Percentage of Hypochromic Red Blood Cells is an Independent Risk Factor for Mortality in Kidney Transplant Recipients
Abstract
There are no published studies on the associations between anemia or iron status parameters and important long-term outcomes in kidney transplant recipients (KTR). We prospectively studied 438 KTR from a large transplant clinic for all-cause mortality and kidney allograft loss. Hemoglobin and iron status parameters (serum iron, transferrin, transferrin saturation, ferritin, percentage of hypochromic red blood cells [%HRBC]) were assessed at baseline as were important demographic, clinical and laboratory characteristics. The Austrian Dialysis and Transplant Registry and the Eurotransplant database were used to ascertain immunological and transplantation-related parameters and to ascertain death and allograft rejection. Cox proportional hazard models were used for analyses. Over 7.8 years of follow-up, 129 deaths (29.5%) occurred and 208 grafts (47.5%) were lost. From multivariate analyses, we found that anemia (hemoglobin <10g/dL) was not associated with mortality or graft loss. Among the iron status parameters, only %HRBC was associated with greater all-cause mortality. Patients with HRBC >10% had twice the mortality risk (HR: 2.06; 95%CI: 1.12–3.79) compared to patients with HRBC <5%. Neither of the iron status parameters were associated with allograft rejection. In conclusion, we found that %HRBC was an independent risk factor for mortality in KTR, while other iron status parameters or anemia were not associated with risk. Larger studies on the association between anemia and these outcomes are warranted.
Introduction
Anemia is a consequence of erythropoietin deficiency in patients with chronic kidney disease and thus a common complication in patients with end-stage renal disease (ESRD) (1). Prior to the introduction of recombinant human erythropoietin (rh-Epo) into routine medical care, the therapeutic options available for the treatment of renal anemia used to be rather limited and associated with much-dreaded side effects. Even in the era of rh-Epo, anemia appears to be associated with poor outcomes in patients on dialysis. Several observational studies have demonstrated that high achieved hemoglobin concentration (11–13 g/dL) correlated with improved outcomes in hemodialysis patients, including lower mortality and hospitalization rates (2–5). However, data from randomized trials in patients on dialysis, in which rh-Epo was used to raise hemoglobin levels, are more equivocal. For example, two large randomized trials found no difference in mortality rates between hemoglobin targets of 9–10.5 g/dL and 13–14 g/dL, with a trend toward higher mortality in the high hemoglobin arm of the larger study (6,7). Thus, the more recent efforts in improving the quality and outcomes of care in the renal replacement therapy (RRT) population have focused on defining an optimal target hemoglobin level, that would integrate the evidence of poor outcomes associated with both too low and near-normal levels of hemoglobin. Similarly important factors in that decision-making process are the aim for optimal management of iron metabolism in order to optimize the benefit from rh-Epo treatment, and the reality of limited health care budgets that call for a wise allocation of resources to maximize the health status of entire populations (8).
Interestingly, little is known about the role of iron deficiency as a prognostic factor in patients on RRT. Anemia can be assumed to be, at least in part, on the causal pathway between iron deficiency and important outcomes. Similarly, anemia and iron deficiency may be mutual confounders in such associations. Thus, particular care needs to be applied when studying the associations between iron deficiency, anemia and important outcomes. In the general population, iron deficiency is associated with an increased risk of low birth weight, prematurity and perinatal mortality. In iron-depleted infants, there is convincing evidence of impaired psychomotor development and cognitive performance. Finally, iron deficient adults have a decreased work capacity and less efficient response to exercise (9–11).
