Limited role for fibroblast growth factor 23 in assessing prognosis in heart failure patients: data from the TIME‐CHF trial

Aim: Fibroblast growth factor 23 (FGF23) is an intensively studied biomarker at the crossroads of cardiovascular disease, heart failure (HF) and chronic kidney disease. Independent associations between increasing FGF23 levels and cardiovascular events were found in many, but not all studies. By analysing data from the TIME‐CHF cohort, we sought to investigate the prognostic value of FGF23 in an elderly, multimorbid HF patient cohort. We determined differences between intact (iFGF23) and C‐terminal FGF23 (cFGF23) regarding their prognostic value and their levels over time in different HF subgroups according to left ventricular ejection fraction (LVEF). Methods and results: In this multicentre trial of 622 patients with symptomatic HF aged ≥60 years, we determined iFGF23 and cFGF23 at baseline, 3, 6 and 12‐month follow‐up. In unadjusted analyses, cFGF23 significantly predicted all HF‐related outcomes at all time points. The predictive value of iFGF23 was less and not statistically significant at baseline. After multivariable adjustments, the association between both cFGF23 and iFGF23 and outcome lost statistical significance apart from cFGF23 at month 3. Overall, patients with preserved and mid‐range LVEF had higher levels of iFGF23 and cFGF23 than those with reduced LVEF. Levels decreased significantly during the first 3 months in mid‐range and reduced LVEF patients, but did not significantly change over time in those with preserved LVEF. Conclusions: Fibroblast growth factor 23 is of limited value regarding risk prediction in this elderly HF population. Potentially heterogeneous roles of FGF23 in different LVEF groups deserve further investigation.

Keywords: Fibroblast growth factor 23; Prognosis; Heart failure; Hospitalisation; Mortality; Risk assessment

Introduction

The canonical role of the osteocyte‐derived circulating hormone fibroblast growth factor 23 (FGF23) is augmenting renal phosphate excretion by blocking tubular reabsorption and down‐regulation of 1,25 dihydroxy‐vitamin D synthesis.[1] However, the so‐called off‐target (more specifically cardiac) effects of FGF23 have reached a significant level of interest.[[2], [4]] Numerous cohort studies exist, describing the association of high circulating levels of FGF23 in serum with dismal outcome and prognosis in various patient populations, including patients with chronic kidney disease, patients with heart failure (HF), patients with acute coronary syndrome, critically ill patients, or in the general population.[[5], [7], [9], [11], [13], [15], [17]] These data have suggested an independent association between high FGF23 levels and the risk for increased cardiovascular morbidity events (such as hospitalization for HF) and mortality in all the investigated patient groups.[18] Notably, the link between FGF23 and HF‐associated endpoints was reported to be especially strong, while the association to atherosclerosis‐related cardiovascular endpoints was modest.[[19]] In HF patients the association between FGF23 levels and outcome was modified by variable Klotho levels and high FGF23 plus low Klotho levels showed specifically strong impact upon renin–angiotensin–aldosterone system inhibitor efficacy in terms of endpoint reduction.[21] Experimental in vivo data have supported these human cohort data by attributing a direct and causal role to FGF23 to left ventricular hypertrophy and dysfunction.[[22]]

The vast majority of previous studies investigating the association of serum FGF23 with outcome relied upon measuring C‐terminal FGF23 (cFGF23). Enzyme‐linked immunosorbent assays (ELISAs) measuring cFGF23 identify both the biologically active hormone (intact FGF23, iFGF23) as well as its degradation products. Early analysis performed in dialysis patients attributed a comparable value regarding risk assessment to both iFGF23 and cFGF23,[24] and a small cohort study in 100 HF patients recently confirmed this assumption.[14] However, the biological role of iFGF23 differs substantially from that of cFGF23[25] and, hence, relevant differences in the potency as biomarkers and risk predictors between the two forms are likely. Preliminary work in a cohort of patients undergoing coronary angiography for acute coronary syndrome suggested that the prognostic relevance of cFGF23 is superior for HF patients with reduced (HFrEF) compared to those with preserved ejection fraction (HFpEF),[26] but no previous study has so far taken into account the new European Society of Cardiology (ESC) ejection fraction strata of HF, which encompass three HF classes based on left ventricular ejection fraction (LVEF) into consideration.[27]

