Association of anemia and erythropoiesis stimulating agents with inflammatory biomarkers in chronic kidney disease

Association of anemia and erythropoiesis stimulating agents withinflammatory biomarkers in chronic kidney disease

Sai Ram Keithi-Reddy1, Francesco Addabbo2, Tejas V. Patel1, Bharati V. Mittal1, Michael S.Goligorsky2, and Ajay K. Singh11Renal Division, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School,Boston, Massachusetts, USA2Department of Medicine and Department of Pharmacology, Renal Research Institute, New YorkMedical College, Valhalla, New York, USA

AbstractInflammatory cytokines are important predictors of cardiovascular mortality especially in patientswith chronic kidney disease. Here we explored the relationship of anemia and epoetin treatment toinflammatory cytokine levels in patients with chronic kidney disease. One hundred non-dialysispatients with chronic kidney disease over 18 years of age were evenly split into anemic and non-anemic cohorts. Of the 50 anemic patients, 23 were receiving erythropoiesis stimulating agentstreatments. Levels of tumor necrosis factor (TNF)-α were found to be significantly higher and serumalbumin was significantly lower with trends towards higher interleukin (IL)-6 and IL-8 in anemiccompared to non-anemic patients. Further analysis by multiple logistic regression found that anemicpatients treated with erythropoiesis stimulating agents had significantly higher odds for the uppertwo quartiles for IL-6, IL-8 and TNF-α compared to non-anemic patients. Our study found that theanemia of chronic kidney disease was associated with up regulation of TNF-α, and possibly IL-6 andIL-8 along with increased levels of these proinflammatory cytokines in patients treated with epoetin.

Keywordsanemia; erythropoietin; inflammation; chronic kidney disease; C-reactive protein

Cardiovascular disease (CVD) remains the principal cause of mortality in patients with chronickidney disease (CKD). The association of anemia with cardiovascular outcomes is well known,but underlying mechanisms are not well understood.1 On the other hand, targeting of a higherhemoglobin with higher doses of erythropoiesis-stimulating agents (ESAs) worsenscardiovascular outcomes in CKD.2,3 The reason for higher mortality in patients targeted forthe higher hemoglobin is an area of intense research. Possibilities include a heightened

© 2008 International Society of NephrologyCorrespondence: Ajay K. Singh, Renal Division, Brigham and Women’s Hospital, 75 Francis Street, Boston, Massachusette 02115,USA. E-mail: [email protected]: SK, MSG, and AKS designed the research; FA performed molecular studies; SK and AKS analyzed data and wrote thepaper; TVP assisted with statistical analysis; BVM and AKS performed the clinical studies.DISCLOSUREDr Singh reports receiving consulting fees from Ortho Biotech Clinical Affairs/Johnson and Johnson, Amgen, Roche, Merck, Abbott,Watson and lecture fees from Ortho Biotech Clinical Affairs/Johnson and Johnson, Roche, Amgen, and Watson; serving on advisoryboards for Ortho Biotech Clinical Affairs, Roche, Watson, AMAG, and Amgen; and receiving grant support from Ortho Biotech ClinicalAffairs, Roche, Johnson & Johnson, Amgen, Watson. Dr Singh is the Medical Director of Dialysis Clinics Inc. Dr Goligorsky, Dr Patel,Dr Mittal, Dr Addabbo, and Dr Keithi-Reddy report no conflicts of interests.

NIH Public AccessAuthor ManuscriptKidney Int. Author manuscript; available in PMC 2009 September 8.

Published in final edited form as:Kidney Int. 2008 September ; 74(6): 782–790. doi:10.1038/ki.2008.245.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

prothrombotic tendency because of higher hemoglobin levels or a potentially toxic effect ofESAs on the cardiovascular system.4 The potential influence of inflammation in CKD patientswith anemia on ESAs treatment is not well understood.

Evidence supports a role for inflammation in atherosclerosis. 5,6 Inflammation has also beenshown to play a role in atherosclerosis in all stages of CKD.7 Some of the biomarkers studiedin relation to inflammation in CKD include interleukin (IL)-1β, IL-6, IL-8, tumor necrosisfactor (TNF)-α, and C-reactive protein.8,9 Elevated levels of these markers are detectable evenamong patients without evidence of clinical or sub-clinical CVD. Thus, inflammation presentsas an early phenomenon among CKD patients and might play a role in promotingatherosclerosis. These mechanisms may, in part, explain the high incidence of cardiovascularevents in CKD patients and the high prevalence of CVD in end-stage renal disease patients.

We hypothesized that exposure to ESA therapy may result in proinflammatory cytokineactivation. Consequently, in this study, we evaluated the levels of several bio-inflammatorymarkers in CKD patients with and without anemia who were either treated with ESA therapyor were ESA-naive.

RESULTSPatient characteristics by CKD-associated anemia

The baseline characteristic of the CKD cohort is presented in Table 1. Anemia was defined bymodified World Health Organization (WHO) criteria as less than 12 g/dl in women and lessthan 13 g/dl in men or treatment with ESAs. Of the 100 patients analyzed, 50 patients wereanemic by the modified WHO criteria. The mean Hgb level in anemic patients was 11.2 g/dland in non-anemic patients was 13.8 g/dl. Anemic patients tended to be predominantly maleand significantly older by age, with a mean age of 64 years. These patients also had asignificantly reduced estimated glomerular filtration rate (eGFR) calculated using theModification of Diet in Renal Disease II equation and were predominantly in stage 3 and stage4 CKD (eGFR 15–60 ml/min per 1.73m2). Non-anemic patients were predominantly in stage2 and stage 3 CKD (eGFR 30–90 ml/min per 1.73m2). No other significant differences wereobserved. In all, 23% (N=23) of patients were treated with ESAs; of ESA-treated patients, 57%(N=13) were on epoetin-α (median dose of 10,000 U/week) and 43% (N=10) on darbepoetin.

Proinflammatory markers and anemiaA non-parametric test was used for comparison as the data were not normally distributed. Foranemic (by modified WHO classification) vs non-anemic patients, the median (interquartilerange) levels for IL-6 were 4.65 pg/ml (0.64, 12.38) and 1.09 pg/ml (0.64, 8.34) (P=0.12),respectively; for IL-8 were 7.85 pg/ml (2.64, 13.17) vs 3.66 pg/ml (1.37, 10.15) (P=0.05),respectively; and for TNF-α were 4.54 pg/ml (1.89, 7.18) vs 2.36 pg/ml (1.10, 4.54) (P=0.019),respectively. Notably, serum albumin levels were 4.1 g/100 ml (3.90, 4.30) and 4.40 g/100 ml(4.07, 4.52) (P<0.002) for anemic and non-anemic patients respectively, whereas the serumferritin levels were not different: 96 ng/ml (52, 282) and 93 ng/ml (53, 151) (P=0.56) in anemicand non-anemic subjects respectively.

