Erythropoietin prevents dementia in hemodialysis patients: a nationwide population-based study

Erythropoietic medications such as including erythropoietin (EPO) are known to be neuroprotective and to correlate with improved cognitive functions. However, it is not known whether supplementation with EPO reduces the risk of dementia in end-stage renal disease (ESRD) patients receiving hemodialysis (HD). Here, we determined whether EPO levels correlate with the incidence of different dementia subtypes, including Alzheimer’s disease (AD), vascular dementia (VaD), and unspecified dementia (UnD), and whether such associations vary with annual cumulatively defined daily doses (DDDs) of EPO for ESRD patients receiving HD. This retrospective study included data from 43,906 adult ESRD patients who received HD between 1999 and 2010. Using hazard ratios and Cox regression models, we found that patients receiving EPO had a 39% lower risk of general dementia than those in the non-EPO group. Similarly, the risks of VaD and UnD was lower for patients in the EPO cohort. The risk of dementia was further reduced in HD patients treated with EPO in combination with iron. Our results suggest that the use of EPO medications in HD patients is associated with a reduced risk of VaD and UnD, but not AD, regardless of whether EPO is used alone or in combination with iron.


INTRODUCTION
In maintenance hemodialysis (HD) patients, renal anemia is generally treated with erythropoietic medications, including erythropoietin (EPO) and intravenous iron supplements. The introduction of EPOs has revolutionized the care of anemic patients with chronic kidney disease (CKD) and almost completely eradicated the severe anemia of end-stage renal disease (ESRD) patients. Moreover, EPOs decrease the need for recurrent blood transfusions and the risk of iron overload and may improve the patients' quality of life. Today, EPOs and adjuvant iron therapy are the main treatments for anemia associated with CKD [1,2].
AGING Treatment of anemia with EPO was associated with improved neuropsychological test performance and electroencephalography measurements in uncontrolled studies of patients with ESRD conducted in the early 1990s [3,4]. One study suggested that normalization of hemoglobin (Hb) using EPO was associated with further improvements in cognitive function [5], and other studies have suggested that EPO may exert neuroprotective effects independently of raising Hb levels [3,4].
Due to its protective effects on cognitive function, EPO has been used in the treatment of neuropsychiatric disorders with cognitive impairments, including schizophrenia [6]. Given that the EPO receptor is widely expressed in the nervous system and that EPO easily crosses the intact blood-brain barrier [7], EPO supplementation is able to rescue cognitive decline in aged rats and restore impaired memory in vascular dementia (VaD) rat models [8,9]. In addition, previous studies illustrated that EPO can attenuate hippocampal neuronal loss, neuroinflammation, and cholinergic deficit in rats [10], and can function as a neuroprotectant against amyloid beta (Aβ) toxicity [11,12], which is a principal consideration for the development of treatments for Alzheimer's disease (AD).
The data concerning the effects of EPO and intravenous iron supplementation on specific dementia subtypes among HD patients are scarce. Therefore, we conducted a total population-based retrospective cohort study to test whether EPO and intravenous iron supplementation correlate with the risk of various dementia subtypes including AD, VaD, and unspecified dementia (UnD) in HD patients.

RESULTS
The demographic characteristics and comorbidities of the HD patients in our cohorts are shown in Table 1. We recruited 43,906 HD patients who took EPO and 11,676 HD patients who took iron during the study periods. Of these, 11,189 patients in the EPO cohort (25.5%) and 487 patients in the non-EPO cohort (1.1%) were administered iron therapy for anemia ( Figure 1). The most frequent comorbidities were hypertension (88.9%), anemia (51.6%), diabetes mellitus (51.2%), and hyperlipidemia (41.9%). Less than 5% of the study population had a history of atrial fibrillation (AF), and nearly 53.9% were aged 60 years or older.

Annual cumulative exposure to EPO and iron
Biosimilar EPOs can be discriminated from the endogenous protein by slight differences that may include posttranslational modification; consequently, they may be similar-but not identical-to the originator EPOs [13]. A previous population-based study indicated that EPO consumption by HD patients was similar for biosimilar and originator EPOs [14]. Defined daily dose (DDD) is the assumed average maintenance dose per day for a drug used for its main indication in adults [15]. To investigate the effect of dose, and to avoid higher cumulative EPO doses with longer patient follow-up periods, the cumulative use of EPOs was calculated as total prescribed annual DDDs (i.e., the same as total dispensed DDD under this system). Upon dementia diagnosis, the cumulative EPO dosage was recorded as the total of annual DDDs from drug initiation to the day before the diagnosis. Patients who took EPO at least once after the index date were defined as EPO users; the remaining patients were defined as non-EPO users. Similarly, patients who received intravenous iron at least once after the index date were defined as iron users; the remaining patients were defined as non-iron users. Upon dementia diagnosis, the cumulative iron dosage was recorded as the total annual DDDs from drug initiation to the day before the diagnosis.
Therefore, using the daily records of prescribed erythropoietic medications during the follow-up period, we categorized patients who had received erythropoietic medications as the treatment group and those who did not receive the drugs as the control group.