Several unsolved issues remain about the reliability of FGF23 as a biomarker in HF given that recent studies question a direct, causal and independent role of FGF23 as causal cardiac risk factor.[[28], [30]] Data from the TIME‐CHF (Trial of Intensified vs. Standard Medical Therapy in Elderly Patients With Congestive Heart Failure) cohort – a large prospective HF cohort with sophisticated long‐term follow‐up – can help to close these gaps.[[32]] TIME‐CHF offers the opportunity to target the following questions: Is there a relevant longitudinal change in FGF23 levels in HF patients? Does the reliability of FGF23 as a biomarker in HF differ between iFGF23 and cFGF23? Does the performance of FGF23 as a biomarker differ depending on strata of LVEF and HF subclass? Does FGF23 have a distinct role in prognosis assessment compared to established HF biomarkers such as B‐type natriuretic peptide (BNP)?

Methods

The TIME‐CHF cohort

Details about the TIME‐CHF study were described previously.[[32]] In brief, the study included patients aged 60 years or older with dyspnoea [New York Heart Association (NYHA) class ≥II, a history of hospitalization for HF within the last year, and an N‐terminal proBNP (NT‐proBNP) level of ≥400 ng/L in patients aged <75 years and a level of ≥800 ng/L in patients aged ≥75 years. Based on these criteria, 622 outpatients were included. Of these, 499 had an LVEF ≤45% by echocardiography, which was the cut‐off for the definition of reduced LVEF in the original publication.[32] For the present analysis, patients were re‐classified into three groups based on LVEF as suggested by the most recent ESC guideline definition[27]; i.e. LVEF <40% = HFrEF, LVEF 40–49% = HF with mid‐range ejection fraction (HFmrEF), LVEF ≥50% = HFpEF. TIME‐CHF randomized patients to intensified NT‐proBNP‐guided vs. standard symptom‐guided therapy for 18 months, with further follow‐up for up to 5.5 years. Both treatment groups were stratified into two age groups of 60–74 years and ≥75 years. The study was approved by the Ethics Committees of each centre, and each patient gave written informed consent before entering the study. The primary result of the TIME‐CHF trial was that HF therapy guided by NT‐proBNP or symptom‐guided therapy did not significantly improve the primary endpoint.[34] For each patient, every hospitalization, including cause of hospitalization and mortality, was recorded up to 5.5 years.[35]

Biochemistry

Blood samples were taken and stored at −80°C according to standard procedures.

From plasma samples collected at baseline, 3, 6 and 12 months, iFGF23 and cFGF23 were measured by TECO Medical using commercially available ELISAs (Immutopics, San Clemente, CA, USA).[[36]] Intra‐assay coefficient of variation (CV) for iFGF23 was 4.1% for 43 pg/mL and 2.0% for 426 pg/mL, while inter‐assay CV were 9.1% for 46 pg/mL and 3.5% for 441 pg/mL. For cFGF23, CV was 2.4% for 33.7 RU/mL and 1.4% for 302 RU/mL, while inter‐assay CV were 4.7% for 33.7 RU/mL and 2.4% for 293 pg/mL. Additional biomarkers were measured as previous described.[38] The present analysis is based on the currently available content of the biobank: not in all patients at all visits plasma was left and could be analysed. The number of included patients is reported for each analysis (refer to the Results section).

C‐terminal FGF23 was measured in RU/mL, which was converted to pg/mL by multiplying by 1.5 and further converted to pmol/L by multiplying by 0.03831. Intact FGF23 was measured in pg/mL, which was converted to pmol/L by multiplying by 0.03831.