As there are different classifications for anemia and definitions may be viewed as relativelyarbitrary, we performed four sensitivity analyses for different hemoglobin levels (Table 2).The first sensitivity analysis was performed by strictly classifying subjects based on the WHOcriteria with or without including treatment with ESAs as a cause of anemia (WHO definitionof anemia is ≤13 g/dl for adult male and ≤12 g/dl an adult female): 48 patients were classifiedas anemic. The levels of IL-8 (P=0.0013) and TNF-α (0.016) were significantly higher butthere was no difference for IL-6 (P=0.11), whereas serum albumin levels were significantly

Keithi-Reddy et al. Page 2

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

lower (P<0.001) in patients with anemia. However, the levels of serum ferritin were notdifferent in these groups. The second sensitivity analysis was performed based on the NationalKidney Foundation criteria for anemia (anemia is <13.5 g/dl for an adult male and <12 g/dlfor an adult female): 54 patients were classified as anemic. The level of TNF-α (P=0.049) wassignificantly higher and serum albumin levels significantly lower (P=0.005) in these patientswith National Kidney Foundation-defined anemia. There was no significant difference in IL-6and IL-8 levels and serum ferritin levels. In the third and fourth sensitivity analyses, we usedlower hemoglobin values to describe anemia, defining CKD anemia for both males and femalesas ≤12 and ≤11 g/dl, respectively. The level of TNF-α remained significantly elevated in CKDanemic patients, and IL-6 and IL-8 levels were not significantly different in anemic subjects.

We evaluated C-reactive protein (CRP) levels in CKD patients with and without anemia(N=74). CRP levels were divided into tertiles. Tertiles for CRP was based on the distributionof CRP in the sample, as the cut points are less well defined, consistent with the previousliterature.10 The median (interquartile range) levels for CRP were 1.60 mg/l (0.60, 4.32) inanemic patients and 1.19 mg/l (0.46, 3.42). Among anemic subjects, 35.9% were in the thirdtertile whereas 33% were in the first tertile for CRP (Mantel–Haenszel test for linearassociation; P=0.58). Thus, CRP trends were not significantly affected by the presence ofanemia in our CKD cohort (Figure 1a).

We also evaluated IL-10, an important anti-inflammatory cytokine that limits inflammationand is related to IL-6, TNF-α, and CRP. IL-10 levels were 4.14 pg/ml (1.80, 8.68) in anemicpatients and 2.88 pg/ml (1.11, 5.86) in non-anemic patients (P=0.13).

We observed an inverse relationship between hemoglobin levels and inflammatory markerssuch as IL-8 (Spearman’s correlation coefficient (rs)=−0.28; P=0.004), TNF-α (rs=−0.32;P<0.001), IL-6 (rs= −0.17; P=0.09), and CRP levels (rs=−0.19; P=0.09) and positivecorrelation between Hgb levels and serum albumin levels (rs=0.35; P<0.001). In contrast,although there was a significant inverse correlation between GFR and TNF-α (rs=−0.34;P<0.001), other inflammatory marker such as IL-8 (rs= −0.13; P=0.20), IL-6 (rs= −0.003;P=0.97), CRP (rs=−0.15; P=0.19), and serum albumin (rs= −0.02; P=0.83) did not demonstrateany correlation with GFR.

Proinflammatory markers and therapy with ESAsWe examined the potential effect of ESA therapy on proinflammatory markers. The medianlevels of inflammatory markers in patients on ESAs and ESA-naive patients were 6.86 pg/ml(0.64, 16.8) and 2.24 pg/ml (0.64, 7.73) for IL-6; 9.35 pg/ml (4.86, 13.7) and 7.83 pg/ml (2.36,12.98) for IL-8; 4.99 pg/ml (1.95, 7.23) and 3.20 pg/ml (1.87, 6.74) for TNF-α; 4.20 pg/ml(3.90, 4.40) and 4.10 pg/ml (3.90, 4.20) for serum albumin; and 109.5 pg/ml (52, 354.0) and90 pg/ml (51, 227) for serum ferritin levels, respectively. The markers were divided intoquartiles, and the odds ratio (OR) for patients on ESA having an inflammatory biomarker inthe upper two quartiles were compared with ESA-naive and non-anemic patients (Figure 2).In all, 71.4% of the patients on ESAs were in upper two quartiles for IL-6 compared to 51.9%of ESA-naive anemic and 40.4% of non-anemic patients with unadjusted OR of 3.6 (95%confidence interval (CI), 1.2–11.0; P=0.02). Likewise, 76.2% of the patients on ESAs were inthe upper two quartiles of IL-8 compared to 59.3% of ESA-naive anemic and 34.6% of non-anemic patients with unadjusted OR of 6.0 (95% CI, 1.9–19.1; P=0.002). A total of 71.4% ofthe patients on ESAs were in upper two quartiles for TNF-α compared to 55.6% of ESA-naiveanemic and 38.5% of non-anemic patients with unadjusted OR of 4.0 (95% CI, 1.3–12.0;P=0.01). However, ferritin was not significantly different with 60% of the patients on ESAsin upper two quartiles for ferritin compared to 44% of ESA-naive anemic and 48.9% of non-anemic patients with unadjusted OR of 1.5 (95% CI, 0.5–4.5; P=0.40). A total of 61.9% of thepatients on ESAs were in the lower two quartiles for albumin compared to 77.8% of ESA-naive

Keithi-Reddy et al. Page 3

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

anemic and 34.6% of non-anemic patients. It is worth noting that the ESA-naive patients hadhigher OR at 6.0 (95% CI, 1.0–8.7; P=0.001) compared to patients on ESAs whose unadjustedOR was 3.0 (95% CI, 1.0–8.7; P=0.03) for serum albumin.

