In search of potential predictors of erythropoiesis-stimulating agents (ESAs) hyporesponsiveness: a population-based study

Erythropoiesis-stimulating agents (ESAs) are biological products, analogues of human erythropoietin, produced by cell lines using the recombinant DNA technology. ESAs are approved for the treatment of anaemia related to chronic kidney disease (CKD) or chemotherapy-induced in cancer patients. According to the Italian Medicines Agency, ESAs are indicated when hemoglobin (Hb) levels are lower than 11 g/dl in CKD patients and lower than 10 g/dl in cancer patients. In Italy, for both indications, haemoglobinemia has to range between 11 and 12 g/dl [1], avoiding a rise in Hb values greater than 2 g/dl over a four-week period.

Generally, the term “ESA hyporesponsive” refers to patients who need high doses of ESAs (25–100% higher doses than what recommended) to increase and/or maintain their Hb levels within the acceptable range [2]. More specifically, the Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines define patients as ESA hyporesponsive if they do not experience an increase in Hb levels within the first month of ESA treatment, using an appropriate weight-based dosing (not graded) [3].

ESA hyporesponsiveness could be acute or chronic, but, to date, there is neither consensus nor shared position on the definition of the chronic condition in particular [3]. Based on the definition gave by Sibbel et al., “4 months of continuous ESA hyporesponsiveness [defined considering both Hb concentrations and ESAs doses] can be used to differentiate acute from chronic hyporesponsiveness” [4].

A previously published population-based study, conducted on Italian administrative healthcare databases, evaluated the comparative effectiveness of both biosimilar and originator ESAs in CKD and cancer patients. Results highlighted that, in clinical practice, around 20% of ESA users were non-responders, defined as subjects experiencing no variations or a reduction in Hb levels within the first 3 months of ESA treatment. Furthermore, no differences were observed between different type of ESAs (i.e., biosimilars or originators), in terms of ESA responsiveness [5].

In patients with conservative end-stage renal disease, as well as in dialysis patients, ESA hyporesponsiveness and Hb level variability may lead to cardiovascular complications, increasing the risk of all-cause mortality, due to the required higher doses of ESA [6‐8].

In both CKD and cancer patients, several factors may contribute to ESA hyporesponsiveness, such as iron deficiency, inflammation and malnutrition status, while chronic hyperparathyroidism may affect ESA response in CKD patients, specifically [9, 10].

Debate is still on-going regarding the potential effects of renin-angiotensin system inhibitors, such as angiotensin-converting enzyme (ACE) inhibitors or angiotensin II-receptor antagonists (ARBs), on the development of anaemia in patients with renal disease [11].

This naturalistic population-based study was aimed at identifying which factors could be associated to ESA hyporesponsiveness in anaemic patients with CKD or cancer, in the general population of two Italian Local Health Units (LHUs).

On a total population of 1,538,812 subjects registered in Treviso and Caserta LHUs, 10,714 (0.7%) received at least one ESA dispensing during the years 2009–2015; of these, 1080 (10.1%) incident ESA users were included in the study, based on the above-mentioned inclusion criteria [CKD = 616 (57.0%); cancer = 464 (43.0%)] (Fig. 1).

As shown in Table 1, ESAs were in general more frequently used by males among CKD patients, and by females among cancer patients. Regarding age distribution, incident ESA users with CKD appeared to be on average older (mean age ± SD: 72.6 ± 14.7) than patients with cancer (66.9 ± 12.2). Although most of ESA users started ESA treatment having baseline Hb values within the range recommended by the Italian guidelines (Hb < 10 g/dL in cancer and Hb < 11 g/dL in CKD), 18.5% (N = 114) of CKD patients and 10.3% (N = 48) of cancer patients started ESA treatment with baseline Hb values ≥11 g/dL. Around 45% of incident ESA users received a biosimilar ESA at ID, irrespective of indication of use. In general, CKD patients were more likely to be hospitalized than cancer patients (66.2% vs. 56.2%), especially due to non-cardiovascular diseases. As compared to cancer patients, CKD patients were more likely to be affected by chronic comorbidities, such as hypertension (93.2% vs. 66.8%) and diabetes mellitus (41.9% of vs. 25.0%). Among ESA users with CKD, 410 (66.5%) were affected by stage IV-V CKD or were on dialysis. Instead, more than one third of cancer patients were affected by solid malignant neoplasms, although for most of cancer patients the type of tumor was not known (N = 208; 44.8%). CKD patients were more likely to be treated with iron preparations (CKD: 18.3%; cancer: 8.4%) or anti-hypertensive drugs (ACE inhibitor or ARBs) (CKD: 43.0%; cancer: 30.8%) than cancer patients. Considering laboratory parameters, no differences were found among cancer and CKD patients.

The target Hb value, as recommended by the Italian Medicines Agency, was reached on average between 45 and 60 days after ID and was thereafter stable during follow-up (Fig. 2).