Statistical analysis

Data were analysed using SPSS version 25.0 (IBM SPSS Inc., Chicago, IL, USA). Normally distributed data are reported as mean and standard deviations (SD), non‐normally distributed data as median and interquartile range (IQR). Both cFGF23 and iFGF23 values were not normally distributed and were therefore log‐transformed for further analysis. Categorical data are presented as proportions (%).

Differences between groups were assessed using analysis of variance (ANOVA) for continuous data with a normal distribution, and Kruskal–Wallis test was used for continuous data with no normal distribution. Pearson χ2 test was used for categorical data. A mixed model for repeated measures (RM) with a between subject factor was used to assess longitudinal biomarker variance and differences between the three groups. Mauchly's test of variance was used to assess if sphericity was violated, and if this was the case, the Greenhouse–Geisser correction was used. Analyses were corrected for multiple testing comparisons using the Bonferroni method.

First, a RM analysis from the original data was performed in the three groups, excluding patients with any missing data. Second, the RM analysis was performed after multiple imputation (n = 5) of missing biomarker concentrations at follow‐up. Herein, missing biomarker levels of non‐deceased subjects were replaced and depicted as separate lines to show the potential influence of missing data.

Survival was measured using the Kaplan–Meier method. Hazard ratios were calculated using Cox regression analyses for absolute biomarker concentrations. Spearman correlation was used to test associations between both cFGF23 and iFGF23 and other biomarkers. All P‐values were two sided. A P‐value of <0.05 was considered to be statistically significant. Data were analysed using SPSS version 25.0 (IBM SPSS Inc.). Power calculations for our Cox proportional hazard models were performed using XL‐STAT; the method used is described in its package insert (https://www.xlstat.com/en/solutions/features/cox‐model). The following input was used herein: SD of log‐transformed cFGF23 which was 0.71; event rates and sample size as described above; alpha was set at 0.05. For multivariable models, the power calculation additionally uses the R2 for a linear regression model of the included covariates regressed to log cFGF23 which was 0.17.

Results

Comparison between patients with heart failure with reduced, mid‐range and preserved ejection...

Differences regarding baseline characteristics (anthropomorphic and clinical data) between these three groups based on LVEF are listed in Table1. Compared to HFrEF patients, patients with HFpEF were older, more likely to be female, were less likely to have an ischaemic cause of HF, and were less likely to be smokers. They had higher systolic blood pressure and a higher prevalence of atrial fibrillation. Characteristics of patients with HFmrEF were overall intermediate between HFrEF and HFpEF. Table2 depicts serum levels of cardiac, renal, and inflammatory biomarkers at baseline stratified within the three strata of HF. Compared to HFrEF patients, patients with HFmrEF showed higher levels of cFGF23 and NT‐proBNP, while renal function parameters and haemoglobin levels were lower (Table2). Intact FGF23 levels at baseline were not significantly different between the three HF classes.

Baseline characteristics in patients with heart failure with reduced (<40%), mid range (40–49%) and preserved ejection fraction (≥50%)