As described above, CRP levels were classified into tertiles. The analysis revealed that 50%of the patients on ESAs were in the upper tertile of CRP, compared to 33% of patients whowere ESA-naive (P=0.013 by the Mantel–Haenszel’s linear by linear association method), andthe unadjusted OR of patient on ESA having the highest tertile of CRP levels was 4.0 (Figure1b). The median CRP levels in anemic subjects and non-anemic subjects were 1.71 and 1.18mg/l, respectively. The median level of CRP in patients who were ESA-treated vs patients whowere ESA-naive was 2.65 and 0.98 mg/l, respectively.

We performed analyses of anemic subjects who were either treated with ESA or were ESA-naive to explore whether there were differences in IL-10 levels; our data revealed no significantdifference in IL-10 levels. The median IL-10 levels in anemic (by modified WHOclassification) and non-anemic subjects were 4.14 and 2.88 pg/ml, respectively (P=0.13). Themedian levels of IL-10 in ESA-treated and ESA-naive patients were similar at 4.14 pg/ml(P=0.66).

Among the patients treated with ESAs, the duration or the dose of ESA therapy did not correlatewith any of the inflammatory markers.

Determinants of elevated levels of inflammatory markersBy the Spearman’s rank correlation test, IL-8 levels positively correlated with IL-6 (rs=0.63;P<0.001) and TNF-α levels (rs=0.51; P<0.001). In light of a close correlation between thesemarkers, a composite of upper two quartiles of a combination of IL-6, IL-8 and TNF-α wasconsidered as composite outcome variable for multivariate logistic regression. Anemic patientsnot on ESAs had modest OR of reaching the composite outcome variable that did not achievestatistical significance (OR 1.8 (0.57–5.8); P=0.30). Patients on ESAs had a higher OR ofhaving the composite outcome variable (OR 9.1 (95% CI, 1.12–75.3); P=0.03) after weadjusted for age, gender, GFR, hemoglobin, presence of diabetes, hypertension, and otherinflammatory markers such as albumin, ferritin, and platelets (Table 3). We also used othercomposite outcome variables in sensitivity analyses. When a composite of third tertile of IL-6,IL-8, and TNF-α was used as an outcome variable, use of ESAs was associated with anunadjusted OR of 3.4 (95% CI, 1.1–10.1; P=0.02) and an adjusted OR of 3.6 (95% CI, 1.1–11.9; P=0.03); when a composite of upper quartile of IL-6, IL-8, and TNF-α was used as anoutcome variable, unadjusted OR was 3.0 (95% CI, 1.0–8.7; P=0.03) and adjusted OR was 3.1(95% CI, 1.0–9.4; P=0.03) for the association with ESA therapy.

DISCUSSIONIn this study, we explored the inter-relationships between anemia, ESA treatment, and markersof inflammation in CKD patients. We observed that the TNF-α level was higher and the levelsof serum albumin lower in anemic compared to non-anemic subjects. The directionality ofthese findings persisted when sensitivity analyses were performed for various classificationsof anemia. The levels of IL-6, IL-8, IL-1β, CRP, and serum ferritin did not differ, althoughthere were trends for increased IL-6 and IL-8, between anemic and non-anemic patientssuggesting that heightened TNF-α level was unlikely to be explained by a generalized andheightened state of inflammation attributable to renal dysfunction. Indeed, these cytokinesshowed stronger inverse correlation with hemoglobin than GFR. Analysis of anemic patientstreated with ESA compared to ESA-naive patients revealed increased OR for IL-6, IL-8, CRP,and TNF-α with ESA treatment. On multiple logistic regression analysis, ESA-treated anemic

Keithi-Reddy et al. Page 4

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

patients had a strong association for the composite outcome variable of having the upper twoquartiles of IL-6, IL-8, and TNF-α with a multivariate OR of 9.1.

TNF-α was significantly higher and arguably there was a trend for higher IL-6 and IL-8 levelsin our anemic compared to non-anemic subjects with CKD. These cytokines have beenimplicated as non-traditional CVD risk factors associated with an increased risk ofatherosclerosis.11 Indeed, IL-6 and TNF-α stimulate liver production of CRP. Highly sensitiveCRP is clinically used to identify asymptomatic individuals at risk of coronary events.12,13

IL-8 is a potent angiogenic factor that induces migration and proliferation of endothelial cellsand smooth muscle cells contributing to plaque formation in atherosclerosis.14 IL-10 mediatesseveral anti-atherogenic pathways.15 Furthermore, these cytokines are collectively consideredpowerful predictors of carotid atherosclerotic burden and all-cause cardiovascular mortality inboth CKD and non-CKD subjects.16–18 Among these markers, in studies of CKD patients,CRP, IL-6, and serum albumin have been demonstrated to have a strong association withcardiovascular mortality.19,20 Other markers of inflammation in CKD including IL-8, TNF-α, and IL-10 have shown either subtle or equivocal association with CVD. It is well knownthat in CKD patients, inflammation forms a complex with malnutrition and atherosclerosis andportends a poor prognosis.21 One of the hypotheses promulgated to explain increasedinflammation in CKD patients is the reduced clearance of cytokines as the GFR declines.22,23 Ershler24 suggested that increased prevalence of anemia in elderly individuals is due toincreased levels of IL-6 with age. As anemic subjects in our study were older and had lowerGFR than non-anemic subjects, one may consider GFR and age as confounding factors.However, our analysis suggested a potential relationship between heightened levels ofproinflammatory cytokines with anemia in patients with CKD, independent of the effect of ageand renal impairment. Higher TNF-α level and the trend for higher IL-6 and IL-8 levels(positive acute-phase proteins) and a trend for lower serum albumin levels (negative acute-phase protein) in anemic subjects persisted even after sensitivity analyses for differentclassifications of anemia, although, only TNF-α remained statistically significant by allclassifications for anemia. The association between anemia and activation of proinflammatorycytokines has been observed both in patients with cancer and in those patients with autoimmunediseases. In a study by Maccio et al.25, 91 ovarian cancer patients and age-matched 95 healthycontrols were evaluated for relationship between anemia and inflammatory markers. IL-6,TNF-α, CRP, and IL-1β negatively correlated with hemoglobin levels. On multivariateregression analysis, the stage of the disease and the levels of IL-6 were strongly associatedwith lower hemoglobin levels.25,26 Circumstantial evidence from studies on autoimmunediseases points to inflammation related to baselinedisease that then leads to anemia.26 Studiessuggest that inflammatory cytokines interfere with both proliferation and differentiation oferythroid precursors by several mechanisms, including induction of apoptosis, downregulationof ESA receptors, and reduced activity and synthesis of ESAs.27 On the other hand, states ofvascular congestion and chronic heart failures have been shown to have elevated levels of theseinflammatory markers when the renal function declines.28–30 Whether anemia triggers theseinflammatory mediators in patients with CKD by causing subtle vascular congestive statesremains unclear. Of the inflammatory markers, IL-6 is important by virtue of its actions notonly on cardiovascular system but also on worsening of anemia. It induces hepcidin mRNA inhepatocytes leading to iron-deficient erythropoiesis.31 IL-6 also has other effects onerythropoiesis such as inhibition of the development of burst-forming units-erythroid andcolony-forming units-erythroid in bone marrow cultures; the trend seen with CRP levels inanemic subjects may also be important because of the described relationships between CRPand cardiovascular mortality.5,6,18,22,32,33 IL-10 physiologically antagonizes these effects.34

IL-10 showed higher trend in anemic subjects suggesting that anti-inflammatory pathwayswere also affected to negate the effect of proinflammatory cytokines.

