Introduction
Chronic kidney disease (CKD) is a globally significant burden that affects society, with an estimated prevalence of 10–13% worldwide. Moreover, the number of patients with CKD requiring renal replacement therapy is estimated to be 4.9–7.0 million [
1,
2]. Anemia is a common complication among patients with CKD not requiring dialysis [
3,
4]. The prevalence of anemia increases as the stage of CKD progresses, from 8% at stage 1 to 53% at stage 5 [
5]. Observational studies have suggested that anemia may be a biomarker independently associated with increased cardiovascular (CV) and kidney events [
6‐
8]. Erythropoiesis-stimulating agents (ESAs) have been widely used to treat renal anemia in patients with CKD on dialysis and those not requiring dialysis. However, interventional studies using ESAs in patients with CKD not requiring dialysis have reported conflicting results [
9‐
14], and the optimal target hemoglobin levels for patients with CKD are unknown [
10‐
13].
A small randomized controlled trial (RCT) conducted by Gouva et al. has demonstrated favorable effects of early intervention using erythropoietin alfa on renal outcomes [
9]. Thereafter, three large RCTs failed to show the clinical benefits of targeting higher hemoglobin levels. In the Cardiovascular Risk Reduction by Early Anemia Treatment with Epoetin Beta (CREATE) trial, using epoetin beta to target high hemoglobin levels (13.0–15.0 g/dl) vs. low hemoglobin levels (10.5–11.5 g/dl) also failed to reduce the incidence of CV events, while the number of patients requiring dialysis therapy significantly increased [
11]. In the Correction of Hemoglobin and Outcomes in Renal Insufficiency (CHOIR) trial, patients with CKD not requiring dialysis were randomly assigned to target either a high (13.0–13.5 g/dl) or a low hemoglobin level (10.5–11.0 g/dl) using epoetin alfa [
12]. However, targeting higher hemoglobin levels was associated with a significantly higher risk of a composite outcome of death and CV events. In the Trial to Reduce Cardiovascular Events with Aranesp Therapy (TREAT), patients with CKD not requiring dialysis who have diabetes were randomly assigned to a placebo and a group receiving darbepoetin alfa to achieve a hemoglobin level of ≥ 13 g/dl [
13]. Similarly, darbepoetin alfa failed to reduce the risk of the two primary composite outcomes (CV outcome [death or a CV event] and renal outcome [death or a renal event]). However, it significantly increased the risk of stroke. A meta-analysis report has concluded that targeting higher hemoglobin levels with ESAs reduces the need for blood transfusions but increases the risk of CV and kidney events in patients with CKD [
16,
17].
Recently, hypoxia-inducible factor prolyl-hydroxylase inhibitors (HIF-PHIs) have been available. Almost all the studies have successfully demonstrated that the efficiency of HIF-PHIs in treating anemia is comparable with that of ESAs [
18‐
21]. However, no clinical studies have ever shown the clear advantage of HIF-PHIs over ESAs or placebo on the kidney or CV outcomes in patients with CKD not requiring dialysis. Moreover, the optimal target hemoglobin levels using HIF-PHIs for patients with CKD not requiring dialysis have not been previously reported [
22].
Tsubakihara et al. conducted an RCT in Japan to investigate the renal protective effects [
14,
15]. Although the primary analysis was negative, post-hoc analyses demonstrated that maintaining hemoglobin levels (11–13 g/dl) with darbepoetin alfa improved renal outcomes as compared to maintaining hemoglobin levels (9–11 g/dl) with epoetin alfa in patients at stage 5 CKD not requiring dialysis, particularly those without diabetes. Based on these findings, we conducted an RCT, namely ‘Prevention of end-stage kidney disease (ESKD) by Darbepoetin Alfa in CKD Patients with Non-diabetic Kidney Disease (PREDICT) trial.’ This trial aimed to prove our hypothesis that targeting a higher hemoglobin level (11–13 g/dl) with darbepoetin alfa would prevent ESKD as compared to targeting a lower hemoglobin level (9–11 g/dl) in patients with advanced CKD without diabetes [
23]. The primary analysis revealed that targeting a higher hemoglobin level did not significantly improve renal outcomes compared with targeting a lower hemoglobin level [
24]. It is noteworthy that in the PREDICT trial, the prognosis of the high-hemoglobin group demonstrated a tendency to improve (hazard ratio [HR] 0.78; 95% confidence interval [CI] 0.60–1.03), whereas the prognosis in the three studies mentioned above [
11‐
13] yielded opposite results.