	HFrEF (n = 242)	HFmrEF (n = 67)	HFpEF (n = 73)	P‐value
Age (years)	76 ± 7	79 ± 6	80 ± 7	<0.001
Female gender	81 (34%)	29 (43%)	46 (63%)	<0.001
BMI (kg/m2)	25.2 ± 4.3	25.3 ± 4.6	27.4 ± 5.5	0.001
Ischaemic HF	142 (59%)	36 (54%)	26 (36%)	0.002
History of MI	117 (48%)	30 (45%)	24 (33%)	0.07
LVEF (%)	27 ± 6	42 ± 3	57 ± 6	<0.001
Smoking				0.004
Previous	123 (51%)	33 (49%)	26 (36%)
Current	24 (10%)	11 (16%)	3 (4%)
Diabetes	90 (37%)	26 (39%)	27 (37%)	0.97
Stroke/TIA	39 (16%)	9 (13%)	13 (18%)	0.78
COPD	50 (21%)	15 (22%)	11 (15%)	0.49
PAOD	41 (17%)	14 (21%)	17 (23%)	0.43
Cancer	29 (12%)	12 (18%)	11 (15%)	0.42
Osteoporosis	21 (9%)	8 (12%)	13 (18%)	0.09
Depression	31 (13%)	8 (12%)	6 (8%)	0.57
Kidney disease	138 (57%)	46 (69%)	44 (60%)	0.23
Anaemia	62 (26%)	30 (45%)	28 (38%)	0.004
Charlson score >3	97 (40%)	21 (31%)	27 (37%)	0.28
NYHA class				0.45
II	46 (19%)	19 (28%)	13 (18%)
III	157 (65%)	40 (60%)	50 (68%)
IV	39 (16%)	8 (12%)	10 (14%)
History of oedema	114 (48%)	35 (53%)	38 (54%)	0.56
Rales	96 (40%)	29 (44%)	38 (52%)	0.18
Elevated central venous pressure	84 (35%)	23 (35%)	23 (32%)	0.95
Systolic BP (mmHg)	116 ± 18	128 ± 19	139 ± 23*	<0.001
Heart rate (bpm)	77 ± 14	73 ± 14	73 ± 12	0.10
Heart rhythm				0.03
Sinus rhythm	162 (67%)	35 (52%)	36 (49%)
Atrial fibrillation	66 (27%)	28 (42%)	31 (43%)
Pacemaker	13 (5%)	4 (6%)	6 (8%)

1 BMI, body mass index; BP, blood pressure; COPD, chronic obstructive pulmonary disease; HF, heart failure; HFmrEF, heart failure with mid‐range ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; LVEF, left ventricular ejection fraction; MI, myocardial infarction; NYHA, New York Heart Association; PAOD, peripheral arterial obstructive disease; TIA, transient ischaemic attack.

• 2 * P  < 0.05 vs. HFrEF,
• 3 † P  < 0.05 vs. HFmrEF.

Biomarker values in patients with heart failure with reduced (<40%), mid range (40–49%) and preserved ejection fraction (≥50%)

	HFrEF (n = 242)	HFmrEF (n = 67)	HFpEF (n = 73)	P‐value
iFGF23 (pmoL/L)	2.8 (1.9–4.2)	3.5 (2.4–5.3)	3.0 (2.1–4.1)	0.09
cFGF23 (pmoL/L)	13.1 (7.7–31.7)	23.0 (11.1–42.5)	15.5 (8.4–34.8)	0.01
NT‐proBNP (ng/L)	4189 (2152–7716)	5010 (2728–7335)	2141 (1497–4165)	<0.001
hs‐TnT (ng/L)	35 (19–60)	47 (22–118)	27 (17–44)	0.002
IL‐6 (ng/L)	7.1 (3.9–14.0)	8.4 (5.4–15.7)	8.1 (5.1–15.4)	0.19
hs‐CRP (mg/L)	6.7 (2.4–15.5)	8.7 (2.7–20.4)	8.2 (3.0–25.9)	0.25
eGFR (mL/min/1.73 m2)	53 (38–68)	46 (33–59)	51 (37–64)	0.03
Cystatin C (mg/L)	1.7 (1.4–2.1)	2.0 (1.7–2.4)	1.9 (1.6–2.3)	0.002
Uric acid (mmol/L)	7.8 (6.0–9.3)	7.9 (6.4–9.9)	7.6 (6.4–9.5)	0.51
BUN (mmol/L)	10.1 (8.0–13.6)	11.0 (7.4–14.6)	10.1 (7.8–13.7)	0.91
Haemoglobin (g/L)	133 (122–144)	122 (108–136)	119 (105–135)	<0.001
sTfR/log ferritin	1.9 (1.4–2.8)	1.9 (1.4–3.1)	2.0 (1.5–3.2)	0.24

• 4 Values are given as median (interquartile range).
• 5 BUN, blood urea nitrogen; cFGF23, C‐terminal fibroblast growth factor 23; eGFR, estimated glomerular filtration rate; HFmrEF, heart failure with mid‐range ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; hs‐CRP, high‐sensitivity C‐reactive protein C; hs‐TnT, high‐sensitivity troponin T; iFGF23, intact fibroblast growth factor 23; IL‐6, interleukin‐6; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; sTfR, soluble transferrin receptor.
• 6 * P  < 0.05 vs. HFrEF,
• 7 † P  < 0.05 vs. HFmrEF.