Keithi-Reddy et al. Page 5

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

The relation between ESAs and inflammation was also explored in our study, although thenumber of patients that we studied was relatively limited. Although the anti-apoptotic effectof ESAs on erythroid precursors in the bone marrow is well described,35 the potential effectof ESAs on cytokines is less well understood. Long-term administration of ESAs has beenassociated with decreased levels of TNF-α in subjects with CKD.36 Such an anti-inflammatoryeffect of ESAs has been taken advantage of in treating anemia and decreasing disease activityin rheumatoid arthritis.37 On the other hand, erythropoietin has been reported to enhanceinflammation and ischemia-induced neovascularization by increasing the mobilization ofendothelial progenitor cells.38 Recent studies suggest a potential proinflammatory effect forESAs in other disease states.39 ESAs have demonstrated vascular effects through induction ofmonocyte chemoattractant protein-1, which enhances atherogenicity.40,41 Furthermore, ESAsactivate vascular smooth muscle cells, endothelium, and platelets enhancing thrombogenityand loss of vasodilatory potential.40–44 In another study, when 1270 subjects aged 65 years orolder were evaluated, the number of elevated inflammatory markers (CRP, IL-6, IL-1β, andTNF-α) was associated with progressively higher levels of erythropoietin levels in non-anemicnon-CKD participants, independent of age, sex, and hemoglobin. Out of these inflammatorymarkers, IL-6 had a very strong association.45 Although the pro- and anti-inflammatory rolesof ESAs are inconclusive, the responsiveness of the erythroid progenitors to ESAs is bluntedin inflamed severe chronic disease patients due to inflammatory cytokines interfering with theESA–receptor signal transduction pathways.46 As hemoglobin levels are inversely correlatedwith these cytokines, one would expect reduced levels of these cytokines with the use of ESAs,as ESA therapy increases hemoglobin levels. However, in our study, the use of ESAs wasassociated with increased OR for the upper two quartiles of IL-6, IL-8, and TNF-α and theupper tertile for CRP levels. The mechanism for ESA-associated increase in cytokine levels isnot clear. One possibility is that of confounding by indication, that is, severely inflamed patientswere more severely anemic and therefore required ESA therapy. An alternative hypothesis isthat induction of proinflammatory cytokines by ESAs occurs as a consequence of ESAs actingvia erythropoetin receptors on macrophages (or other cell types expressing these cytokines),which on activation secrete IL-6, IL-8, and TNF-α. If increased synthesis of these cytokines isdemonstrated to result from ESA treatment, this could be important as these cytokines aremajor acute-phase reactant proteins and markers of the malnutrition–inflammation–cachexiasyndrome. Indeed, heightened IL-6, IL-8, and TNF-α levels have been associated with anincreased risk of death in the dialysis population.47,48 Furthermore, this could be oneexplanation for the observation of increased risk for all-cause and cardiovascular mortality inpatients targeted to higher hemoglobin levels with higher doses of ESAs.2,3,49 In our study,we also observed that IL-10 was not significantly elevated in subjects on ESAs, perhapssuggesting that inflammation counter-regulatory mechanisms were not significantly triggeredin patients on ESAs.

Our study had several strengths. To our knowledge, this study is the first to describe theassociation and interaction of ESAs, anemia, and inflammatory cytokines in the setting of CKD.Although our study does not indicate causality, we believe that, conceptually, it is hypothesis-generating. Our study evaluated different definitions of anemia, thus enhancing its internalvalidity. Finally, the assays were performed using defined validated techniques with limitedintra-assay variability. An important limitation to this study is its observational design and thatwe studied prevalent patients. The effect of unmeasured confounders should limit inferencesregarding causality. The issue of confounding by indication also requires us to be cautious inthe interpretation of the data; indeed, our observation on associations between proinflammatorycytokines and anemia and ESAs requires further exploration and confirmation by longitudinalprospective studies.

In summary, our data suggest an association between anemia and increased levels of theproinflammatory cytokine TNF-α. We also observed an association between higher levels of

Keithi-Reddy et al. Page 6

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

proinflammatory cytokines and ESA treatment in our anemic patients, raising a possibleexplanation for why ESA therapy may be associated with adverse effects under certaincircumstances.

PATIENTS AND METHODSThe Brigham CKD cohort is a prospective observational cohort established in 2004 that isdesigned to study the influence of specific modifiable factors on the risk of progression of CKDand the risk of CVD in patients with CKD. Inclusion criteria include patients aged ≥18 yearsand CKD defined as GFR < 60 ml/min per 1.73m2 or GFR≥ 60 ml/min per 1.73m2 withalbuminuria. eGFR was calculated by the Modification of Diet in Renal Disease II equation(eGFR (ml/min per 1.73m2)=186 × [Cr (mg/100 ml)]−1.154 × (age)−0.203 × (1.21 if subject isblack)). Exclusion criteria included unwillingness to provide informed consent, patients whowere hospitalized or had a thrombotic event or cardiovascular event 1 month prior toenrollment, institutionalized subjects (nursing home resident, skilled nursing facility resident,prisoner), those with a life expectancy ≤3 years, known human immunodeficiency virus patient,an active malignant neoplastic disease other than localized non-melanoma skin cancer, patientson immunotherapy or immunosuppressive treatment for a primary or secondary renal disease,a previous diagnosis of multiple myeloma, and those patients enrolled in an intervention study.One hundred patients out of 142 who had blood samples were included for the analysis. Thestudy was approved by the institutional review board at Partners Healthcare and an informedconsent was taken from each patient. Anemia was defined as Hgb<12 g/dl in women andHgb<13 g/dl in men on treatment with ESAs; modification of WHO criteria. For testing ouranalysis, which tests the effect of ESAs on inflammation in anemia, the WHO definition foranemia was used. Treatment with ESAs and the dose utilized were ascertained by review ofthe longitudinal medical record, direct contact with the patient, and reviewing the pharmacyrecord.

