Postepy Hig Med Dosw. (online), 2013; 67: 1076-1082
Original Article
Full Text PDF  

Different mortality predictor pattern in hemodialysis and peritoneal dialysis diabetic patients in 4-year prospective observation
Odmienne czynniki zagrożenia śmiertelnością u chorych na cukrzycę leczonych hemodializą i dializą otrzewnową w 4-letniej prospektywnej obserwacji
Katarzyna Madziarska1  ABCDEF, Waclaw Weyde2  ADE, Jozef Penar1  BF, Ewa Zukowska-Szczechowska3  B, Magdalena Krajewska1  BF, Tomasz Golebiowski1  B, Renata Klak1  B, Slawomir Zmonarski1  B, Cyprian Kozyra4  C, Marian Klinger1  ADE
1Department of Nephrology and Transplantation Medicine, Wroclaw Medical University, Wroclaw, Poland
2aculty of Dentistry, Wroclaw Medical University, Wroclaw, Poland
3Internal Medicine, Diabetology and Nephrology, Medical University of Silesia, Zabrze, Poland
4Department of Statistics, Wroclaw University of Economics, Wroclaw, Poland
Corresponding author
Katarzyna Madziarska, MD, PhD, Department of Nephrology and Transplantation Medicine, Wroclaw Medical University, 213 Borowska St., 50-556 Wroclaw, Poland; e-mail: kmadziarska@wp.pl

Authors' Contribution:
A - Study Design, B - Data Collection, C - Statistical Analysis, D - Data Interpretation, E - Manuscript Preparation, F - Literature Search, G - Funds Collection

Received:  2013.06.24
Accepted:  2013.10.07
Published:  2013.11.20

Summary
Introduction: The aim was to identify factors carrying an ominous prognosis in a cohort of diabetic patients (pts) on a hemodialysis (HD) and peritoneal dialysis (PD) program.
Materials and methods: We analyzed survival rates of 61 diabetic dialysis pts (35 HD/26 PD). The participants were matched in baseline characteristics, standard indicators of dialysis care and laboratory para­meters. The studied group was prospectively observed up to 4 years.
Results: 21 pts (34.4%) survived the whole observation period. The annual mortality rate was 23.2%, with no difference between HD and PD. Irrespective of dialysis modality, the only factor associated with mortality in the Cox proportional hazard model was serum albumin lowe­ring. Referring to dialysis modality, the HD survivors were characterized by lower IL-6 level, higher albumin concentration, and increased serum cholesterol values with higher chole­sterol left in multivariate analysis; under PD therapy the only factor significantly associa­ted with mortality was older age. In contrast to HD treatment, elevated cholesterol was a universal finding in PD patients, significantly above levels in HD, with a slight tendency to lower values in PD survivors.
Conclusions:
1. A difference in mortality predictor pattern appeared in diabetic patients treated by PD and HD.
2. In the PD group more advanced age had a decisive negative impact on survival whereas in the HD group the outlook was dependent on factors related to nutrition and inflammation.
3. Elevated cholesterol level was associated with survival benefit in HD patients, being a com­mon abnormality in the PD group, without positive prognostic significance.
Key words: diabetic patients • hemodialysis • peritoneal dialysis • survival