Correlation analysis to other biomarkers related to heart failure

Table3 depicts how both iFGF23 and cFGF23 correlate with traditional HF biomarkers and with other biochemical indicators for relevant co‐morbidities. Both iFGF23 and cFGF23 correlated positively with markers of cardiac damage and stress (NT‐proBNP and high sensitivity troponin), inflammation (interleukin‐6 and high‐sensitivity C‐reactive protein), as well as declining renal function (negative correlation with estimated glomerular filtration rate and positive correlation with cystatin C, uric acid, and urea). Anaemia and iron metabolism parameters also showed significant correlations to both subtypes of FGF23 (Table3).

Spearman ρ correlation between different biomarkers in all patients at baseline (n  = 382)

	iFGF23	cFGF23
NT‐proBNP	0.11	0.33
hs‐TnT	0.18	0.22
IL‐6	0.13	0.41
hs‐CRP	0.04	0.20
eGFR	−0.42	−0.39
Cystatin C	0.43	0.48
Uric acid	0.32	0.37
Urea	0.40	0.37
Haemoglobin	−0.07	−0.21
sTfR/log ferritin	0.21	0.49
cFGF23	0.44	–

• 8 cFGF23, C‐terminal fibroblast growth factor 23; eGFR, estimated glomerular filtration rate; hs‐CRP, high‐sensitivity C‐reactive protein C; hs‐TnT, high‐sensitivity troponin T; iFGF23, intact fibroblast growth factor 23; IL‐6, interleukin‐6; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; sTfR, soluble transferrin receptor.
• 9 * P  < 0.05,
• 10 † P  < 0.001.

Influence of co‐morbidities and severity of heart failure on fibroblast growth factor 23 leve...

Patients with diabetes, chronic obstructive pulmonary disease and ischaemic HF did not differ in their levels of both iFGF23 and cFGF23 compared to non‐diabetics, not having chronic obstructive pulmonary disease or patients with non‐ischaemic HF aetiology, respectively (Table4). Gender (higher in females), age (higher >75 years), atrial fibrillation, history of renal failure or anaemia, atherosclerosis, and NYHA class (increasing levels with higher NYHA class) had significant impact upon cFGF23 levels but only some of these factors upon iFGF23.

Intact and C‐terminal fibroblast growth factor 23 levels at baseline in subgroups, irrespective of left ventricular ejection fraction

	iFGF23 (pmoL/L)	cFGF23 (pmoL/L)
	No	Yes	P‐value	No	Yes	P‐value
Diabetes	3.0 (1.9–4.6)	3.0 (2.2–4.0)	0.67	14.7 (8.5–32.1)	17.1 (7.9–37.1)	0.49
COPD	3.0 (2.1–4.3)	2.8 (1.9–4.4)	0.40	15.8 (8.1–35.6)	12.1 (8.3–28.9)	0.48
History of renal failure	2.3 (1.7–3.5)	3.4 (2.4–5.0)	<0.001	10.5 (6.6–21.0)	20.7 (9.9–42.5)	<0.001
Atherosclerosis	2.7 (1.7–4.1)	3.1 (2.2–4.4)	0.02	12.1 (6.6–24.5)	16.7 (8.9–37.0)	0.01
History of anaemia	2.8 (2.0–4.2)	3.1 (2.1–4.8)	0.31	13.6 (6.8–30.3)	21.3 (10.4–44.9)	<0.001
Male gender	3.0 (2.0–4.4)	2.9 (2.0–4.3)	0.91	18.7 (9.6–40.2)	12.8 (7.5–30.4)	0.006
Age >75 years	2.8 (2.0–3.9)	3.2 (2.1–4.6)	0.08	11.3 (6.7–32.7)	17.3 (9.6–34.8)	0.004
Ischaemic HF	2.8 (2.1–4.1)	3.2 (1.9–4.6)	0.44	14.1 (7.4–32.8)	16.1 (8.9–35.1)	0.31
Atrial fibrillation	2.8 (1.9–4.0)	3.4 (2.3–4.9)	0.01	12.4 (6.7–27.4)	22.2 (11.3–48.0)	<0.001
NYHA class		0.13		<0.001
II	2.8 (2.1–3.7)		9.4 (6.5–23.3)
III	3.0 (2.0–4.5)		15.7 (8.9–32.8)
IV	3.2 (2.3–5.3)		20.9 (12.0–54.4)