Biochemical analysisPlasma was stored at −80°C until analysis. The multiplex human CVDs biomarkers panels 1and 3 were used (HCVD1-67AK-4Plex, HCVD3-67CK-6Plex; Linco Inc. (Milliporecorporation, Chicago, IL, USA) part of Millipore, MA, USA) for the simultaneousquantification of the several analytes. Of these, inflammatory markers include IL-1β, IL-6,IL-8, TNF-α, and IL-10. All examined analytes were tested individually and in combinationto ensure that there were no crossreactions. All measurements were performed in duplicate andthe investigators were blinded to clinical characteristics of the patients being studied. Briefly,the multibiomarker and cytokine standards were resuspended in provided assay buffer and thendifferently serially diluted. A total of 25 µl of reconstituted standard, Quality Controls, orsample (HCVD-1 required a 1:50 serum dilution) was added to each well of a 96-well platewith 25 µl of the bead suspension. The plate was sealed, covered with aluminum foil, andincubated overnight (16–18 h) with agitation on a plate shaker at 4°C. Then the plate waswashed twice with 200 µl/well of wash buffer, removing buffer by vacuum filtration(<100mmHg) between each wash. This was followed by addition of 25 µl of a detectionantibody cocktail into each well and incubation at room temperature for 60 min. A total of 25µl of a streptavidin–phycoerythrin solution was added to each well and incubated at roomtemperature for 30 min. The plate was then analyzed on the Luminex IS100 analyzer (LuminexInc., Austin, TX, USA). The data were saved and evaluated as median fluorescence intensityusing appropriate curve-fitting software (Luminex 100IS software version 2.3). A five-parameter logistic method with weighting was used.

C-reactive protein in blood was determined by a highly sensitive latex-based immunoassay(Dade Behring, Newark, DE, USA) in collaboration with Paul Ridker/Nader Rifai laboratories

Keithi-Reddy et al. Page 7

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

at the Children’s hospital, Harvard Medical School, Boston. The concentration of CRP wasdetermined using an immunoturbidimetric assay on the Hitachi 917 analyzer (RocheDiagnostics, Indianapolis, IN, USA), using reagents and calibrators from DiaSorin (Stillwater,MN, USA). In this assay, an antigen–antibody reaction occurs between CRP in the sample andan anti-CRP antibody that has been sensitized to latex particles, and agglutination results. Thisantigen–antibody complex causes an increase in light scattering, which is detectedspectrophotometrically, with the magnitude of the change being proportional to theconcentration of CRP in the sample. This assay has a sensitivity of 0.03 mg/l. The day-to-dayvariabilities of the assay at concentrations of 0.91, 3.07, and 13.38 mg/l are 2.81, 1.61, and1.1%, respectively. This method has been shown efficacious in predicting risk of recurrentcoronary events.33

Statistical analysisContinuous data were analyzed with independent sample t-test, and categorical data wereanalyzed with χ2 statistic in the analysis of demographic characteristics. The levels of IL-6,IL-8, TNF-α, IL-10, serum ferritin, albumin, and highly sensitive CRP levels deviated fromnormal distribution and hence we used Wilcoxon rank-sum test (also called Mann–WhitneyU test) to compare anemic and non-anemic subjects and subjects treated with ESAs and epo-naive. Spearman’s correlation coefficients were calculated for pairs of continuous variables.Logistic regression was used to calculate OR for reaching a composite end point of upper twoquartiles (third and fourth quartile vs first and second quartiles) of IL-6 or IL-8 or TNF-α.Mantel–Haenszel linear by linear association was used to assess the inflammatory markers intertiles or quartiles. Bar diagrams were used to represent upper two and lower two quartiles ofTNF-α, IL-6, IL-8, albumin, and ferritin, and unadjusted OR were calculated for the effect ofESAs. SPSS version 15 (SPSS Inc., Chicago, IL, USA) was used to analyze the data.

ACKNOWLEDGMENTSS.R.K. has received support from the International Society of Nephrology. Studies were also supported in part by NIHgrants DK54602 and DK45462 (M.S.G.). We are also highly indebted to Dr Paul Ridker for helping in the analysisof CRP levels. We are thankful to Pierre Debrosse, Patricia Pankievich, Paula Hertello and Natasha Tyagi formaintaining the CKD cohort.

REFERENCES1. Jurkovitz CT, Abramson JL, Vaccarino LV, et al. Association of high serum creatinine and anemia

increases the risk of coronary events: results from the prospective community-based atherosclerosisrisk in communities (ARIC) study. J Am Soc Nephrol 2003;14:2919–2925. [PubMed: 14569102]

2. Singh AK, Szczech L, Tang KL, et al. Correction of anemia with epoetin alfa in chronic kidney disease.N Engl J Med 2006;355:2085–2098. [PubMed: 17108343]

3. Drueke TB, Locatelli F, Clyne N, et al. Normalization of hemoglobin level in patients with chronickidney disease and anemia. N Engl J Med 2006;355:2071–2084. [PubMed: 17108342]

4. Smith KJ, Bleyer AJ, Little WC, et al. The cardiovascular effects of erythropoietin. Cardiovasc Res2003;59:538–548. [PubMed: 14499855]

5. Ridker PM, Cook NR. Biomarkers for prediction of cardiovascular events. N Engl J Med2007;356:1472–1473. [PubMed: 17415901]author reply 1474–1475

6. Ridker PM. C-reactive protein and the prediction of cardiovascular events among those at intermediaterisk: moving an inflammatory hypothesis toward consensus. J Am Coll Cardiol 2007;49:2129–2138.[PubMed: 17531663]

7. Tonelli M, Sacks F, Pfeffer M, et al. Biomarkers of inflammation and progression of chronic kidneydisease. Kidney Int 2005;68:237–245. [PubMed: 15954913]

8. Lonnemann G, Engler-Blum G, Muller GA, et al. Cytokines in human renal interstitial fibrosis. II.Intrinsic interleukin (IL)-1 synthesis and IL-1-dependent production of IL-6 and IL-8 by culturedkidney fibroblasts. Kidney Int 1995;47:845–854. [PubMed: 7752583]