Factors associated with dementia incidence
The results of our analysis examining the association between the use of EPO or intravenous iron and the risk of developing dementia are shown in Table 2. Stratified Cox proportional hazard regressions showed that the hazard ratio (HR) for dementia in HD patients who took EPO within the follow-up period was 0.48 [95% confidence interval (CI) 0.42 to 0.54; p<0.0001] in comparison with the HD patients in the non-EPO cohort. We further analyzed the HR value after adjusting the data for diabetes, hyperlipidemia, hypertension, anemia, coronary artery disease, heart failure, and AF. The HR value obtained was 0.61 (95% CI 0.54-0.70; p<0.0001) in HD patients who took EPO. Table 2 also shows the association between the annual DDDs and the risk of dementia. Patients who exhibited higher annual DDDs of EPO exhibited a decreased risk of dementia (28%-47%). Analyzing the annual DDDs of EPO indicated that the low-, medium-and high-dose groups exhibited reduced dementia rates compared with the non-EPO cohort, suggesting a reduction of risk for annual various DDDs of EPO. An additional analysis demonstrated an association between iron use and a reduced dementia risk (HR: 0.75; 95% CI, 0.65-0.86). In the analyses examining risks associated with different iron dosage, only low and medium doses of iron (<5 annual DDDs and 5-13 annual DDDs), but not high doses (≥14 annual DDDs), were associated with a reduced risk of dementia (39%-47%) in patients with HD using the non-iron cohort as a reference group.

Association between combined EPO and iron intake and general risk of dementia
The risk of incident dementia in relation to the combination of EPO and iron use was evaluated in HD patients, as compared with those who used neither drug ( Table 2). The Cox proportional hazards model revealed that after adjusting for age, sex, and comorbidities, the risk of incident dementia was reduced both among HD patients who had received both EPO and iron (adjusted HR = 0.49, 95% CI, 0.41-0.58), and those who had used EPO but not iron (adjusted HR = 0.68, 95% CI, 0.59-0.78). In addition, HD patients using EPO and iron treatments experienced a delayed onset of dementia or prevented it altogether (log-rank test, p < 0.0001, Figure 2). Table 3 shows the results of HR analysis for HD patient cohorts by dementia subtype. In comparison with non-EPO patients, HD patients treated with EPO were less likely to experience some subtypes of dementia during the follow-up period after the index healthcare use. Of note, the adjusted HRs for VaD and UnD in patients with HD were 0.44 (p<0.0001) and 0.65 (p<0.0001), respectively. Moreover, iron supplementation correlated with a reduced risk of UnD.

Risk of dementia subtypes upon EPO or Iron supplementation
We estimated whether EPO supplementation correlated with reductions in risk for specific dementia subtypes based on the tertile of annual DDDs for EPO use. We observed a reduced risk of VaD (adjusted HR range from 0.34 to 0.53) and UnD (adjusted HR range from 0.56 to 0.77) among EPO users (Table 3).
Furthermore, the Cox proportional hazards model revealed a reduced risk of UnD among HD patients who underwent iron supplementation, adjusting for age, sex, and comorbidities (adjusted HR = 0.74, 95% CI, 0.64-0.87) ( Table 3).