• 11 Values are given as median (interquartile range).
• 12 cFGF23, C‐terminal fibroblast growth factor 23; COPD, chronic obstructive pulmonary disease; HF, heart failure; iFGF23, intact fibroblast growth factor 23; NYHA, New York Heart Association.

Prognostic impact of fibroblast growth factor 23 on cardiovascular outcome

Of the 382 patients included in this analysis, 137 (36%) died during follow‐up, 194 (51%) died or were hospitalized due to HF, and 300 (79%) died or were hospitalized due to any cause. In the patient group surviving 12 months (n = 324), 86 (27%) died, 138 (43%) either died or were hospitalized due to HF, and 250 (77%) were hospitalized due to any cause. The prognostic values of iFGF23 and cFGF23 regarding survival and HF‐related hospitalization are depicted in Table5. In unadjusted analyses, log‐transformed levels of cFGF23 concentration significantly predicted outcome. Log‐transformed levels of iFGF23 also predicted outcomes at most, but not all time points in unadjusted analysis (Table5). After multivariable adjustment including NT‐proBNP, the association between cFGF23 and outcome was less and for most analyses no longer of statistical significance (Table5). There was no significant difference regarding the predictive value of both iFGF23 and cFGF‐23 between the three HF strata (HFrEF, HFmrEF, HFpEF) for any outcome (data not shown). Sample size for univariable analysis was sufficient at all time points and for all outcomes with a power of >90%. Sample size for multivariable analyses was limited (i.e. <80%) when the effect size fell below a hazard ratio of approximately 1.5 at baseline and 1.6 at month 12 for a 10‐fold increase in FGF23 levels.

Hazard ratio per log 10 of intact and C‐terminal fibroblast growth factor 23 in all patients at baseline, 3, 6 and 12 months regarding all‐cause survival, heart failure hospitalization free survival and all‐cause hospitalization free survival

	cFGF23		iFGF23
	Univariable Cox regression	Multivariable Cox regression	Univariable Cox regression	Multivariable Cox regression
	HR (95% CI)	P‐value	HR (95% CI)	P‐value	HR (95% CI)	P‐value	HR (95% CI)	P‐value
Survival
Baseline	2.09 (1.46‐3.01)	<0.001	1.14 (0.75‐1.74)	0.55	1.22 (0.68‐2.19)	0.51	0.52 (0.27‐0.99)	0.05
Month 3	3.58 (2.39‐5.37)	<0.001	2.27 (1.30‐3.96)	0.004	6.02 (2.96‐12.3)	<0.001	2.02 (0.71‐5.74)	0.19
Month 6	5.96 (3.68‐9.67)	<0.001	1.56 (0.77‐3.18)	0.22	4.69 (2.65‐8.28)	<0.001	0.92 (0.38‐2.21)	0.84
Month 12	1.96 (1.17‐3.28)	0.01	0.73 (0.34‐1.52)	0.39	3.43 (1.78‐6.62)	<0.001	1.12 (0.47‐2.66)	0.80
HF hospitalization free survival
Baseline	1.99 (1.46‐2.71)	<0.001	1.15 (0.80‐1.65)	0.45	1.27 (0.80‐2.03)	0.32	0.52 (0.31‐0.89)	0.02
Month 3	3.95 (2.74‐5.69)	<0.001	2.53 (1.57‐4.09)	<0.001	4.42 (2.45‐8.00)	<0.001	1.35 (0.59‐3.06)	0.48
Month 6	5.51 (3.75‐8.08)	<0.001	1.96 (1.15‐3.33)	0.01	3.69 (2.34‐5.84)	<0.001	0.82 (0.42‐1.58)	0.56
Month 12	2.85 (1.81‐4.48)	<0.001	1.35 (0.75‐2.42)	0.31	2.73 (1.53‐4.88)	0.001	0.96 (0.45‐2.03)	0.91
All‐cause hospitalization free survival
Baseline	1.50 (1.18‐1.91)	0.001	1.09 (0.83‐1.45)	0.54	1.46 (1.00‐2.13)	0.94	0.94 (0.61‐1.45)	0.78
Month 3	2.63 (1.96‐3.52)	<0.001	1.63 (1.13‐2.35)	0.008	2.44 (1.51‐3.93)	1.41	1.41 (0.75‐2.66)	0.29
Month 6	2.32 (1.68‐3.22)	<0.001	1.21 (0.79‐1.86)	0.38	2.40 (1.57‐3.65)	1.38	1.38 (0.79‐2.39)	0.26
Month 12	1.85 (1.30‐2.64)	0.001	0.93 (0.60‐1.43)	0.73	1.78 (1.12‐2.83)	0.95	0.95 (0.52‐1.74)	0.87