Abbreviations:
BMI - body mass index, BP - blood pressure, CRP - C-reactive protein, Hb - hemoglobin, HD - hemodialysis, IL-6 - interleukin 6, Kt/V - adequacy of dialysis, NT-proBNP - N-terminal pro­-B-type natriuretic peptide, PD - peritoneal dialysis, pro-ANP - serum pro-atrial natriuretic peptide.
Introduction
The number of patients requiring renal replacement therapy is growing continuously.
According to the last report from the ERA-EDTA Registry, the annual incidence of patients starting dialysis in the year 2008 was 122 per million population (pmp), with a significantly increasing proportion of elderly patients, mean age 64, and accompanied by a pronounced rise in the proportion of pa­tients with diabetes, up to one third. These two tendencies, ageing and marked increase of type 2 diabetic patients, are responsible for ominous prognosis in dialysis programs. The five-year survival in patients with type 2 diabetes on dialy­sis was around 30% compared to 46% in non-diabetics [13]. Therefore, the aim of the current study was to identify fac­tors exhibiting a marked impact on survival in a high risk cohort of 61 prevalent dialysis patients with diabetes melli­tus. The included patients were prospectively observed for 4 years. An additional goal was to reveal possible differences in the efficacy of hemodialysis (HD) and peritoneal dialysis (PD) treatment, because both dialysis modalities were nearly equally distributed.
Materials and Methods
Study population
The study group was created from 61 prevalent diabetic dia­lysis patients who had been on a chronic HD or PD program. The patients were recruited from three dialysis centers in south-west Poland (Wroclaw, Zabrze, Walbrzych).
The selected patients carried pronounced cardiovascular risk - all patients with diabetes above the age of 40 with ad­vanced Monckeberg type intravascular calcifications in the forearm arteries (confirmed by X-ray); 26 patients (42.62%) were above the age of 70.
The group was formed in 2006 and prospectively observed during the subsequent 4 years.
The patients included in the study were free of active in­fection, symptomatic coronary disease, overt heart failure, history of malignancy, and diseases requiring immunosup­pressive treatment. Clinical data of patients were extracted from the hospital records.
Study measurements
The impact of the following factors on survival was tested: ba­seline characteristics (age, gender), residual diuresis, duration of dialysis, all standard indicators of dialysis care (blood pres­sure, Hb, body mass index (BMI), adequacy of dialysis - Kt/V) and the following laboratory parameters: serum pro-atrial natriuretic peptide (pro-ANP), serum N-terminal pro-B-type natriuretic peptide (NT-proBNP), serum C-reactive protein (CRP), interleukin 6 (IL-6), serum albumin and cholesterol.
All HD patients were dialyzed using a native arteriovenous fistula fulfilling a single-pool Kt/V >= 1.3. All PD patients were treated by continuous ambulatory peritoneal dialysis (CAPD) achieving weekly Kt/V > 1.7. Kt/V was calculated from three consecutive measurements at monthly intervals.
Laboratory methods
Routine laboratory tests (Hb [g/dl], CRP [mg/l], serum al­bumin [g/dl], and cholesterol [mmol/l]) were measured in the Central Hospital Laboratory as part of the standard care.
In addition, serum pro-ANP (amino terminal 1-98 ANP frag­ment) and serum NT-proBNP were assessed by ELISA (BIO­MEDICA, Vienna, Austria); interleukin-6 (IL-6) was measured by ELISA (R&D Systems, Minneapolis, USA).
The blood samples were taken in HD patients before the mi­dweek dialysis session, and in PD patients during a control visit at the outpatient clinic during morning hours before the first fluid exchange. Blood pressure measurements were performed three times with a standard sphygmomanome­ter, and the mean value from the last two readings was used for calculations.
Statistical analysis
The statistical analysis was performed with Statistica 9.1 software.
The statistical test to compare means of qualitative varia­bles in two groups was the nonparametric Mann-Whitney U test. Survival analysis was conducted using the multiva­riate Cox proportional hazard model to investigate the in­fluence of qualitative variables on risk of death. Backward stepwise removal of non-significant variables was used in multivariate Cox regression, which could result in a univa­riate regression model.
Statistical significance was recognized with a p-value less than 0.05. For quantitative variables, results are given as mean+SD.
Results
We analyzed data of 61 diabetic patients on dialysis (24 F; 37 M - all Caucasian), median age 68, who had been on mainte­nance dialysis at least 2 months (median period 17 months), with preserved median residual diuresis 500 ml/24 h.
Demographic, clinical and laboratory parameters of the in­vestigated participants at the study entry are presented in Table 1. In the investigated cohort 14 patients had type 1 diabetes and 47 type 2; the proportion was 7/28 in the HD group, and 7/19 in the PD group. The data are shown sepa­rately for the HD (35) and PD (26) groups. There were two significant differences between the groups at the start of the study: significantly lower albumin level (p=0.024) and signifi­cantly higher cholesterol concentration (p<0.0001). In the search for the possible cause of these distinctions the records from dialysis initiation were checked. The albumin concen­tration was at dialysis onset essentially similar in both groups (3.54±0.32 g/dl in PD versus 3.56±0.41 g/dl in HD), whereas cholesterol values were significantly higher in PD patients (5.56±0.85 mmol/l versus 4.77±1.20 mmol/l - in HD; p=0.003).
Table 1. Clinical characteristics of study groups

Twenty-one patients from the initial cohort of 61 (34.4%) survived the whole 48-month observation period: 12 HD patients (34.3%) and 9 PD patients (34.6%).
Table 2 contains the clinical characteristics of the de­ceased and surviving patients treated by HD. The sur­vivors were characterized by a lower IL-6 level (p=0.04), higher albumin concentration (p=0.03), and increased cholesterol (p=0.004).
Table 2. Clinical characteristics of deceased and surviving patients treated by HD