In this prespecified secondary analysis of the PREDICT trial, we aimed to further clarify the effects of targeting hemoglobin levels using darbepoetin alfa on renal outcomes, including the primary endpoint of composite renal endpoint in the per-protocol set (PPS) and the secondary endpoints of eGFR and proteinuria slopes in the full analysis set (FAS) and PPS, in patients with advanced CKD without diabetes.
Discussion
In this prespecified secondary analysis of the PREDICT trial, we found that targeting a higher hemoglobin level at 11–13 g/dl with darbepoetin alfa did not improve eGFR and proteinuria slopes as compared to targeting a lower hemoglobin level at 9–11 g/dl in patients with advanced CKD without diabetes. These were in line with the results of the primary analysis in our study that a higher hemoglobin target did not reduce the composite renal outcome [
24]. On the other hand, the PPS analysis that excluded patients with off-target hemoglobin levels demonstrated a 36% reduction in the composite renal endpoint and an improvement in eGFR slope of 1 ml/min/1.73 m
2/year in the high-hemoglobin group compared to the low-hemoglobin group. These were in contrast to the results of previous RCTs indicating that higher hemoglobin targeting was linked to a relatively worse prognosis [
11‐
13,
16].
In the FAS analyses of the eGFR and proteinuria slopes, which was the prespecified secondary endpoints, we could not demonstrate the clear benefit of higher hemoglobin targeting using darbepoetin alfa. The composite renal endpoint was reduced by 22% in the high-hemoglobulin group (vs. low-hemoglobin group) as reported previously [
24], and the eGFR decline was 0.59 ml/min/1.73 m
2/year smaller as presented in this study, although not statistically significant (
P = 0.08 and
P = 0.075, respectively). Regarding the proteinuria slope analysis, the mean baseline value was only 58 mg/dL. Therefore, it is difficult to draw any conclusion about the anti-proteinuric effects in this study. Our previous report has also revealed that the rates of CV events were not significantly different between the groups. The PPS analyses in this study suggested that high-hemoglobin targeting was associated with better kidney outcomes among patients with advanced CKD without diabetes and who maintained the target hemoglobin levels without violating the protocol. Darbepoetin alfa, when used properly to maintain the target hemoglobin level at 11–13 g/dl, may exert good effects on the kidney as previously described [
14,
15].
As aforementioned, a meta-analysis has demonstrated the potential harm of targeting a higher hemoglobin level (> 13 g/dl) using ESAs in patients with CKD not requiring dialysis [
16]. One of the main differences was the target level in the high-hemoglobin group; the target in the three major RCTs was > 13 g/dl [
11‐
13], whereas in our study, the target was not normalization but the middle range at 11–13 g/dl. Taken together with our previous report [
24], the PREDICT study demonstrates that maintaining hemoglobin levels at 11–13 g/dl compared with 9–11 g/dl did not at least worsen the prognosis of the patients.
However, the results of the PPS analysis need to be interpreted with caution. Hypo-responsiveness to ESAs was reported to be associated with worse kidney outcomes and CV prognosis [
25,
26]. Patients unable to achieve the target hemoglobin level may have been hypo-responsive to ESAs. Especially in the high-hemoglobin group, 57 out of 200 patients were excluded because the lower hemoglobin levels than the target. Therefore, it is still difficult to conclude from the PPS analysis that maintaining hemoglobin levels at 11–13 g/dl protects the kidney more than at 9–11 g/dl.
Concerning the mechanisms of ESAs on the CV prognosis and renal outcomes, both beneficial and harmful effects have been postulated. The beneficial effects for organ protection may be independent of anemia correction but can be attributed to the non-hematological effects of rHuEPO that prevent tissue damage [
27]. The potential harm of ESAs may include elevated blood pressure, increased viscosity, and increased platelet number and aggregation [
28]. The balance of these factors may have led to different results in RCTs, including this study, depending on the setting of the trial, such as target hemoglobin levels, dosage and kinds of ESA reagents, as well as participants’ characteristics with different risks for CV and renal events. Two RCTs involving patients with CKD receiving kidney transplantation have revealed the renoprotective effects of maintaining normal hemoglobin levels with ESAs on renal outcomes [
29,
30]. The participants were relatively young and had a lower risk of CV events than those in other RCTs involving patients with CKD in the pre-dialysis stage.
Regarding the other factors, male sex and higher urinary protein levels at baseline were independently associated with worse renal outcomes in line with those of previous studies [
31,
32]. Our study also showed that older age was associated with better renal outcomes after adjusting for other clinical factors. This is probably because those who had hypertensive nephrosclerosis with slower CKD progression were relatively old, while those who had glomerulonephritis with rapid CKD progression were relatively young [
33].