• 13 cFGF23, C‐terminal fibroblast growth factor 23; CI, confidence interval; HF, heart failure; iFGF23, intact fibroblast growth factor 23; HR, hazard ratio.
• 14 * Adjusted for log10 N‐terminal pro‐B‐type natriuretic peptide, New York Heart Association class, systolic blood pressure, serum creatinine, age, diabetes, history of cancer, history of anaemia.

Longitudinal changes of fibroblast growth factor 23 over time

In 560 patients [HFrEF: 369 (66%); HFmrEF: 96 (17%); HFpEF: 95 (17%)], iFGF23 and cFGF23 were measured at least once at any of the four follow‐up visits. As compared to those with both FGF23 forms measured at baseline (n = 382, online supplementary Table S1), the other 178 (32%) patients had less often a history of renal failure (49% vs. 60%, P = 0.02) or anaemia (20% vs. 31%, P = 0.006), were less symptomatic (NYHA class II, III, and IV, respectively: 32%, 57%, 11% vs. 20%, 65%, 15%; P = 0.007), less often ankle oedema (39% vs. 50%, P = 0.04) and haemoglobin was higher (136 ± 16 g/L vs. 128 ± 18 g/L, P < 0.001) and cystatin C lower [median 1.6 (IQR 1.4–2.0) vs. 1.8 (IQR 1.5–2.2), P = 0.007].

Online supplementaryTable S1 depicts the levels of iFGF23 and cFGF23 in HFrEF, HFmrEF and HFpEF at different time points during follow‐up. Online supplementary FiguresS1 and S2 show the log‐transformed mean values of the available data as well five additional analyses with imputed data. Online supplementary FigureS1 shows that the observed patterns over time for cFGF23 in the three groups based on LVEF were not significantly influenced by missing data. In contrast, accuracy of imputation was less for iFGF23, particularly in patients with HFmrEF and HFpEF (online supplementary Figure S2).

Both iFGF23 and cFGF23 showed a decrease at month 3 of the study in HFrEF patients and remained stable or slightly increased later in follow‐up. A similar pattern was seen in HFmrEF patients regarding cFGF23, whereas in HFpEF changes were small. iFGF23 increased over time in both HFmrEF and HFpEF, but accuracy was less for these analyses due to missing data.

Discussion

Overall, the present TIME‐CHF data do not support the previous appraisal about FGF23 being a strong and independent predictor of HF‐related outcome and prognosis. In fact, the predictive power for both iFGF23 and cFGF23 was significantly reduced or disappeared with multifactorial adjustment including NT‐proBNP. Moreover, our data show that risk assessment analyses using FGF23 in HF should carefully differentiate between iFGF23 and cFGF23, since reliability and performance as well as associations with co‐morbidities were meaningfully different between the two forms.