Interestingly, studies of the epidemiology of anemia in kidney transplant recipients (KTR) have only recently become available. It has been found that anemia, dependent on the time since transplantation and the definition of anemia used, was prevalent in 20–40% of KTR, and it was shown that KTR received rh-Epo therapy only infrequently even if their hemoglobin or hematocrit levels were quite low (12–17). To date, no prospective studies on the putative association between anemia and important outcomes in KTR, such as patient and graft survival are available. The possible association between iron status and patient and graft survival is also unknown. Only an association between anemia and de novo congestive heart failure has recently been reported (18). We conducted the present study to specifically fill this void. In a large cohort of KTR, we prospectively evaluated the prognostic value of hemoglobin and iron status parameters on all-cause mortality and kidney allograft loss. A priori, we hypothesized that anemia, defined as a hemoglobin concentration <10 g/dL would be associated with inferior outcomes. Similarly, we hypothesized that a higher percentage of hypochromic red blood cells, an indicator for iron availability and utilization (19), would be associated with greater mortality and allograft loss in KTR.
Methods
Study population
We prospectively collected detailed information on 438 consecutive stable KTR who presented to our transplant clinic over a period of 4 weeks in 1995, as previously reported (13,20). All patients consented to being included in the study. During the baseline visit, we ascertained each participant's age and gender (all patients were white/Caucasian), weight, height, underlying native kidney disease, time since last kidney transplantation and how many such procedures they had undergone previously. Each patient's body mass index (BMI) was calculated as the weight in kilograms divided by the squared height in meters. We noted whether a patient received any rh-Epo or iron therapy at that time, as well as detailed information about the product, route, dosage and duration of that therapy. The exact immunosuppressant regimen was also noted. A blood sample was taken during that visit and blood chemistry values were determined immediately at a single laboratory using standard methods: serum-creatinine, C-reactive protein, serum-iron, transferrin and ferritin. Transferrin saturation (TSAT) was calculated as ([iron level/transferrin level]* 70.9). We calculated each patient's creatinine clearance using the Cockcroft-Gault formula and standardized to a body surface area of 1.73 m2 (21). Whole blood counts and percentages of hypochromic red blood cells (%HRBC) were analyzed with a Technicon H*2 hematology analyzer (Bayer Diagnostics, Tarrytown, NY).
Study follow-up
From the day of the baseline visit, patients were followed longitudinally using the Austrian Dialysis and Transplant Registry (OeDTR) and the registry of the Eurotransplant Foundation, the joint organ procurement agency for Austria, Belgium, Germany, Luxemburg, The Netherlands and Slovenia. The Austrian Dialysis and Transplant Registry routinely collects longitudinal information on all dialysis patients and KTR residing in Austria. Follow-up in this database has been 100% complete for many years, and reliable information on timing and occurrence of patient death and modality switches, such as re-initiation of maintenance dialysis after kidney graft failure, is available for study. From the Eurotransplant Foundation database, we obtained information on the organ donor (donor age, gender, living vs. cadaveric donor), and on the specific circumstances of the transplantation procedure (cold ischemia time, number and type of human leukocyte antigen [HLA]-mismatches, and recipient panel reactive antibody [PRA]-titer).
Statistical analyses
All statistical analyses used the SAS for Windows statistical software package (release 8.2; The SAS Corporation, Inc., Cary, NC). Spearman correlation was used to describe pair-wise relationships between continuous variables. We built univariate and multivariate Cox proportional hazards models to describe the crude and independent relationships between the main exposure variables, hemoglobin and iron status parameters (serum iron, transferrin, TSAT, ferritin %, HRBC), and the outcomes of interest (22). These were all-cause mortality, and kidney allograft loss, which was defined as the composite endpoint of patient death and re-initiation of maintenance dialysis. Additional analyses using re-initiation of maintenance dialysis as the outcome and death as a censoring rather than an event-indicator were also conducted (death-censored graft failure). Hazards ratios (HR) were used as the measure of association and provided together with the corresponding 95% confidence interval (CI).
