption that
the epidemiology of cardiovascular risk factors is the same in ESRD
individuals as in the general population may be flawed, because a
survival bias, a form of selection bias, may heavily influence the
epidemiologic constellations in this small proportion of CKD
survivors (i.e., the dialysis patients). A similar trend is observed
in elderly individuals, who have reached advanced age possibly due
to survival advantages, and they also are reported to have a
different epidemiology with different cardiovascular disease risk
factors than the general population [11,
13, 34].
It should be noted that many epidemiologic studies in ESRD
patients are based on a mixed pool of incident and prevalent
dialysis patients (i.e., those who have started dialysis recently
and those who have been on dialysis for a number of years are
treated equally in such studies). This may lead to what
epidemiologists call "incidence-prevalence" or "survival" bias,
which per se is a form of selection bias [97].
In other words, those incident patients who had more risk factors or
were more vulnerable to such risks did not survive long enough to be
included in a cross-sectional study that encompasses many prevalent
patients who have been on dialysis for a number of years and who
hence over-represent ESRD patients who have exceptional survival
advantages. Indeed, the majority of incident dialysis patients die,
especially due to the exceptionally high ESRD mortality rate,
currently approximately 20% in the United States [1].
Hence, even longitudinal studies that are based on a mixture of
incident and prevalent dialysis patients may be heavily influenced
by a survival bias [98].
Survival bias can potentially exert a strong influence on both
epidemiologic studies and clinical trials, especially when patients
who have varying degrees of duration of dialysis are enrolled
together [97,
98]. In this regard, it is possible that the concept of reverse
epidemiology is a consequence of systematic errors in epidemiologic
data analysis [83].
However, some epidemiologic data based exclusively on incident
dialysis patients still show the same risk factor reversal [25],
indicating that this phenomenon cannot be solely explained by
survival bias.
Time discrepancies among competitive risk
factors
Survival advantages that exist in obese, hypertensive,
hypercholesterolemic, hypercreatininemic, or hyperhomocysteinemic
dialysis patients may, in the short-term, outweigh the harmful
effects of these risk factors on cardiovascular disease in the
long-term. Since dialysis patients have a mortality risk that is
greater than the general population [2],
the long-term effects of these risk factors on future mortality may
be overwhelmed by the short-term effects of other factors on
dialysis mortality. Indeed, it may be difficult, if not impossible,
to observe a significantly greater life expectancy with reduction of
the traditional risk factors in dialysis patients who have a short
life expectancy, even when such a risk factor reduction is
beneficial in the general population who have a normal life
expectancy.
An example of this time discrepancy in non-ESRD populations may
concern obesity and underweight. In the general population in the
United States, as well as in most industrialized countries,
manifestations of over-nutrition such as overweight/obesity and
hypercholesterolemia are major risk factors for cardiovascular
mortality [15-19,
56, 57]. These are also countries where people have a greater
life expectancy compared to individuals in other parts of the world.
Studies of risk factors of cardiovascular mortality are essentially
based on these populations. In contrast, in developing countries,
which represent the majority of the world's population,
under-nutrition is still a powerful determinant of poor clinical
outcome and morbidity and mortality, leading to a shorter life
expectancy [99-101].
A hypothetical demonstration of the combined effect of competitive
risk factors (over- versus under-nutrition) is presented in Figure
5.
Thus, in dialysis patients who have a short life expectancy, any
factor that may improve short-term survival, such as high blood
pressure, obesity, and hypercholesterolemia may exert a desirable
effect on longevity, whereas conditions that are traditionally
associated with long-term survival such as rigorous blood pressure
control, low body weight-for-height, and low serum cholesterol and
homocysteine, may be less relevant. Such factors may even be harmful
if they cause, represent, or aggravate states of under-nutrition,
hypotension or cardiac failure.