The information on the deceased and surviving PD pa­tients is shown in Table 3. The only factor significantly affecting survival was more advanced age in deceased patients (p=0.05).
Table 3. Clinical characteristics of deceased and surviving patients treated by PD

The mean annual mortality rate was 23.2%, equal in HD (22.8%) and PD (23.1%).
After the end of the 4-year observation period, type of dialysis modality (HD vs. PD) did not exert a significant impact on patient survival (Fig. 1); the log-rank test p-value equals 0.83.
Figure 1. Kaplan Meier Survival Function. The difference in mortality between the diabetic patients during 4-year observation period according to modality of dialysis treatment - PD versus HD

However, in the detailed analysis a tendency to higher mortality appeared in the PD group in the first observa­tion year, in which 8 patients died (30.8%) compared with 5 deceased patients in the HD group (14.3%), although the difference did not reach significance.
It is worth noting that after 4 years of follow-up surviving PD patients were significantly younger than HD survivors (58.22 ± 10.50 y versus 69.83 ± 8.38 y; p+0.017), although at baseline HD and PD patients did not differ in respect of age. Addition­ally, the PD survivors exhibited significantly lower albumin level than HD survivors (3.36+0.48 versus 3.98+0.45; p=0.012).
The most common cause of death was cardiovascular complications - 18 pts (45% of all deaths) (7 PD/11 HD), followed by infection - 12 pts (30%) (6 PD/6 HD), ma­lignancy - 5 pts (12.5%) (1 PD/4 HD) and other reasons (gastrointestinal bleeding) - 5 pts (12.5%) (3 PD/2 HD).
In the backward stepwise selected multivariate Cox pro­portional hazard regression model (Table 4) low serum albumin concentration was the only factor with a sig­nificant negative impact on 4-year survival in the whole diabetic cohort (p=0.047). Referring to dialysis modality the significant mortality predictor in the HD group was low cholesterol level (p=0.004), and exclusively older age (p=0.047) in the PD group.
Table 4. Cox proportional hazard regression model. Dependent variable: survival time from beginning of investigation