For all multivariate analyses, we used automated stepwise model selection procedures that would only include variables below a multivariate p < 0.20 in the outcomes model. Hemoglobin and iron status parameters were then manually introduced into these pre-selected models. Due to the collinearity among hemoglobin and the iron status parameters (13), we decided to first use only one of these parameters in the multivariate models at a time. For additional analyses, we built models that included hemoglobin as well as one iron metabolism parameter each. Age and estimated creatinine clearance, likely the most important confounders of the studied associations, were used as continuous variables and forced into all models. Each variable that was not automatically selected into an outcome model was also manually introduced to ensure that confounding control was as optimal as possible given all the information which were available for study.
Results
The mean age among the 438 stable KTR was 51.6 years, and nearly 60% were male. At the time of study inclusion, the median time since transplantation was 4.4 years. The mean creatinine was 1.8 mg/dL and the average estimated GFR was 52.8 mL/min per 1.73m2. A detailed description of the study population, their donors and the actual transplant procedure is provided in Table 1.
| Variable | Count (proportion in %) or mean (±SD) |
|---|---|
| Age (years) | 51.6 (±13.3) |
| Gender (male) | 261 (59.6%) |
| Native kidney disease | |
| Glomerulonephritis | 144 (32.9%) |
| Polycystic kidney disease | 56 (12.8%) |
| Interstitial nephritis | 36 (8.2%) |
| Diabetic nephropathy | 26 (5.9%) |
| Various other (specified) | 56 (12.8%) |
| Unknown | 120 (27.4%) |
| Body mass index (kg/m2) | 25.5 (±4.2) |
| C-reactive protein (≥0.5 mg/dL) | 75 (17.1%) |
| Glomerular filtration rate (mL/min per 1.73m2) | 53 (±19) |
| Hemoglobin (g/dL) | 13.0 (±2.1) |
| Serum iron (μg/dL) | 82 (±33) |
| Serum transferrin (mg/dL) | 267 (±54) |
| TSAT (%) | 24 (±11) |
| Serum ferritin (μg/L) | 196 (±375) |
| <30 μg/L | 106 (24.3%) |
| 30–300 μg/L | 270 (61.6%) |
| >300 μg/L | 62 (14.2%) |
| Hypochromic red blood cells (%) | 2.5 (±5.8) |
| <5% | 393 (89.7%) |
| 5%–10% | 21 (4.8%) |
| >10% | 24 (5.5%) |
| First kidney transplant | 356 (81.3%) |
| Years since transplantation | 4.9 (±3.7) |
| Cold ischemia time >20 h | 213 (48.6%) |
| Panel reactive antibodies >50% | 39 (8.9%) |
| Donor age (years) | 38 (±15) |
| Donor gender (male) | 284 (64.8%) |
| Donor type (cadaveric) | 417 (95.2%) |
| HLA mismatch | |
| Locus A (any) | 65.3% |
| Locus B (any) | 74.1% |
| Locus DR (any) | 39.4% |
| Immunosuppressive regimen | |
| Cyclosporine A + corticosteroid + azathioprine | 269 (61.4%) |
| Cyclosporine A + corticosteroid | 123 (28.1%) |
| Other | 46 (10.5%) |
| Erythropoietin therapy (any dose and duration) | 29 (6.6%) |
| Iron substitution therapy (any dose and duration) | 30 (6.9%) |
- Abbreviations: TSAT, transferrin saturation.
Pair-wise Spearman correlation coefficients among several iron status parameters and hemoglobin from the study cohort are displayed in Table 2. The percentage of HRBC showed the best correlation with hemoglobin levels (r =−0.28, p < 0.001), while the correlation between hemoglobin and ferritin was particularly poor (r =−0.10, p = 0.04), and no correlation was found between TSAT and hemoglobin levels (p = 0.51).