Recently, HIF-PHIs have been utilized in treating renal anemia in patients with CKD not requiring dialysis. Many preclinical studies have shown the potential benefit of HIF-PHIs for ischemic organ damage, including the heart and kidney [
35‐
36]. To date, clinical studies have not demonstrated the detrimental effects of HIF-PHIs [
17‐
21,
37,
38]. Since the mechanism of HIF-PHIs function is different from that of ESAs [
34,
39,
40], the data of hemoglobin targeting studies obtained by ESA treatment cannot be directly applied to understand the potential benefit or harm of HIF-PHIs in the clinical setting. RCTs to analyze the effects of targeting high vs. low hemoglobin levels on the prognosis of patients with CKD not requiring dialysis using HIF-PHIs will provide clinically useful information. Due to the difference in the relatively lower incidence of CV events in patients with CKD in Japan than in Western countries [
41], clinical trials focusing on renal outcomes by targeting high hemoglobin levels with HIF-PHIs should be conducted in each region.
Our study had several potential limitations. First, only patients with CKD in Japan were included in the study, and Japan has a much lower incidence of CV events. Therefore, our results may underestimate the potential harm of CV diseases and may not be generalized to all patients with CKD in other places. Nevertheless, this could also be a strength of our study. Since the rates of CV events in our study were low, only 8% in the high- and 7% in the low-hemoglobin groups [
24], the net effects of higher hemoglobin levels on renal outcomes were focused on without considering the indirect effects of CV events. Second, the difference in achieved hemoglobin levels was not more significant than expected, leading to insufficient power to detect the group differences in outcomes. The PREDICT trial was planned and conducted based on the data from the previous study performed in Japan [
14], where the difference in the hemoglobin levels between high- and low-hemoglobin groups was approximately 2 g/dl. However, the actual difference in this study was only 1.2 g/dl in the FAS and 1.7 g/dl in the PPS. Third, a prespecified method for making up the PPS may not be the best. Patients who could not reach the higher target levels were excluded due to the violation of protocol. As mentioned above, the patients may have been hyporesponsive to ESA therapy and did not necessarily intend to violate protocols. Due to the selection bias from this, the PPS analysis may overestimate the true effects of targeting higher hemoglobin levels.
Acknowledgements
The authors would like to express our deepest gratitude to the patients, investigators, and staff at the study sites for their contribution to the study. The following investigators participated in the PREDICT trial (the numbers of enrolled patients are in parentheses): Terumasa Hayashi, Osaka General Medical Center (38); Takeyuki Hiramatsu, Konan Kosei Hospital (35); Hirofumi Tamai, Anjo Kosei Hospital (31); Yoshiyasu Iida, Yokkaichi Municipal Hospital (30); Tomohiro Naruse, Kasugai Municipal Hospital (18); Hideto Oishi, Komaki City Hospital (18); Shoichi Maruyama, Nagoya University Hospital (17); Shunya Uchida, Teikyo University School of Medicine (16); Hideaki Shimizu, Chubu Rosai Hospital (15); Kunio Morozumi, Japanese Red Cross Nagoya Daini Hospital (13); Hisashi Kurata, Toyota Kosei Hospital (12); Nobuhito Hirawa, Yokohama City University Medical Center (12); Saori Nishio, Hokkaido University Hospital (12); Yukio Yuzawa, Fujita Health University (11); Ichiei Narita, School of Medicine, Niigata University (10); Makoto Mizutani, Handa City Hospital (9); Isao Aoyama, Chukyo Hospital (9); Hideaki Yoshida, Sapporo Medical University Hospital (9); Kouji Kaneda, Oita Red Cross Hospital (9); Masaomi Nangaku, The University of Tokyo Graduate School of Medicine (8); Hideki Hirakata, Japanese Red Cross