Discussion
The results of our study clearly indicate the gloomy long-term outlook of dialysis treatment in high cardiovascular risk diabetic patients, without a significant difference between PD and HD effect on survival. Two thirds of pa­tients died during four-year prospective observation (at a median time of 65 months from the beginning of dialysis treatment), with an annual mortality rate of 23.2% dur­ing the study period.
In the Cox proportional hazard regression model (Table 4) low serum albumin concentration was the only factor with a significant negative impact on 4-year survival in the whole diabetic cohort (Table 4; p=0.047). Referring to dialysis modality in the multivariate analysis the signifi­cant mortality predictor in the HD group was low choles­terol level (p=0.004), and exclusively older age (p=0.047) in the PD group. Despite general equality between PD and HD in terms of survival, the separate analysis of both groups revealed differences of particular clinical impor­tance. First, during the median time of 16.5 months be­tween dialysis start and study onset a drop of serum al­bumin in the PD group occurred, whereas in the similar time of 17 months no change of albumin level appeared in the HD group. This tendency to lowering of albumin concentration with time on PD was responsible for a sig­nificantly inferior albumin level in the PD compared to the HD group at the study onset. This negative event ap­peared in PD diabetic patients without augmentation of the inflammatory process, which was not different in PD and HD groups, exhibiting similar CRP and Il-6 values (Table 1). It shows that most patients with diabetes are not able to compensate for the peritoneal and urinary albumin loss. The occurrence of a lower albumin level was a universal feature in the PD group, losing due to this high distribution the impact on survival. This was the reason that the only significant difference between PD survivors and deceased patients was more advanced age in those who died (67.65±10.15 y versus 58.22±10.50 y; p=0.05). Older age retained its negative effect on PD pa­tient survival as the only independent factor in the Cox proportional hazard regression model (Table 4; p=0.047). The association of older age with mortality in the PD group was also revealed by significantly younger age of PD survivors compared to HD survivors (58.22 ± 10.50 y versus 69.83 ± 8.38 y, p=0.017). Additionally, the PD sur­vivors exhibited a significantly lower albumin level than HD survivors (p=0.012), which in turn reflects common hypoalbuminemia presence under PD treatment.
A different pattern of survival predictors appeared in patients with diabetes treated by HD. The HD survivors were characterized by a lower IL-6 level (p=0.04), higher albumin concentration (p=0.03), and increased choles­terol (p=0.004). This shows that long-term survival in HD pts is associated with two categories of variables, reflect­ing inflammation (IL-6, albumin) and nutrition (choles­terol), whereas age had no influence. In this context the different significance of hypercholesterolemia in PD pa­tients versus HD should be emphasized. Elevated choles­terol was a universal finding in PD patients, with very significantly higher values (p<0.0001) compared to the HD group, and a slight tendency to lower values in PD survivors. In PD patients hypercholesterolemia seems to appear as an element of atherosclerotic dyslipidemia, whereas in maintenance HD its occurrence is probably more connected with good nourishment. This difference may explain the failure to prove the advantage of statin application in HD populations in randomized control tri­als [3]. It also suggests more potential benefits from statin therapy in PD patients. In addition, it is noteworthy that IL-6 appeared to be a more sensitive inflammation indica­tor than CRP. In the Cox proportional hazard regression model low cholesterol level remained as the independent mortality predictor in HD pts (Table 4; p=0.004), again in­dicating a possible connection between cholesterol and nourishment in HD treatment. It should be mentioned that the other considered parameters - systolic blood pressure, residual diuresis, Hb, CRP, serum pro-ANP and NT-proBNP - did not exert an effect on survival. This is different to the situation observed in our previous study [9] encompassing elderly dialysis patients (median age 77 y) in whom paramount for survival appeared to be ad­equate extracellular volume control and as independent mortality predictors were found higher plasma pro-ANP, lower residual diuresis and lower systolic blood pressure.
The literature on the effects of dialysis modality on sur­vival in patients with diabetes is equivocal [10,15,16], with a study suggesting better outlook under PD treatment [2,5,8,11,18], and others signifying more beneficial results in HD programs [1,4,12]. Our study confirms the crucial importance of the initial characteristics of the diabetic co­hort [6,14,17,18], which may be responsible for contradic­tory results and is also in line with the United States and Canadian registries showing that HD is associated with better survival for diabetics aged 45 and older [16,19].
While conscious of our study limitations connected with the small number of included patients, we believe that it provides a couple of clinically important indications for appropriate application of dialysis treatment in diabet­ics. A strength of the study derives from the fact that the observed diabetic patients formed a homogeneous group with pronounced cardiovascular risk (Monckeberg type intravascular calcifications in all persons, with 26 patients (43%) over 70 years old). The association between the se­rum albumin level and mortality is well known from the seminal paper published by Lowrie and Lew [7]. However, in the current interpretation more emphasis is put in HD on the connections between hypoalbuminemia and inflam­mation than on the nutrition state. This relationship also appeared in the multivariate analysis in the present study when all patients were collectively considered indepen­dently of dialysis modality type. The division according to dialysis modality reveals clearly that distinct categories of metabolic disorders occur under PD and HD treatment. Therefore, from the methodological point of view the eval­uations of morbidity and mortality risk factors are more appropriate when they consider PD and HD separately.
In summary, the PD modality is more suitable for younger patients with diabetes mellitus below 60 y of age with albumin concentration in the upper half of the normal range; for those not fulfilling such characteristics, HD seems to be the preferred option. During performance of the HD program in diabetics, particular attention should be concentrated on maintaining appropriate nutrition with simultaneous inflammation control.
Conclusions
1. There is a difference in the mortality predictor pat­tern in high risk cohort diabetic patients dependent on dialysis modality.

2. In the PD program more advanced age has a decisive negative impact on survival, whereas during HD the­rapy the outlook is associated with variables reflecting nutrition and inflammation.