| Hemoglobin | %HRBC | Serum iron | Ferritin | Transferrin | TSAT | |
|---|---|---|---|---|---|---|
| Hemoglobin | 1.0 | r =−0.28, p < 0.001 | r =+0.14, p = 0.03 | r =−0.10, p = 0.04 | r =+0.20, p < 0.001 | r =−0.03, p = 0.51 |
| %HRBC | – | 1.0 | r =−0.42, p < 0.001 | r =−0.20, p < 0.001 | r =+0.04, p = 0.36 | r =−0.35, p < 0.001 |
| Serum iron | – | – | 1.0 | r =+0.29, p < 0.001 | r =−0.07, p = 0.12 | r =+0.86, p < 0.001 |
| Ferritin | – | – | – | 1.0 | r =−0.52, p < 0.001 | r =+0.45, p < 0.001 |
| Transferrin | – | – | – | – | 1.0 | r =−0.47, p < 0.001 |
| TSAT | – | – | – | – | – | 1.0 |
- Abbreviations: %HRBC, percentage of hypochromic red blood cells; TSAT, transferrin saturation.
The cohort was followed-up over a median of 7.8 years. Over 2952 person-years, 129 patients died (crude mortality rate = 43.7/1000 person-years) and over 2635 person-years of follow-up, 208 kidney allografts were lost (crude rate of allograft loss = 78.9/1000 person-years).
Analyses of kidney transplant recipient mortality
From univariate Cox proportional hazards models, we failed to detect an association between hemoglobin and all-cause mortality (HR = 0.96; 95% CI: 0.88–1.05), but from categorical analysis of quartiles of hemoglobin, it became apparent that the association with mortality was not linear. Using our predefined definition of anemia, hemoglobin <10 g/dL was associated with a 77% greater mortality risk (HR = 1.77; 95% CI: 1.01–3.07). Among the iron status parameters, higher serum iron level was associated with lower mortality (HRfor each 10 μg/dL increase= 0.93; 95% CI: 0.88–0.99), while a greater %HRBC was associated with a higher risk for death (HR = 1.03; 95% CI: 1.01–1.05). The other iron status parameters were not associated with mortality in univariate analyses (Table 3).
| Univariate | Multivariate1 | |||||
|---|---|---|---|---|---|---|
| Hazards ratio | 95% CI | p-value | Hazards ratio | 95% CI | p-value | |
| Hemoglobin (for each 1 g/dL increase) | 0.96 | 0.88–1.05 | 0.35 | 1.01 | 0.90–1.14 | 0.86 |
| Serum iron (for each 10 μg/dL increase) | 0.93 | 0.88–0.99 | 0.02 | 0.95 | 0.89–1.01 | 0.11 |
| Ferritin (for each 100 μg/L increase) | 0.99 | 0.94–1.04 | 0.57 | 0.99 | 0.93–1.04 | 0.60 |
| Ferritin | ||||||
| <30 μg/L | 0.76 | 0.50–1.18 | 0.22 | 0.99 | 0.63–1.57 | 0.97 |
| 30–300 μg/L | 1.0 | Referent | – | 1.0 | Referent | – |
| >300 μg/L | 0.88 | 0.52–1.48 | 0.63 | 1.13 | 0.65–1.95 | 0.67 |
| Transferrin (for each 10 mg/dL increase) | 1.01 | 0.98–1.05 | 0.47 | 1.01 | 0.98–1.05 | 0.51 |
| TSAT (for each 1% increase) | 0.98 | 0.97–1.00 | 0.08 | 0.99 | 0.97–1.01 | 0.26 |
| %HRBC (for each 1% increase) | 1.03 | 1.01–1.05 | <0.001 | 1.02 | 1.00–1.04 | 0.03 |
| HRBC | ||||||
| <5% | 1.0 | Referent | – | 1.0 | Referent | – |
| 5–10% | 1.22 | 0.60–2.62 | 0.61 | 1.10 | 0.50–2.45 | 0.81 |
| >10% | 2.23 | 1.25–3.96 | 0.006 | 2.06 | 1.12–3.79 | 0.02 |
- 1Multivariate models simultaneously adjusting for recipient age, gender, C-reactive protein, body mass index, glomerular filtration rate, native kidney disease, time since transplantation, donor age, HLA-DR mismatch and panel reactive antibody titer. Each variable listed in Table 3 introduced separately into full model. Abbreviations: %HRBC, percentage of hypochromic red blood cells; TSAT, transferrin saturation.