Keithi-Reddy et al. Page 8

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

9. Lonnemann G, Shapiro L, Engler-Blum G, et al. Cytokines in human renal interstitial fibrosis. I.Interleukin-1 is a paracrine growth factor for cultured fibrosis-derived kidney fibroblasts. Kidney Int1995;47:837–844. [PubMed: 7752582]

10. Pradhan AD, Cook NR, Buring JE, et al. C-reactive protein is independently associated with fastinginsulin in nondiabetic women. Arterioscl Thromb Vasc Biol 2003;23:650–655. [PubMed: 12692007]

11. Haddy N, Sass C, Droesch S, et al. IL-6, TNF-alpha and atherosclerosis risk indicators in a healthyfamily population: the STANISLAS cohort. Atherosclerosis 2003;170:277–283. [PubMed:14612208]

12. Libby P, Ridker PM, Maseri A. Inflammation and atherosclerosis. Circulation 2002;105:1135–1143.[PubMed: 11877368]

13. Tousoulis D, Charakida M, Stefanadis C. Endothelial function and inflammation in coronary arterydisease. Heart 2006;92:441–444. [PubMed: 16159981]

14. Simonini A, Moscucci M, Muller DW, et al. IL-8 is an angiogenic factor in human coronaryatherectomy tissue. Circulation 2000;101:1519–1526. [PubMed: 10747344]

15. Schonbeck U, Libby P. Inflammation, immunity, and HMG-CoA reductase inhibitors: statins asantiinflammatory agents? Circulation 2004;109:II18–II26. [PubMed: 15173059]

16. Addabbo F, Mallamaci F, Leonardis D, et al. Searching for biomarker patterns characterizing carotidatherosclerotic burden in patients with reduced renal function. Nephrol Dial Transplant2007;22:3521–3526. [PubMed: 17636056]

17. Iseki K, Tozawa M, Yoshi S, et al. Serum C-reactive protein (CRP) and risk of death in chronicdialysis patients. Nephrol Dial Transplant 1999;14:1956–1960. [PubMed: 10462277]

18. Ridker PM, Hennekens CH, Buring JE, et al. C-reactive protein and other markers of inflammationin the prediction of cardiovascular disease in women. N Engl J Med 2000;342:836–843. [PubMed:10733371]

19. Yeun JY, Levine RA, Mantadilok V, et al. C-Reactive protein predicts all-cause and cardiovascularmortality in hemodialysis patients. Am J Kidney Dis 2000;35:469–476. [PubMed: 10692273]

20. Kalantar-Zadeh K. Inflammatory marker mania in chronic kidney disease: pentraxins at the crossroadof universal soldiers of inflammation. Clin J Am Soc Nephrol 2007;2:872–875. [PubMed: 17702716]

21. Kalantar-Zadeh K, Kopple JD. Relative contributions of nutrition and inflammation to clinicaloutcome in dialysis patients. Am J Kidney Dis 2001;38:1343–1350. [PubMed: 11728973]

22. Panichi V, Migliori M, De Pietro S, et al. C reactive protein in patients with chronic renal diseases.Ren Fail 2001;23:551–562. [PubMed: 11499569]

23. Bemelmans MH, Gouma DJ, Buurman WA. Influence of nephrectomy on tumor necrosis factorclearance in a murine model. J Immunol 1993;150:2007–2017. [PubMed: 8436831]

24. Ershler WB. Biological interactions of aging and anemia: a focus on cytokines. J Am Geriatr Soc2003;51:S18–S21. [PubMed: 12588568]

25. Maccio A, Madeddu C, Massa D, et al. Hemoglobin levels correlate with interleukin-6 levels inpatients with advanced untreated epithelial ovarian cancer: role of inflammation in cancer-relatedanemia. Blood 2005;106:362–367. [PubMed: 15774616]

26. Davis D, Charles PJ, Potter A, et al. Anaemia of chronic disease in rheumatoid arthritis: in vivo effectsof tumour necrosis factor alpha blockade. Br J Rheumatol 1997;36:950–956. [PubMed: 9376990]

27. Weiss G, Goodnough LT. Anemia of chronic disease. N Engl J Med 2005;352:1011–1023. [PubMed:15758012]

28. Sato Y, Takatsu Y, Kataoka K, et al. Serial circulating concentrations of C-reactive protein, interleukin(IL)-4, and IL-6 in patients with acute left heart decompensation. Clin Cardiol 1999;22:811–813.[PubMed: 10626084]

29. Hasper D, Hummel M, Kleber FX, et al. Systemic inflammation in patients with heart failure. EurHeart J 1998;19:761–765. [PubMed: 9717010]

30. Yilmaz MI, Carrero JJ, Axelsson J, et al. Low-grade inflammation in chronic kidney disease patientsbefore the start of renal replacement therapy: sources and consequences. Clin Nephrol 2007;68:1–9.[PubMed: 17703829]

31. Nemeth E, Valore EV, Territo M, et al. Hepcidin, a putative mediator of anemia of inflammation, isa type II acute-phase protein. Blood 2003;101:2461–2463. [PubMed: 12433676]

Keithi-Reddy et al. Page 9

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

32. Pearson TA, Mensah GA, Alexander RW, et al. Markers of inflammation and cardiovascular disease:application to clinical and public health practice: a statement for healthcare professionals from theCenters for Disease Control and Prevention and the American Heart Association. Circulation2003;107:499–511. [PubMed: 12551878]

33. Rifai N, Tracy RP, Ridker PM. Clinical efficacy of an automated high-sensitivity C-reactive proteinassay. Clin Chem 1999;45:2136–2141. [PubMed: 10585345]

34. Voulgari PV, Kolios G, Papadopoulos GK, et al. Role of cytokines in the pathogenesis of anemia ofchronic disease in rheumatoid arthritis. Clin Immunol 1999;92:153–160. [PubMed: 10444359]

35. Krantz SB. Erythropoietin. Blood 1991;77:419–434. [PubMed: 1991159]36. Ludwig H, Fritz E, Kotzmann H, et al. Erythropoietin treatment of anemia associated with multiple

myeloma. N Engl J Med 1990;322:1693–1699. [PubMed: 2342535]37. Kaltwasser JP, Kessler U, Gottschalk R, et al. Effect of recombinant human erythropoietin and

intravenous iron on anemia and disease activity in rheumatoid arthritis. J Rheumatol 2001;28:2430–2436. [PubMed: 11708414]