3. Elevated cholesterol level has different prognostic si­gnificance depending on the dialysis mode, exhibiting a survival advantage in HD patients.
REFERENCES
[1] Bloembergen W.E., Port F.K., Mauger E.A., Wolfe R.A.: A comparison of mortality between patients treated with hemodialysis and peritoneal dialysis. J. Am. Soc. Nephrol., 1995; 6: 177-183
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[2] Collins A.J., Hao W., Xia H., Ebben J.P., Everson S.E., Constantini E.G., Ma J.Z.: Mortality risks of peritoneal dialysis and hemodialysis. Am. J. Kidney Dis., 1999; 34: 1065-1074
[PubMed]  
[3] Epstein M., Vaziri N.D.: Statins in the management of dyslipidemia associated with chronic kidney disease. Nat. Rev. Nephrol., 2012; 8: 214-223
[PubMed]  
[4] Ganesh S.K., Hulbert-Shearon T., Port F.K., Eagle K., Stack A.G.: Mortality differences by dialysis modality among Incident ESRD patients with and without coronary artery disease. J. Am. Soc. Nephrol., 2003; 14: 415-424
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[5] Heaf J.G., Lokkegaard H., Madsen M.: Initial survival advantage of peritoneal dialysis relative to haemodialysis. Nephrol. Dial. Transplant., 2002; 17: 112-117
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[6] Jaar B.G., Coresh J., Plantinga L.C., Fink N.E., Klag M.J., Levey A.S., Levin N.W., Sadler J.H., Kliger A., Powe N.R.: Comparing the risk for death with peritoneal dialysis and hemodialysis in a national cohort of patients with chronic kidney disease. Ann. Intern. Med., 2005; 143: 174-183
[PubMed]  
[7] Lowrie E.G., Lew N.L.: Death risk in hemodialysis patients: the predictive value of commonly measured variables and an evaluation of Heath rate differences between facilities. Am. J. Kidney Dis., 1990; 15: 458-482
[PubMed]  
[8] Lukowsky L.R., Mehrotra R., Kheifets L., Arah O.A., Nissenson A.R., Kalantar-Zadeh K.: Comparing mortality of peritoneal and hemodialysis patients in the first 2 years of dialysis therapy: a marginal structural model analysis. Clin. J. Am. Soc. Nephrol., 2013; 8: 619-628
[PubMed]  
[9] Madziarska K., Weyde W., Krajewska M., Zukowska Szczechowska E., Gosek K., Penar J., Klak R., Golebiowski T., Kozyra C., Klinger M.: Elderly dialysis patients: analysis of factors affecting long-term survival in 4-year prospective observation. Int. Urol. Nephrol., 2012; 44: 955-961
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[10] Noordzij M., Jager K.J.: Survival comparisons between haemodialysis and peritoneal dialysis. Nephrol. Dial. Transplant., 2012; 27: 3385-3387
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[11] Schaubel D.E., Morrison H.I., Fenton S.S.: Comparing mortality rates on CAPD/CCPD and hemodialysis. The Canadian experience: Fact or fiction? Perit. Dial. Int., 1998; 18: 478-484
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[12] Stack A.G., Molony D.A., Rahman N.S., Dosekun A., Murthy B.: Impact of dialysis modality on survival of new ESRD patients with congestive heart failure in the United States. Kidney Int., 2003; 64: 1071-1079
[PubMed]  
[13] Stel V.S., van de Luijtgaarden M.W., Wanner C., Jager K.J.: The 2008 ERA-EDTA Registry Annual Report-a précis; on behalf of the European Renal Registry Investigators. NDT Plus, 2011; 4: 1-13
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[14] Termorshuizen F., Korevaar J.C., Dekker F.W., Van Manen J.G., Boeschoten E.W., Krediet R.T.: Hemodialysis and peritoneal dialysis: comparison of adjusted mortality rates according to the duration of dialysis: analysis of The Netherlands Cooperative Study on the Adequacy of Dialysis 2. J. Am. Soc. Nephrol., 2003; 14: 2851-2860
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[15] Tomaszuk-Kazberuk A., Bachorzewska-Gajewska H., Malyszko J., Malyszko J., Mysliwiec M., Musial W.J.: Impact of diabetes mellitus on survival in patients with end-stage renal disease: a three-year follow-up. Kidney Blood Press. Res., 2011; 34: 83-86
[PubMed]  
[16] Vonesh E.F., Snyder J.J., Foley R.N., Collins A.J.: Mortality studies comparing peritoneal dialysis and hemodialysis: what do they tell us?. Kidney Int. Suppl., 2006; 70: S3-S11
[PubMed]  
[17] Xue J.L., Chen S.C., Ebben J.P., Constantini E.G., Everson S.E., Frazier E.T., Agodoa L.Y., Collins A.J.: Peritoneal and hemodialysis: I. Differences in patient characteristics at initiation. Kidney Int., 2002; 61: 734-740
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[18] Xue J.L., Everson S.E., Constantini E.G., Ebben J.P., Chen S.C., Agodoa L.Y., Collins A.J.: Peritoneal and hemodialysis: II. Mortality risk associated with initial patient characteristics. Kidney Int., 2002; 61: 741-746
[PubMed]  [Full Text HTML]  [Full Text PDF]  
[19] Yeates K., Zhu N., Vonesh E., Trpeski L., Blake P., Fenton S.: Hemodialysis and peritoneal dialysis are associated with similar outcomes for end-stage renal disease treatment in Canada. Nephrol. Dial. Transplant., 2012; 27: 3568-3575
[PubMed]  [Full Text HTML]  [Full Text PDF]  
The authors have no potential conflicts of interest to declare.