The automated model selection process yielded a multivariate model that included recipient age, gender, C-reactive protein, BMI, glomerular filtration rate, native kidney disease, time since transplantation, donor age, HLA-DR mismatch and PRA titer as factors each being independently associated with mortality at p < 0.20. We then added hemoglobin and each of the iron status parameters individually into that model. From these multivariate analyses (Table 3), we found that %HRBC was the only iron status parameter to be significantly associated with all-cause mortality (HR = 1.02; 95% CI: 1.00–1.04). When using categorical indicators of HRBC percentage, we found that having HRBC >10% at baseline conferred twice the mortality risk (HR = 2.06; 95% CI: 1.12–3.79) compared to patients with HRBC <5%. All other iron status parameters failed to demonstrate an independent association with mortality risk. Hemoglobin <10 g/dL showed a non-significant trend towards greater mortality (HR = 1.88; 95% CI: 0.98–3.61). Finally, we added both hemoglobin and %HRBC into the multivariate model. The association between %HRBC and mortality remained unchanged, whereas the association with hemoglobin <10 g/dL was attenuated (HR = 1.62; 95% CI: 0.83–3.17) and remained not associated with mortality risk. No effect modification was found between %HRBC and hemoglobin (p = 0.50). When adding the treatment indicator for rh-Epo therapy, we found a trend towards a greater mortality among those receiving rh-Epo therapy at baseline (HR = 1.93; 95% CI: 0.87–4.29; p = 0.11). No association was found for baseline iron therapy (HR = 1.29; 95% CI: 0.61–2.69; p = 0.51). Including these treatment indicators to the models, or restricting the models to patients without rh-Epo and/or iron treatment at baseline, did not materially change the results.
Analyses of kidney allograft loss
From the univariate analyses of allograft loss (defined as the earliest of patient death, return to dialysis or re-transplantation), we found that higher hemoglobin levels were associated with a 16% reduced risk of graft loss (per 1g/dL increase; HR = 0.84; 95% CI: 0.78–0.90). A greater %HRBC was associated with a greater risk of graft loss (HR = 1.02; 95% CI: 1.00–1.04). All other iron status parameters as well as hemoglobin <10 g/dL were not associated with this outcome (Table 4).
| Univariate | Multivariate1 | |||||
|---|---|---|---|---|---|---|
| Hazards ratio | 95% CI | p-value | Hazards ratio | 95% CI | p-value | |
| Hemoglobin (for each 1 g/dL increase) | 0.84 | 0.78–0.90 | <0.001 | 0.94 | 0.86–1.02 | 0.15 |
| Serum iron (for each 10 μg/dL increase) | 0.97 | 0.93–1.01 | 0.11 | 0.98 | 0.93–1.02 | 0.30 |
| Ferritin (for each 100 μg/L increase) | 1.01 | 0.98–1.05 | 0.49 | 1.0 | 0.97–1.03 | 0.97 |
| Ferritin | ||||||
| <30 μg/L | 0.75 | 0.54–1.06 | 0.10 | 0.87 | 0.61–1.25 | 0.46 |
| 30–300 μg/L | 1.0 | Referent | – | 1.0 | Referent | – |
| >300 μg/L | 0.94 | 0.63–1.39 | 0.75 | 1.03 | 0.68–1.55 | 0.89 |
| Transferrin (for each 10 mg/dL increase) | 0.99 | 0.96–1.01 | 0.33 | 1.01 | 0.99–1.04 | 0.36 |
| TSAT (for each 1% increase) | 1.00 | 0.99–1.01 | 0.85 | 1.00 | 0.97–1.01 | 0.68 |
| %HRBC (for each 1% increase) | 1.02 | 1.00–1.04 | 0.02 | 1.01 | 0.99–1.03 | 0.42 |
| HRBC | ||||||
| <5% | 1.0 | Referent | – | 1.0 | Referent | – |
| 5–10% | 0.90 | 0.46–1.76 | 0.76 | 0.74 | 0.37–1.46 | 0.38 |
| >10% | 1.45 | 0.86–2.46 | 0.16 | 1.26 | 0.72–2.19 | 0.42 |
- 1Multivariate models simultaneously adjusting for recipient age, gender, C-reactive protein, body mass index, glomerular filtration rate, native kidney disease and panel reactive antibody titer. Each variable listed in Table 4 introduced separately into full model.