38. Heeschen C, Aicher A, Lehmann R, et al. Erythropoietin is a potent physiologic stimulus forendothelial progenitor cell mobilization. Blood 2003;102:1340–1346. [PubMed: 12702503]

39. Swaminathan S, Ahmed I, McCarthy JT, et al. Nephrogenic fibrosing dermopathy and high-doseerythropoietin therapy. Ann Intern Med 2006;145:234–235. [PubMed: 16880475]

40. De Marchi S, Cecchin E, Falleti E, et al. Long-term effects of erythropoietin therapy on fistula stenosisand plasma concentrations of PDGF and MCP-1 in hemodialysis patients. J Am Soc Nephrol1997;8:1147–1156. [PubMed: 9219165]

41. Desai A, Zhao Y, Lankford HA, et al. Nitric oxide suppresses EPO-induced monocyte chemoattractantprotein-1 in endothelial cells: implications for atherogenesis in chronic renal disease. Lab Invest2006;86:369–379. [PubMed: 16482103]

42. Stohlawetz PJ, Dzirlo L, Hergovich N, et al. Effects of erythropoietin on platelet reactivity andthrombopoiesis in humans. Blood 2000;95:2983–2989. [PubMed: 10779449]

43. Anagnostou A, Lee ES, Kessimian N, et al. Erythropoietin has a mitogenic and positive chemotacticeffect on endothelial cells. Proc Natl Acad Sci USA 1990;87:5978–5982. [PubMed: 2165612]

44. Ammarguellat F, Llovera M, Kelly PA, et al. Low doses of EPO activate MAP kinases but not JAK2-STAT5 in rat vascular smooth muscle cells. Biochem Biophys Res Commun 2001;284:1031–1038.[PubMed: 11409898]

45. Ferrucci L, Guralnik JM, Woodman RC, et al. Proinflammatory state and circulating erythropoietinin persons with and without anemia. Am J Med 2005;118:1288. [PubMed: 16271918]

46. Jelkmann W. Proinflammatory cytokines lowering erythropoietin production. J Interferon CytokineRes 1998;18:555–559. [PubMed: 9726435]

47. Stenvinkel P, Ketteler M, Johnson RJ, et al. IL-10, IL-6, and TNF-alpha: central factors in the alteredcytokine network of uremia—the good, the bad, and the ugly. K Int 2005;67:1216–1233.

48. Gauldie J, Richards C, Harnish D, et al. Interferon beta 2/B-cell stimulatory factor type 2 sharesidentity with monocyte-derived hepatocyte-stimulating factor and regulates the major acute phaseprotein response in liver cells. Proc Ntl Acad Sci USA 1987;84:7251–7255.

49. Besarab A, Bolton WK, Browne JK, et al. The effects of normal as compared with low hematocritvalues in patients with cardiac disease who are receiving hemodialysis and epoetin. N Engl J Med1998;339:584–590. [PubMed: 9718377]

Keithi-Reddy et al. Page 10

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 1. Number of anemic patients and patients on ESAs by tertiles of CRP in CKD cohortAmong anemic subjects, 35.9% were in the third tertile whereas 33% were in the first tertilefor CRP (Mantel–Haenszel test for linear association; P=0.58) (a). Fifty percentage of theanemic subjects on ESAs were in the third tertile whereas 33% of epo-naive patients were inthe third tertile for CRP (Mantel–Haenszel test for linear association; P=0.013). Patients in thefirst and third tertiles were compared for P-value (b). CKD, chronic kidney disease; CRP, C-reactive protein; ESAs, erythropoiesis-stimulating agents.

Keithi-Reddy et al. Page 11

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 2. Odds ratios for upper two quartiles for patients treated with ESAsIn all, 71.4% of the patients on ESAs were in upper two quartiles for IL-6 compared to 51.9%of ESA-naive anemic and 40.4% of non-anemic patients with unadjusted odds of 3.6 (a). Inall, 76.2% of the patients on ESAs were in upper two quartiles for IL-8 compared to 59.3% ofESA-naive anemic and 34.6% of non-anemic patients with unadjusted odds of 6.0 (b). In total,71.4% of the patients on ESAs were in upper two quartiles for TNF-α compared to 55.6% ofESA-naive anemic and 38.5% of non-anemic patients with unadjusted odds of 4.0 (c). Sixtypercentage of the patients on ESAs were in the upper two quartiles for Ferritin compared to44% of ESA-naive anemic and 48.9% of non-anemic patients with unadjusted odds of 1.5(d). A total of 61.9% of the patients on ESAs were in the lower two quartiles for albumin

Keithi-Reddy et al. Page 12

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

compared to 77.8% of ESA-naive anemic and 34.6% of non-anemic patients with unadjustedodds of 6.0 (e). Odds ratios are for the use of ESAs. *Lower two and upper two quartiles arerepresented in reverse order for the convenience of readers. ESAs, erythropoiesis-stimulatingagents; IL, interleukin; TNF-α, tumor necrosis factor-α.

Keithi-Reddy et al. Page 13

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Keithi-Reddy et al. Page 14

Table 1Baseline characteristics of CKD patients stratified by anemic statusa

Characteristic Anemic CKD patients(N=50)

Non-anemic CKDpatients (N=50)

P-value

Age in years 64.2 ± 14.8 53.6 ± 17.2 0.002*

Gender (M/F) 64:36 32:68 0.001*

Race 0.73

White 56.0 54

Black 28 20

Hispanic 10 18

Asian 4 6

Others 2.0 2

Etiology of CKDb 0.43

Diabetic nephropathy 38 24

Hypertension 24 20

Chronic glomerulonephritis 12 24

Chronic interstitial nephritis 10 8

Lupus nephritis 12 16

Others 4 8

Diabetes 38 30 0.26

Hypertension 86 72 0.07

History of smoking 8.2 7 0.23

Serum creatinine (mg/dl) 2.7 ± 1.5 1.7 ± 1.0 <0.001*

MDRD GFR (ml/min per 1.73m2)c 31.4 ± 19.6 53.8 ± 31.6 <0.001*

Platelets 275.8 ± 64.6 231.7 ± 94.8 0.008*

Hemoglobin (g/dl) 11.2 ± 1.2 13.8 ± 1.1 <0.001*

TSAT in % 23.9 ± 9.5 28.1 ± 13.3 0.08

Random blood sugar (mg/dl) 105.4 ± 47.3 100.3 ± 33.0 0.54

Mean duration of ESAs (in months) 6.8 ± 12.4 —

(N=23)

Median dose of ESAs (U/kg)d 8000 (N=23) —

CKD, chronic kidney disease; ESAs, erythropoiesis-stimulating agents; GFR, glomerular filtration rate; MDRD, Modification of Diet in Renal Disease.