Excluding incident ESA users with baseline Hb values higher than recommended (i.e., ≥11 g/dL), we observed that most of subjects included in the study cohort reached, at least once, the target Hb values (11 ≤ Hb levels≤12 g/dL), according to recommendations from Italian guidelines, despite 664 (61.5%) incident ESA users reached Hb levels> 13.0 g/dL, at least once during follow-up (Additional file 1).

Table 2 showed that, for each cohort, the proportion of ESA hyporesponsive patients was similar using the two approaches of ESA hyporesponsiveness. According to the given definition of ESA hyporesponsiveness, the multivariate binary logistic regression showed that the type of dispensed ESA (biosimilar or originator) was not a predictor of ESA response in CKD. Moreover, high baseline Hb values (OR = 1.7, 95% CI: 1.2–2.2; P-value< 0.001) and CRP hematic levels (OR = 1.2, 95% CI: 1.0–1.5; P-value = 0.060) were associated to ESA hyporesponsiveness in CKD (Table 3), while high baseline Hb values (OR = 1.7, 95% CI: 1.1–2.4; P-value = 0.007) and prior ischemic heart disease diagnosis (OR = 2.7, 95% CI: 0.9–7.9; P-value = 0.072) were predictors of ESA hyporesponsiveness in cancer patients (Table 4). On the contrary, ESA hyporesponsiveness was decreased by concomitant use of iron preparations (OR = 0.3, 95% CI: 0.2–0.7; P-value = 0.002) and high dosage of ACE inhibitors/ARBs (OR = 0.5, 95% CI: 0.3–0.9; P-value = 0.022) in CKD patients and by higher levels of albumin and potassium in cancer patients, although not significantly (P-values> 0.005). The discriminatory power of the predictive response of the variables included into the models was good, as confirmed by the ROC curves (Figs. 3-4).

By modifying the ESA hyporesponsiveness definition in the sensitivity analysis, high baseline Hb value was a positive predictor of responsiveness both in CKD (OR = 0.7, 95% CI: 0.5–1.0; P-value = 0.053) and in cancer patients (OR = 0.5, 95% CI: 0.3–0.8; P-value = 0.003); that is, patients with high baseline Hb values had more chance to reach the target Hb values rather than patients starting ESA treatment with low baseline Hb values (Additional files 2, 3, 4, 5). In addition, concomitant use of iron preparations and acidosis condition increased ESA response in CKD patients, with a good predictive power (AUC = 0.6969 ± 0.03; P-value< 0.001). Moreover, males with CKD (OR = 0.5, 95% CI: 0.3–0.9; P-value: 0.011) seemed to be more responsive than females.

Anaemia is a common complication in both cancer and CKD patients and it could contribute to a poor prognosis. ESA therapy represents the main treatment to increase Hb levels in such groups of patients, leading to improvement of quality of life and reducing the risk of cardio- and cerebrovascular complications, as well as the requirement of blood transfusions. However, ESA therapy must be carefully handled due to the increased risk of stroke in older patients having Hb levels above the target range. Indeed, due to the occurrence of ESA resistance, the need for higher doses of ESA may increase the risk of developing cardiovascular diseases and, ultimately, death [13]. Moreover, Minutolo et al. demonstrated that ESA hyporesponsiveness increased the risk of end stage renal disease by 2.5-fold in CKD patients [14].

Our data confirmed that the inflammatory condition and the iron intake affect ESA response. Inflammatory cytokines may affect the development of anemia through suppression of bone marrow erythropoiesis, suppression of erythropoietin production, or interfering with the iron status [15]. Several published studies demonstrated that high levels of CRP in hemodialysis patients were associated with ESA hyporesponsiveness, leading to an increased risk of death [16‐18]. Although the prevalence of ESA hyporesponsiveness in hemodialysis patients is similar to that found in non-dialysis patients, limited studies on the predictors of ESA hyporesponsiveness have been conducted in the latter population [14].

Regarding the iron intake, our results highlighted that the use of iron preparations was a predictive factor of ESA response, whilst serum iron and ferritin were not independently associated to responsiveness to ESA treatment. Although previous studies on hemodialysis patients demonstrated that an altered iron status (in terms of low transferrin saturation levels and/or low ferritin levels), is a common factor inducing ESA hyporesponsiveness [19], there is no general consensus regarding the role of iron status as a predictor of ESA response. A recent study examined the relationship between iron markers, such as transferrin saturation and ferritin levels, and ESA responsiveness. Finding from the study highlighted that transferrin saturation, but not ferritin, was statistically associated to ESA hyporesponsiveness [20]. On the hand, in our study, transferrin saturation covariate was excluded from the analysis, due to the high proportion of missing values (> 50%).