Malnutrition inflammation complex syndrome
(MICS)
Measures of protein-energy malnutrition (PEM) and inflammation
are major predictors of clinical outcome in dialysis patients [3-5,
7, 9]. Dialysis patients with cardiovascular disease have a
higher prevalence of malnutrition and hypoalbuminemia and a lower
protein intake than those without cardiovascular disease [7,
102]. Several studies report strong relationships between
hypoalbuminemia and cardiovascular disease in dialysis patients and
suggest that hypoalbuminemia is an important risk factor for
cardiovascular disease in these individuals [54,
103-105]. Cardiac diseases such as heart failure may engender
anorexia and, if sufficiently severe, may independently induce
muscle wasting, which is also known as cardiac cachexia [102,
106]. However, it is not completely clear how cardiac disease,
inflammation, hypoalbuminemia, and other measures of PEM in dialysis
patients are interrelated [106,
107]. It has been postulated that the common link among these
conditions is inflammation [7,
54, 103, 104, 106, 108]. A common mechanism for the development
of cardiovascular disease and PEM in dialysis patients may be
cytokine activation associated with reduced renal function or other
proinflammatory conditions in dialysis patients. Frequent contact
with dialysis membranes, vascular access grafts or catheters or
dialysate, or peritoneal dialysis fluid may each constitute a
proinflammatory condition [7,
109]. Increased release or activation of inflammatory cytokines,
such as interleukin-6 (IL-6) or tumor necrosis factor- (TNF-), may
suppress appetite, cause muscle proteolysis and hypoalbuminemia, and
may be involved in the processes that lead to atherosclerosis [66,
104, 110]. Nevertheless, the degree to which PEM in dialysis
patients is caused by inflammation is not clear [7,
110]. Some studies suggest that PEM and inflammation each
independently contribute to hypoalbuminemia and subsequently
increase morbidity and mortality [105].
Since both PEM and inflammation are strongly associated with each
other and can change many nutritional measures in the same
direction, and because the relative contributions of measures of
these two conditions to each other and to outcomes in dialysis
patients are not yet well defined, MICS has been suggested to denote
the important contribution of both of these conditions to ESRD
outcome [7,
60]. Some investigators have used other terminologies such as
"malnutrition inflammation atherosclerosis" (MIA) to emphasize the
importance of atherosclerosis as the consequence of MICS [54,
110]. Furthermore, oxidative stress may play an important
integral part in the associations between inflammation, malnutrition
and atherosclerosis, since variations of serum albumin may be
correlated not only with outcome but also with inflammation,
cardiovascular disease, and serum cholesterol and tHcys
concentrations [111,
112]. No matter what it is called or caused by, the theory of
reverse epidemiology in ESRD is compatible with the existence of
MICS or MIA and its interplay with the traditional risk factors.
The existence of paradoxic risk factors could be accentuated by
the MICS, possibly in several ways. First, patients who are
underweight or who have a low serum cholesterol, creatinine, or
homocysteine, may be suffering from the MICS and its poor outcome.
Thus, MICS may both cause these alterations and also be associated
with increased mortality either caused by the illnesses that
engender the MICS or the cardiovascular disease that seems to be
promoted by the MICS [106,
113, 114]. Second, the above paradoxic factors may indicate a
state of under-nutrition, which may predispose to infection or other
inflammatory processes [7].
Finally, it has been argued that when individuals are malnourished,
they are more susceptible to the ravages of inflammatory diseases [113].
Hence, any condition that potentially attenuates the magnitude of
PEM or inflammation should be favorable to dialysis patients. This
notion can explain why there is a reverse epidemiology of
cardiovascular risk factors in dialysis patients, especially since
almost all the so-called nontraditional protective factors, such as
obesity, hypercholesterolemia, hypercreatininemia, and
hyperhomocysteinemia, are related to nutritional status. Suliman et
al [81,
82] have reported a more specific example of the contribution of
the MICS to risk factor reversal concerning with
hyperhomocysteinemia in dialysis patients. In their study, plasma
tHcys levels were shown to be dependent on nutritional status,
protein intake, and serum albumin in hemodialysis patients.
Hemodialysis patients with cardiovascular disease had lower tHcys
levels as well as a higher prevalence of malnutrition and
hypoalbuminemia than those without cardiovascular disease.
Furthermore, in another study plasma tHcys was shown to rise during
treatment of malnourished peritoneal dialysis patients with an amino
acid-containing peritoneal dialysate (1.7 g methionine/day) [115].