Fukuoka Hospital (8); Satoshi Suzuki, Kainan Hospital (8); Hiroki Adachi, Kanazawa Medical University Hospital (7); Eriko Kinugasa, Showa University, Northern Yokohama Hospital (7); Kei Kurata, Tosei General Hospital (7); Hiroshi Morinaga, Okayama University Hospital (6); Yusuke Tsukamoto, Itabashi Chuo Medical Center (6); Kazuhiro Tsuruya, Kyushu University Hospital (5); Ryoichi Ando, Musashino Red Cross Hospital (5); Shizunori Ichida, Japanese Red Cross Nagoya Daiichi Hospital (5); Teiichi Tamura, Yokosuka Kyosai Hospital (5); Takao Masaki, Hiroshima University Hospital (4); Takashi Wada, Kanazawa University Hospital (4); Hirokazu Honda, Showa University Koto Toyosu Hospital (4); Junichiro Yamamoto, Tsushima City Hospital (4); Yoshitaka Isaka, Osaka University Hospital (4); Eri Muso, Tazuke Kofukai Medical Research Institute, Kitano Hospital (4); Yasuhiro Komatsu, St. Luke's International Hospital (4); Norimi Ohashi, Ogaki Municipal Hospital (4); Taiga Hara, Kagawa University Hospital (4); Kiyoshi Ikeda, Ikeda Vascular Access, Dialysis and Internal Medicine Clinic (3); Kazuyoshi Okada, Nihon University School of Medicine (3); Tetsuhiko Yoshida, Hamanomachi Hospital (3); Seiya Okuda, Kurume University Hospital (3); Hiromichi Suzuki, Saitama Medical University (3); Takeshi Nakanishi, Hyogo College of Medicine (3); Harumichi Higashi, St Mary's Hospital (3); Arimasa Shirasaki, Ichinomiya Municipal Hospital (3); Shuichiro Endo, Kyoto University Hospital (2); Yutaka Osawa, Niigata Rinko Hospital (2); Ryuji Aoyagi, Tachikawa General Hospital (2); Yasuhiko Tomino, Juntendo University Hospital (2); Tetsu Akimoto, Jichi Medical University (2); Tsuyoshi Watanabe, Fukushima Medical University (2); Jiro Toyonaga, Iizuka Hospital (2); Motoko Tanaka, Akebono Clinic (2); Yoshitaka Ishibashi, Japanese Red Cross Medical Center (2); Shigehiro Uezono, Miyazaki Prefectural Miyazaki Hospital (2); Masako Sakakibara, Nagoya Memorial Hospital (2); Hajime Yamazaki, Nagaoka Red Cross Hospital (1); Hideki Takano, Tokyo Teishin Hospital (1); Hirofumi Ikeda, Munakata Medical Association Hospital (1); Takuma Takata, Nagaoka Chuo General Hospital (1); Hiroshi Yamashita, Toyota Memorial Hospital (1); Kunihiro Yamagata, University of Tsukuba (1); Toshinobu Sato, Japan Community Health Care Organization, Sendai Hospital (1); Ashio Yoshimura, Showa University Fujigaoka Hospital (1); Keiichi Tamagaki, Kyoto Prefectural University of Medicine (1); Kazuhiro Sonomura, Omihachiman Community Medical Center (1); Akira Iguchi, Ojiya General Hospital (1); Masahito Tamura, Hospital of the University Occupation and Environmental Health (1); Ryota Yasukawa, Sado General Hospital (1); Takanobu Morihiro, Masuko Hospital (1); Manei Oku, Ikeda Hospital (1).
Terumasa Hayashi, Takeyuki Hiramatsu, Hirofumi Tamai, Yoshiyasu Iida, Tomohiro Naruse, Hideto Oishi, Shoichi Maruyama, Shunya Uchida, Hideaki Shimizu, Kunio Morozumi, Hisashi Kurata, Nobuhito Hirawa, Saori Nishio, Yukio Yuzawa, Ichiei Narita, Makoto Mizutani, Isao Aoyama, Hideaki Yoshida, Kouji Kaneda, Masaomi Nangaku, Hideki Hirakata, Satoshi Suzuki, Hiroki Adachi, Eriko Kinugasa, Kei Kurata, Hiroshi Morinaga, Yusuke Tsukamoto, Kazuhiro Tsuruya, Ryoichi Ando, Shizunori Ichida, Teiichi Tamura, Takao Masaki, Takashi Wada, Hirokazu Honda, Junichiro Yamamoto, Yoshitaka Isaka, Eri Muso, Yasuhiro Komatsu, Norimi Ohashi, Taiga Hara, Kiyoshi Ikeda, Kazuyoshi Okada, Tetsuhiko Yoshida, Seiya Okuda, Hiromichi Suzuki, Takeshi Nakanishi, Harumichi Higashi, Arimasa Shirasaki, Shuichiro Endo, Yutaka Osawa, Ryuji Aoyagi, Yasuhiko Tomino, Tetsu Akimoto, Tsuyoshi Watanabe, Jiro Toyonaga, Motoko Tanaka, Yoshitaka Ishibashi, Shigehiro Uezono, Masako Sakakibara, Hajime Yamazaki, Hideki Takano, Hirofumi Ikeda, Takuma Takata, Hiroshi Yamashita, Kunihiro Yamagata, Toshinobu Sato, Ashio Yoshimura, Keiichi Tamagaki, Kazuhiro Sonomura, Akira Iguchi, Masahito Tamura, Ryota Yasukawa, Manei Oku