All continuous data expressed as mean ± s.d.

Categorical data expressed in percentages.

*Significant if P-value is <0.05.

TSAT refers to percentage of transferrin saturation.

aModified WHO classification of anemia is used where anemia is defined as presence of Hgb<12g/100ml in women or Hgb<13g/100ml in men on

treatment with ESAs.

bCKD defined as GFR<60 ml/min per 1.73m2 or GFR≥60 ml/min per 1.73m2 with albuminuria.

cMDRD GFR refers to glomerular filtration rate equation measured by modification of diet in renal disease formula type II: eGFR (ml/min per 1.73m2)

=186 × [Cr (mg/100 ml)]−1.154 × (age)−0.203 × (1.21 if subject is black).

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Keithi-Reddy et al. Page 15

dOf ESA-treated patients, 57% (N=13) were on epoetin-α (median dose of 10, 000 U/week) and 43% (N=10) on darbepoetin (median dose of 7000 U/

week; a conversion of 300 U for each microgram of darbepoetin-α was done).

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Keithi-Reddy et al. Page 16Ta

ble

2In

flam

mat

ory

mar

kers

by

anem

ic st

atus

: sen

sitiv

ity a

naly

sis

Mar

kera

P-va

lue

by M

od.

WH

O c

rite

riab

Ane

mic

(N=5

0)

P-va

lue

by W

HO

crite

riac

Ane

mic

(N=4

8)

P-va

lue

by N

KF

crite

riad

Ane

mic

(N=5

4)

P-va

lue

for

the

cut p

oint

Hgb

<12g

/dl

Ane

mic

(N=3

8)

P-va

lue

for

the

cut p

oint

Hgb

<11g

/dl

Ane

mic

(N=1

5)

IL-6

0.12

0.11

0.28

0.39

0.52

IL-8

0.05

0.01

3*0.

070.

050.

07

TNF-α

0.01

9*0.

016*

0.04

9*0.

016*

0.00

8*

Seru

m fe

rriti

n0.

560.

390.

580.

690.

41

Seru

m a

lbum

in0.

002*

<0.0

01*

0.00

5*0.

001*

0.32

ESA

s, er

ythr

opoi

esis

-stim

ulat

ing

agen

ts; I

L, in

terle

ukin

.

Man

n–W

hitn

ey U

test

was

don

e fo

r the

ana

lysi

s with

leve

ls o

f inf

lam

mat

ory

mar

kers

as d

epen

dent

var

iabl

es.

* Sign

ifica

nt.

a IL-1β

was

not

sign

ifica

ntly

ele

vate

d in

bot

h th

e gr

oups

. All

the

grou

ps h

ad v

alue

s <0.

64 p

g/m

l.

b Mod

ified

WH

O d

efin

ition

of a

nem

ia is

use

d he

re w

here

ane

mia

is d

efin

ed a

s pre

senc

e of

Hgb

<12g

/dl i

n w

omen

or H

gb<1

3g/d

l in

men

or t

reat

men

t with

ESA

s.

c WH

O d

efin

ition

of a

nem

ia: H

gb<1

3g/d

l for

adu

lt m

ale

and

<12g

/dl f

or a

n ad

ult f

emal

e.

d NK

F de

finiti

on o

f ane

mia

: Hgb

<13.

5g/d

l for

adu

lt m

ale

and

<12g

/dl f

or a

n ad

ult f

emal

e.

Kidney Int. Author manuscript; available in PMC 2009 September 8.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Keithi-Reddy et al. Page 17Ta

ble

3O

dds r

atio

for c

ompo

site

end

poi

nt o

f IL-

6 or

IL-8

or T

NF-α

in th

e up

per t

wo

quar

tiles

Para

met

erU

nadj

uste

d O

R (9

5% C

I)P-

valu

eA

djus

ted

OR

(95%

CI)

P-va

lue

Patie

nts n

ot o

n ES

A n

on-a

nem

icR

efer

ence

—R

efer

ence

—

1Pa

tient

s not

on

ESA

ane

mic

2.3

(0.7

5–7.

1)0.

141.

8 (0

.57–

5.8)

0.30

2Pa

tient

s on

ESA

s10

.5 (1

.3–8

5.4)

0.02

*9.

1 (1

.12–

75.3

)0.

03*

3A

ge1.

01 (0

.99–

1.04

)0.

211.

01 (0

.98–

1.05

)0.

41

4M

ale

gend

er1.

75 (0

.68–

4.5)

0.24

1.4

(0.3

8–5.

39)

0.79

5D

iabe

tes

1.04

(0.3

9–2.

7)0.

930.

62 (0

.17–

2.2)

0.47

6H

yper

tens

ion

0.98

(0.3

1–3.

0)0.

980.

51 (0

.11–

2.3)

0.39

6H

emog

lobi

n<11

g/10

0 m

l6.

1 (0

.76–

48.9

)0.

082.

2 (0

.20–

23.9

)0.

50

7G

FR0.

98 (0

.97–

1.00

)0.

110.

99 (0

.97–

1.01

)0.

35

8A

lbum

in0.

61 (0

.19–

1.8)

0.39

0.68

(0.2

0–2.

28)

0.53

9Fe

rriti

n1.

0 (0

.99–

1.1)

0.95

0.99

(0.9

9–1.

00)

0.98

10Pl

atel

ets

0.99

(0.9

9–1.

00)

0.68

1.00

(0.9

9–1.

01)

0.36

CI,

conf

iden

ce in

terv

al; E

SAs,

eryt

hrop

oies

is-s

timul

atin

g ag

ents

; GFR

, glo

mer

ular

filtr

atio

n ra

te; I

L, in

terle

ukin

; OR

, odd

s rat

io.

Mul

tivar

iate

logi

stic

regr

essi

on w

as d

one

by st

ep-w

ise

back

war

d el

imin

atio

n by

log-

likel

ihoo

d ra

tio (L

R) a

nd c

onfir

med

by

forw

ard

LR.

* P<0.

05.

Kidney Int. Author manuscript; available in PMC 2009 September 8.