The presence of MICS may further reinforce this relationship by
suppressing serum albumin, which is an important carrier protein for
plasma tHcys [81,
82, 106].
The puzzling inverse relationship between a low blood pressure
and poor outcome in the dialysis population might also be accounted
for by nutritional status and/or inflammation. Iseki et al [49]
showed a significant association between a low diastolic blood
presssure, hypoalbuminemia, and risk of death in a cohort of
hemodialysis patients. They analyzed the causes of death with regard
to the relationship between serum albumin and blood pressure in 1243
hemodialysis patients who were followed for up to 5 years. The death
rate was inversely correlated with diastolic blood pressure, which
per se was positively correlated with serum albumin and negatively
correlated with age. Hence, hypotension may in some cases be a
manifestation of MICS in hemodialysis patients.
Other hypotheses
Goldberg, Tindira, and Harter [116]
suggest that the paradigm of coronary artery disease in ESRD shifts
from solely traditional risk factors, such as age, diabetes,
hyperlipidemia, and hypertension, to a number of additional factors
that may regulate coronary event rates in a different way in uremia.
Theoretically, hypotension and under-nutrition can contribute to the
increased relative risk of death by several mechanisms, such as
acute coronary syndrome, autoregulation dysfunction, ischemia, and
arrhythmiagenecity [117,
118]. These factors, if indeed altered or exaggerated in uremic
milieu, may provide an environment in dialysis patients, in whom
hypotension and undernutrition may become more influential than the
traditional risk factors in the development of cardiovascular
disese.
Chronic inflammatory processes may independently predispose to
both malnutrition and atherosclerosis. IL-6 predicts
hypocholesterolemia [119],
malnutrition [120,
121], and atherosclerosis [122,
123] in dialysis patients. Inflammation can indeed lead to
hypocholesterolemia in ESRD patients [59,
119]. Moreover, chronic infusion of IL-1 or TNF- (cachectin) has been shown to cause
anorexia, rapid weight loss, and a decline in body protein stores [124].
Hence, given the fact that inflammation is more common in ESRD
patients as compared to the general population [104,
120], inflammation by itself may precede under-nutrition and
lead to the risk factor reversals in individuals with ESRD.
There may be other confounding factors that may explain some of
the risk factor reversals in dialysis patients. Both low body mass
and low blood pressure may be consequences of cigarette smoking [125,
126] and/or congestive heart failure [102,
106], conditions that are associated with increased ESRD
mortality. These conditions may also be associated with an increased
prevalence of inflammation and malnutrition [127,
128]. Heart failure may engender anorexia and may independently
induce cachexia [106].
A low blood pressure, a potential indicator of severe heart disease,
is thus associated with poor outcome [53].
Moreover, a low blood pressure may be caused by autonomic neuropathy
that, in turn, can be the result of uremic toxicity or the ravages
of diabetes [129].
Hence, it may not be normo- or hypotension per se that is
detrimental but rather the underlying cause of low blood pressure
(i.e., cardiac pump failure and/or autonomic neuropathy).
Finally, a less well-substantiated hypothesis represents the
notion that what we consider as reverse epidemiology (the stronger
impact of under nutrition) may indeed be the natural epidemiology in
human and that the so-called conventional epidemiology
(over-nutrition) is a new and unusual phenomenon in human history [130].
In recent decades, excess weight and obesity have become mass
phenomena with a pronounced upward trend in most industrialized
nations. However, despite the detrimental effects of being
overweight, these populations indeed live longer than ever.
Moreover, with aging, the detrimental effects of obesity,
over-nutrition, and hypertension may diminish if not disappear, a
similar trend that can be observed in ESRD population as well.
The reversal of certain key risk factors in dialysis patients
poses serious questions. Is the increasing prevalence of obesity and
its detrimental impact to health in the general population of any
relevance in dialysis patients? Do overnutrition, obesity,
hypertension, or hypercholesterolemia that promote atherosclerosis
and mortality in the general population prevent poor outcome in
dialysis patients, and if so how? Should dialysis patients be
advised to increase their nutrient intake in order to gain weight
and to increase their serum cholesterol, creatinine, and
homocysteine levels? Should their target blood pressure be higher?