DISCUSSION
Our total population-based retrospective cohort study revealed that long-term administration of EPO to HD   patients was inversely associated with their general risk of developing dementia as well as specific dementia subtypes, including VaD and UnD. In addition, we also observed a reduced risk of dementia in HD patients supplementing with intravenous iron. To our knowledge, no prior study has explored the association between EPO supplementation, intravenous iron supplementation, and the risk of dementia subtypes.
Here, after controlling for potential confounders, we found that EPO use of less than 71 annual DDDs, 71-200 annual DDDs, and over 201 annual DDDs in cumulative dose is associated with a 28%, 47%, and 38% risk reduction in dementia, respectively, as compared with not using EPO. However, we found no consistent trends in risk reduction with EPO supplementation of 201 annual DDDs or greater. This paradoxical phenomenon could be attributed to the severity of medical comorbidities, or to a partial response to EPO. For example, patients who used AGING higher annual DDDs very likely had higher resistance to EPO [16] or more medical comorbidities that could cause a poorer response. Indeed, 10%-20% of CKD patients with anemia are resistant to EPOs [17]. Despite their potential for neuroprotection, EPOs might not be sufficient to overcome the adverse effects of severe resistance to EPO [18] or other medical comorbidities. Our study was restricted to analyzing data for HD patients, who already suffer from a high risk of developing dementia; therefore, our results may not be applicable to non-HD patients and further research is need to characterize the effects of EPO supplementation among the general population.
The neuroprotective mechanisms of EPOs include decreased neuronal apoptosis, decreased inflammation, promotion of oligodendrocyte differentiation and maturation, and improved white matter survival [19][20][21][22]. Two randomized, double-blind studies assessed the effect of EPO treatment on cognitive function [6,23]. They showed that intravenouslyadministered EPO leads to better performance in healthy subjects in a test of verbal fluency seven days after treatment [6,23] and that weekly intravenous injections of EPO lead to improved performance in tests of cognitive function in chronic schizophrenic patients. However, large randomized trials to treat anemia in CKD or ESRD patients with EPO AGING supplementation did not evaluate cognitive function [24]. Thus, there is currently insufficient evidence to justify changing current hemoglobin targets to prevent dementia in patients with CKD or ESRD. To our knowledge, our results here are the first to suggest that EPO supplementation correlates with a reduced risk of dementia and dementia subtypes in HD patients, regardless of whether EPO was used alone or combined with iron. Through our present study, after controlling for potential confounders, we also found that iron supplementation of less than 5 annual DDDs and 5-13 annual DDDs in cumulative dose is associated with a 47% and 39% risk reduction in dementia, respectively, as compared with no use of iron. Although EPOs have become the mainstay of anemia therapy in HD patients, iron deficiency and/or insufficient iron bioavailability emerges as a major limiting factor in the effectiveness of these treatments. Therefore, we conducted a subgroup analysis to study the effect of concomitant EPO and iron supplementation. We observed an adjusted HR of 0.49 (95%CI, 0.41-0.58) when compared with the no-EPO and no-iron subgroups. For many HD patients, intravenous administration of iron is often a prerequisite to elicit an optimal response to EPO.
The strengths of this study are its population-based survey with a large sample size, with good follow-up throughout. However, several limitations and precautions are needed for interpreting our results. First, some anemic patients can be asymptomatic and thus might not visit the clinic thereby eluding diagnosis; therefore, the incidence of EPO use in the nonanemic controls was probably overestimated because of the presence of these asymptomatic patients. In addition, unlike a previous study conducted by Kuo KL et al., [25], which analyzed patient hemoglobin levels and medical records, the present study relied on anemia diagnoses by clinicians, which may be less sensitive and delayed. Second, our study was observational in nature and cannot prove causality. Although we adjusted for common health conditions, it's possible that subclinical disease may also have contributed to cognitive decline. Third, hematological data such as ferritin and transferrin saturation (TSAT) were not available for us to consider them in the present study. Thus, it is not possible to evaluate whether iron was administered correctly or not. However, when it comes to prescribing EPOs to HD patients, physicians in Taiwan should follow the NHI reimbursement criteria to keep serum ferritin at levels >100 ng/mL and/or TSAT at levels > 20% during EPO therapy. We believe that the baseline iron parameters in our study might be in accordance with the NHI reimbursement criteria for all EPO users. Fourth, in the current study, the inclusion criteria of patients with dementia were strict, and only patients with at least three outpatient or inpatient claim records of dementiarelated diagnosis codes were included; however, these criteria might still underestimate the number of patients with dementia, particularly among those who rarely visit hospitals or those diagnosed with dementia near the end of 2011. Fifth, patient hemoglobin or hematocrit levels were not available for us to consider them in the present study even though previous studies performed in Taiwan have shown that reasonable hemoglobin targets and favorable outcomes for CKD anemia can be achieved by intravenous iron supplementation [26][27][28][29].
Lastly, we did not analyze the effects of antihypertensive and oral hypoglycemic drugs, which may affect the progression of neurodegenerative diseases. Nonetheless, our study demonstrated a 47% risk reduction for developing dementia in HD patients who used EPO supplementation in the range of 71-200 annual DDDs. Moreover, our results suggest that intravenous iron supplementation correlates with lower risks of dementia in HD patients, especially in combination with EPO.

Data collection
A universal National Health Insurance (NHI) program was implemented in Taiwan in March 1995. Ninety-six percent of the Taiwanese population has been enrolled in this program [30]. By the end of 1996, the Bureau of NHI (BNHI) had contracts with 97% of all Taiwanese hospitals and clinics to join the national health insurance system [31]. The NHRI safeguards the privacy and confidentiality of all beneficiaries and provides health insurance data for research only after ethical approval has been obtained.  [34].

Statistical analysis
The primary endpoint of this study was to determine whether a patient had received ambulatory care visits or had hospitalizations for any type of dementia. We used Pearson's χ 2 test to compare EPO users and iron users with controls in terms of region of residence (urban, suburban, and rural) and selected comorbidities (hypertension, diabetes, hyperlipidemia, coronary heart disease, AF, anemia, and other forms of heart disease) at baseline. We considered these comorbidities only if the condition occurred in an inpatient setting or if there were two or more ambulatory care claims recorded one year before or after the index ambulatory care visit. EPO users and iron users were categorized into tertiles by annual DDDs to explore the potential effects on risk reduction. We measured dementia incidence using Kaplan-Meier analyses during the follow-up period for the HD patients with/without EPO or iron treatments.
The unadjusted HR along with the 95% CI was obtained by evaluating the association between HD patients with different drug usage and risk of dementia during the follow-up period using Cox proportional hazard regression. The adjusted HR was computed after adjusting for days between index date and drug use, age, gender, urbanization level, hypertension, diabetes, hyperlipidemia, coronary heart disease, AF, anemia, and heart failure. We further analyzed dementia incidence rates between cohorts according to dementia subtypes. All data analyses were conducted using the SAS (ver. 9.4) statistical package for Windows (SAS Institute, Cary, NC), and the significance level was set at 0.05 in a two-sided test.