- Abbreviations: %HRBC, percentage of hypochromic red blood cells; TSAT; transferrin saturation.
The automated model selection process yielded a multivariate model that included recipient age, gender, C-reactive protein, BMI, glomerular filtration rate, native kidney disease and PRA titer as factors each being independently associated with allograft loss at a cutoff of p < 0.20. When additionally introducing either hemoglobin or %HRBC, neither of the two parameters was found to be associated with graft loss, illustrating that their univariate associations were both confounded by other prognostic factors in the multivariate model. None of the remaining iron status parameters was associated with allograft loss, either (Table 4). Similar to the analyses of patient survival, baseline rh-Epo therapy showed a trend towards greater graft loss (HR = 1.55; 95% CI: 0.92–2.62; p = 0.10), while no association was found for any baseline iron therapy (HR = 0.94; 95% CI: 0.54–1.65; p = 0.83).
When conducting analyses of allograft loss (return to RRT) where death was used as a censoring indicator rather than an outcome (“immunologic graft loss”), the results were very similar to the ones obtained from the combined endpoint above (Table 5). As before, neither hemoglobin, nor any of the iron status parameters was independently associated with death-censored allograft loss. Interestingly, while no association was found for rh-Epo therapy at baseline (HR = 1.43; 95% CI: 0.80–2.57; p = 0.23), there was a trend toward greater graft survival among those KTR who received any iron therapy at baseline (HR = 0.51; 95% CI: 0.24–1.09; p = 0.08).
| Univariate | Multivariate1 | |||||
|---|---|---|---|---|---|---|
| Hazards ratio | 95% CI | p-value | Hazards Ratio | 95% CI | p-value | |
| Hemoglobin (for each 1 g/dL increase) | 0.75 | 0.68–0.83 | <0.001 | 0.91 | 0.81–1.02 | 0.12 |
| Serum iron (for each 10 μg/dL increase) | 0.98 | 0.93–1.04 | 0.47 | 0.99 | 0.92–1.05 | 0.73 |
| Ferritin (for each 100 μg/L increase) | 1.02 | 0.98–1.06 | 0.26 | 1.01 | 0.97–1.05 | 0.76 |
| Ferritin | ||||||
| <30 μg/L | 0.70 | 0.45–1.10 | 0.12 | 0.88 | 0.55–1.43 | 0.61 |
| 30–300 μg/L | 1.0 | Referent | – | 1.0 | Referent | – |
| >300 μg/L | 0.87 | 0.52–1.48 | 0.61 | 1.08 | 0.62–1.87 | 0.79 |
| Transferrin (for each 10 mg/dL increase) | 0.98 | 0.94–1.01 | 0.17 | 1.01 | 0.98–1.05 | 0.45 |
| TSAT (for each 1% increase) | 1.01 | 0.99–1.02 | 0.43 | 1.00 | 0.98–1.02 | 0.90 |
| %HRBC (for each 1% increase) | 1.02 | 1.00–1.04 | 0.06 | 1.01 | 0.99–1.04 | 0.28 |
| HRBC | ||||||
| <5% | 1.0 | Referent | – | 1.0 | Referent | – |
| 5–10% | 0.66 | 0.24–1.78 | 0.41 | 0.46 | 0.16–1.30 | 0.14 |
| >10% | 1.30 | 0.64–2.67 | 0.47 | 1.31 | 0.63–2.70 | 0.47 |
- 1Multivariate models simultaneously adjusting for recipient age, gender, body mass index, glomerular filtration rate, native kidney disease, donor gender and panel reactive antibody titer. Each variable listed in Table 5 introduced separately into full model. Abbreviations: %HRBC, percentage of hypochromic red blood cells; TSAT, transferrin saturation.