Can these reversed relationships be used to establish therapeutic
goals?
Publication bias may have handicapped or delayed reporting such
paradoxical findings in dialysis patients as the association between
plasma homocysteine and cardiovascular disease in dialysis patients,
since the investigators' first impression upon encountering results
with inversed association may be to consider them erroneous or
flawed and hence be reluctant to report them [131].
However, as more reports indicative of reverse epidemiology in ESRD
have been published recently, more investigators may be encouraged
to report their similar findings. This may explain why more frequent
reports and publications consistent with the reverse epidemiology
have emerged only recently.
It is important to appreciate that some of the discussed risk
factors may represent different biologic or medical phenomena in
ESRD patients as compared to the general population. Serum
creatinine, for instance, is a reflection of renal function in the
general population, whereas it is essentially representative of
skeletal muscle mass and/or meat ingestion as well as the dose of
dialysis in ESRD patients. Similarly, a pre-dialysis blood pressure
measurement may represent a different underlying disease processes
in hemodialysis patients who are often volume overloaded. Thus, a
low predialysis blood pressure in patients who are likely to be
volume expanded is more probably due to a sick heart, whereas in the
general population it may more likely indicate excellent circulatory
homodynamics. Hence, the etiology of "reverse epidemiology" in
dialysis patients may be quite different for various risk factors,
and the term "reverse epidemiology" may be a misnomer. Nevertheless,
it is important to first exhaust the possibility of a single
unifying entity to be accounted for all or most of the
above-mentioned risk factor reversals. We believe that PEM and
inflammation (MICS) are the best candidates.
Despite all these concerns, the evidence is strong that a risk
factor paradox indeed exits in those who reach ESRD and who continue
to have an unacceptably high rate of mortality, currently
approximately 20% in the United States. This high mortality rate has
not substantially been changed in the recent years despite
aggressive efforts toward an optimal management of traditional risk
factors in these individuals. Hence, it is important to explore the
causes of reverse epidemiology and to ascertain how best to reverse
these associations in dialysis patients. We believe that in dialysis
patients more attention should be focused on optimal management of
under-nutrition and inflammation based on the mechanisms responsible
for the reverse epidemiology. However, premature or radical
conclusions to discontinue antihypertensive or antihyperlipidemic
treatment should be avoided until such information is forthcoming.
For instance, the antihypercholesterolemic agent,
3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitor
or such antihypertensive agents as angiotensin-converting enzyme
inhibitors may have an anti-inflammatory effect, which can be
beneficial in the management of the elements of MICS and improved
outcome in dialysis patients irrespective of the existence of
reverse epidemiology [132-135].
It is also important to appreciate that most of the examples of
reverse epidemiology do not apply to renal transplant recipients, in
whom obesity and hyperlipidemia are still reported to be quite
common and strong risk factors for cardiovascular disease and poor
outcome [136].
Although data presented in this review suggest that a higher body
mass, hypertension, and hypercholesterolemia are associated with
reduced morbidity and mortality in the vulnerable population of
ESRD, it is possible that, in the long run, overweight patients may
suffer from more cardiovascular consequences if they could survive
sufficiently long [137].
Therefore, extended observations with sequential measurement of BMI,
blood pressure, and serum levels of cholesterol, creatrinine,
homocysteine and other relevant markers should be helpful to
identify different subgroups of dialysis patients who may have
traditional epidemiology as well as those who have reverse
epidemiology. As more effective treatments for ESRD patents become
available, it is possible that there may be a reversal of the
reverse epidemiology and a return of the traditional epidemiology to
many subgroups of dialysis patients, as is currently found in
kidney-transplant patients. The question may remain unanswered as to
what is indeed the normal epidemiology and what is reverse.
Randomized, prospective, controlled clinical trials to examine the
reversal of the traditional associations or the paradoxic risk
factors will be most beneficial to the maintenance dialysis
patients.
A recent analysis by Lowrie et al [138],
based on 43,334 MHD patients, also showed an improved survival in
those with higher BMI values. |