Minutolo et al. studied for the first time the risk of end-stage renal disease in CKD patients, who were hyporesponsive to ESA treatment. The study findings demonstrated that ESA hyporesponsiveness correlated to an increased risk of end-stage renal disease and the authors suggested that high ESA doses, together with the persistence of anemia, could lead to hypoxia, tubular atrophy and interstitial fibrosis, thus causing the progression of the renal damage. On the other hand, no correlation between the iron markers, CRP levels, serum parathyroid hormone, body mass index and ESA response was found [14].

The influence of gender on ESA response is still controversial. Female gender was associated with ESA hyporesponsiveness in our study cohort. This result is in line with previous studies [19, 21], and may be related to the underlying differences in iron release from reticuloendothelial cells between the two genders [22]. Conversely, other studies demonstrated that males were more likely to be ESA hyporesponsive, in comparison to women [23, 24].

In our study, high doses of ACE inhibitor and/or ARBs were related to ESA responsiveness. This data has been controversially discussed in previous papers. Several studies showing that ACE inhibitors and ARBs are associated to an increase of ESA hyporesponsiveness [23, 25] hypothesized that these anti-hypertensive drugs may interfere with erythropoiesis. It is known that the activation of renin-angiotensin system enhances the erythropoietin productions [26], while its inhibition due to ACE inhibitors may exacerbate anaemia [27]. Moreover, it has been demonstrated that ACE inhibitors may cause an increase in serum N-acetyl-seryl-aspartyl-lysyl-proline (Ac-SDKP) levels, which inhibit the recruitment of pluripotent erythroid cells in bone marrow [28]. Other potential mechanisms by which the considered anti-hypertensive drugs may cause anemia are the serum reduction of specific cytokines, such as interleukin-12, and/or of insulin-like growth factor-1, which physiologically stimulate erythropoiesis [11].

Our results also demonstrated that ESA users with metabolic acidosis (pH < 7.38 and serum HCO³⁻ < 22 mmol/l) had a good ESA response. Due to the lack of evidence explaining such potential association between metabolic acidosis and ESA hyporesponsiveness, further investigations on this potential predictive factor are needed.

Considering cancer patients, we found that higher baseline Hb values were associated with ESA hyporesponsiveness (p-value = 0.007), together with the history of ischemic heart disease, although this correlation is close to be significant (p-value = 0.072). The role of cardiovascular diseases as predictors of ESA hyporesponsiveness has been previously studied [18, 29] in CKD patients and the most liable mechanism is related to an increased production of inflammatory cytokines, such as interleukins 1 and 6, Tumor Necrosis Factor and interferon, which induce apoptosis in erythroid progenitor cells and decrease the iron availability by stimulating hepcidin production [30]. Further analyses are, on the other hand, required to confirm the role of cardiovascular diseases as predictors of ESA hyporesponsiveness in cancer setting.

This study has several strengths. Firstly, we may explore data on ESA dispensing from two large Italian LHUs over a 7-year observation period. Secondly, thanks to the electronic therapeutic plans, information on the exact brand name, number of dispensed packages, and indication for use were available. Moreover, we could explore variations in Hb values as a result of ESA treatment, using real-world data from more than 1000 ESA users. Most of the previous randomized clinical trials were conducted considering CKD patients only, while our study explored the potential predictors of ESA hyporesponsiveness in both CKD and cancer patients. However, some limitations warrant caution. The high frequency of missing values for some variables considered into the study (namely: transferrin saturation for CKD, as well as ferritin and vitamin B12 for cancer) precluded the possibility to test the independent effect of these risk factors on the study outcome. Thus such an issue remains to be investigated in a specifically designed future cohort study. Furthermore, although we tested into the models a series of laboratory risk factors assessed proximally to the Hb measurement, the possibility of residual time dependent confounding due to unmeasured confounders cannot be excluded. Some ESA as well as concomitant drugs (i.e. iron preparations) dispensing might not have been fully captured by the LHUs databases, as these drugs may be initially dispensed directly by the public hospitals or purchased by patients as out of pocket, thus not being traced using the study data sources. However, it is unlikely that this limitation affected the study results, as the potential selection bias is expected to be minimal and non-differential between ESAs responders and hyporesponders.

Finally, since the exact body weight of each ESA user was not available and we could not evaluate the exact ESA dosing regimen, we defined ESA hyporesponsiveness as a decrease in Hb levels and, in the sensitivity analysis, as the failure in achieving Hb values ≥11 g/dL, as reported by Suttorp et al. in a multi-center, prospective study [7].

This study tries to identify some potential predictive factors associated with ESA hyporesponsiveness. Covariates as serum CRP or high levels of baseline Hb were confirmed to be associated with poor response to ESA. A better knowledge of the factors associated with ESA response may help avoiding the use of higher ESA doses, and allow a better anaemia management in order to improve the patients’ quality of life and reduce morbidity and mortality of both CKD and cancer patients.