Discussion
This is the first study of the associations between anemia and iron status parameters and important outcomes in KTR such as mortality and kidney allograft loss. From a large cohort of stable KTR with long follow-up of >7 years, we found that neither hemoglobin, nor anemia, defined as a hemoglobin <10 g/dL were independently associated with either of the outcomes studied. In contrast, %HRBC, a reliable, but rarely used indicator of iron status and metabolic iron utilization (19), was associated with greater all-cause mortality. The percentage of HRBC was not found to be associated with risk of allograft failure. These findings arose from multivariate models that carefully controlled for a comprehensive set of other important factors, both known predictors and likely confounders of these outcomes. In contrast to the positive association between %HRBC and mortality, the other iron status parameters studied failed to demonstrate such an association. Only serum iron showed a reasonably strong association of the same logical direction, similar to the one found for %HRBC, even if the 95% confidence interval crossed the null value. However, it is important to note that this study neither attempts to compare the various iron status parameters regarding their associations with the outcomes of study directly, nor would it be sufficiently powered to do so.
While there are no studies on the associations between markers of iron status and mortality in patients with chronic kidney disease (CKD), ESRD or KTR, a few are available in the general population. From a large random population sample in the US, the data indicated an inverse association between TSAT levels and overall mortality and mortality from cardiovascular causes (23). In a meta-analysis of iron status parameters and death from coronary heart disease, similar trends towards a protective association were identified for patients with higher TSAT or serum iron levels, while there the finding was null for serum ferritin levels (24). However, several studies have not found such associations.
The finding that anemia was not associated with all-cause mortality confirms recent findings in heart transplant patients (25), but is in stark contrast with findings from observational studies on this research question in patients on dialysis. In several of these studies, lower hemoglobin levels or hematocrit have been found to be associated with greater mortality (2–5). It is absolutely possible, that these studies did not include certain factors that might be important confounders of these associations, namely indicators of inflammation and iron availability, amongst others. This notion finds support by the finding that patients with lower rh-Epo dose experienced superior long-term outcomes, compared to patients who required higher doses of the hormone (26). Thus, anemia, or required rh-Epo dose might simply be indicators for complex underlying mechanisms of disease that are, in turn, associated with adverse outcomes. Alternatively, our failure to detect an association between anemia and the outcomes of interest may simply be a consequence of limited statistical power.
It has been proposed and demonstrated, that KTR can be used as a model population for patients with CKD (27). Thus, careful analyses of the associations between anemia and iron status parameters and important clinical endpoints in patients with CKD can be expected with great interest. Meanwhile, overly eager introduction of rh-Epo therapy in this population, which is also a most promising growth market for manufacturers of erythropoiesis stimulating agents, might be premature in the absence of solid scientific evidence. Perhaps, it would be more prudent to prioritize the research agenda on improving the screening for iron deficiency or disturbances of iron utilization, and then conduct trials testing the therapeutic effect of iron supplementation regimen (alone vs. placebo/usual care, or in 2 * 2 factorial designs with rh-Epo) on outcomes in patients with CKD and KTR. The present study would certainly support such an approach.
An interesting minor detail of this study also deserves mention. While patients receiving rh-Epo treatment at study baseline appeared to show a trend towards higher rates of the outcomes studied, the situation was different for the treatment indicator for baseline iron therapy. While there was a slight trend toward greater mortality among those patients receiving iron supplementation, the analyses of death-censored graft loss pointed toward a lower risk (HR = 0.51; 95% CI: 0.24–1.09; p = 0.08). This raises the interesting hypothesis whether iron supplementation might actually reduce immunogenic graft loss. This has to be interpreted in light of the likely scenario that probably only the sickest and the most symptomatic patients drew therapeutic intervention regarding both iron therapy and rh-Epo treatment, as can be seen from the positive associations between either treatment and mortality.
The Spearman correlations among the iron status parameters were rather moderate at best, highlighting the fact that each parameter is influenced by different extraneous factors that make the desired assessment of iron availability for erythropoiesis difficult and rather imprecise. Thus, the medical community has struggled to identify a best, or most reliable measure of iron status. Maybe, %HRBC can fill this void. This measure might have a theoretical advantage over the others in that it is least influenced by nutritional status, concurrent iron therapy, or inflammation. A more detailed discussion on this topic has been published elsewhere (13,28). In brief, the traditional iron status parameters (serum iron, ferritin, transferrin, TSAT) measure iron availability and/or storage, all of which are naturally influenced by several factors (diet, nutritional status, inflammation, etc). However, these markers do not reflect utilization of the available iron for erythropoiesis. The percentage of HRBC aggregates iron availability and utilization into a single measure one step further downstream in the erythropoietic process. Thus, %HRBC measures not only a critical input, but also the quality of the output of the red blood cell production process (19). Unfortunately, while technically easy to perform, this measure is unavailable in most routine health care settings.
This study has certain limitations. We acknowledge the hypothesis generating nature of this study, and its findings will need to be validated in other settings. A related and more technical limitation is the presence of a multiple testing situation. While we specifically hypothesized that %HRBC would be associated with the outcomes studied, and this specific null hypothesis was rejected in the mortality model, other associations were also evaluated. Accounting for multiple testing, the associations between %HRBC (both as continuous or categorical variable) and mortality would not reach significance and leave open the possibility of a chance finding. This fact is one more reason to emphasize the necessity to confirm these findings in other, independent cohorts of KTR. The study inclusion of prevalent rather than incident patients, allows for the presence of time-related biases, mainly survival bias. We deem this to be unlikely, which is supported by our failure to formally detect any violations of the proportionality assumption of the hazards in these Cox models. While we did not detect an association between anemia and the outcomes of interest, the power to do so was probably limited, especially for the analyses of all-cause mortality. Further, the present analysis uses an intent-to-treat approach. It is possible that correction of anemia after baseline occurred, which would lead to a bias toward a null finding of no association. This is a potential limitation that is shared with virtually all other outcomes studies of anemia. While we cannot rule out this possibility, the recent evidence of very low treatment rates of KTR even with severe anemia in several large transplant centers would not suggest so (12–17). Lastly, the portfolio of immunosuppressive therapeutics has been expanded greatly over the past years, and some of the newer immunosuppressants have been implicated in being associated with a greater degree of anemia (12–17). While accounting for the exact immunosuppressive regimen did not change the results in this study, it is uncertain how immunosuppressive regimens that have become available more recently would influence similar studies.
Our study has important implications for the clinical care of these patients and points towards important opportunities for future research. The present findings recommend the introduction of %HRBC assessment into routine care, while certain other iron status parameters (ferritin, TSAT) are confirmed to have limited value. As anemia has not been found to be associated with either all-cause mortality or allograft rejection in KTR, a conservative approach regarding the aggressive introduction of erythropoietic therapies into the kidney transplant population appears prudent. This is even more important, since %HRBC did demonstrate an association with all-cause mortality, even after controlling for hemoglobin, a factor that is likely on the causal pathway between iron deficiency and this outcome. Lastly, our literature review revealed that there are no studies on the association between iron status and outcomes in KTR. Thus, further studies of the relationship between anemia, iron metabolism disturbances and its therapies are warranted.
Acknowledgments
WC Winkelmayer and G Sunder-Plassmann were responsible for study design, data collection and interpretation, and drafting of the manuscript. WC Winkelmayer conducted the data analyses. M Lorenz and R Kramar were responsible for data collection and drafting of the manuscript. WH Hörl was responsible for data interpretation and drafting of the manuscript.