The present data amplify and modify our understanding about FGF23 as a biomarker in HF through several novel aspects: we directly compared FGF23 levels across all three subclasses of HF based on LVEF (as defined by the 2016 ESC guidelines[27]). Of note, we detected significant differences in FGF23 blood levels between HFrEF, HFmrEF, and HFpEF patients. Most importantly, FGF23 did not increase linearly with declining LVEF suggesting that myocardial systolic dysfunction is not a major driving force for increasing FGF23 levels. Previous analyses have shown that other biomarkers also display different levels between HFpEF vs. HFrEF patients (e.g. C‐reactive protein, high‐sensitivity troponin, NT‐proBNP[38]). These data indicate that the different subclasses of HF exhibit varying amounts of activation of distinctive pathophysiological pathways.[38] This assumption is fuelled by remarkably different levels of co‐morbidities and co‐conditions between the three HF strata in TIME‐CHF, i.e. differences in age, gender, glomerular filtration rate, and body mass index as well as distinct prevalence of ischaemic heart disease, smoking, and anaemia and the influence of at least some of these factors on FGF23 levels. Interestingly, our group of HFmrEF patients showed similar baseline levels of troponin, NT‐proBNP, cystatin C and a lower glomerular filtration rate than HFrEF patients. Although this appears paradoxical, it most likely reflects the difficulty in classifying this new entity of HF patients, a phenomenon that has previously been reported and has sometimes been attributed to HF being a continuum and HFmrEF a transition phase to either HFpEF or HFrEF.

Indeed, several relevant co‐morbidities and concomitant conditions were associated with FGF23 level, e.g. signs of inflammation, declining renal function, and increasing levels of iron deficiency. Lower glomerular filtration rate, higher inflammatory markers and higher prevalence of anaemia were clearly related to higher FGF23 levels, specifically in the HFmrEF subgroup.[39] Similarly, there are some data suggesting that the association between FGF23 and incident HF is stronger for HFpEF than for HFrEF.[40] Since these associations were specifically strong for cFGF23 rather than for iFGF23 levels, we speculate that such co‐existing conditions could primarily influence the rate of FGF23 degradation rather than its synthesis. This hypothesis is supported by recent findings that iFGF23 and cFGF23 levels react differently to intravenous iron replenishment in HF patients with iron deficiency.[41] How exactly co‐existing conditions influence FGF23 metabolism needs to be determined and the biological relevance is still unclear.

Two other aspects support our skepticism about FGF23 as a stand‐alone and causal HF biomarker: first, the predictive power of cFGF23 was clearly superior to iFGF23, though the biological activity is stronger for the intact molecule.[25] Second, the prognostic impact of NT‐proBNP outperforms both forms of FGF23 in terms of hospitalization and mortality prediction. After multivariable adjustment, the association between FGF23 and outcome largely disappeared. Although statistical power may have not been sufficient to find small prognostic impact in our multivariable models, the potential multivariable effect size of FGF23 is so small that we would need a very large cohort to reveal a statistical significant contribution to the model. With regard to clinical relevance, we can therefore conclude that the association between FGF23 and adverse outcome is limited at best in the light of the prognostic value of other markers included in our multivariable models, e.g. NT‐proBNP. As such, the addition of FGF23 to the assessment of prognosis is neglectable.

Hence, the present data are in contrast to some of the previously published cohort studies which suggested a strong and independent association with all‐cause mortality, cardiovascular mortality, or need for hospitalization[[6], [12], [15]] even after adjustment for (NT‐pro) BNP. Our data are, however, in line with the CARE FOR HOMe study that initially showed a strong association between FGF23 and the development of HF in patients with chronic kidney disease,[19] but this association was lost when adjusting for NT‐proBNP.[42]

Previous studies investigating the role of FGF23 in HF patients were based only upon single‐spot FGF23 measurements at baseline. Noteworthy, TIME‐CHF also offers the exclusive opportunity to analyse serial FGF23 measurements in HF patients. In fact, the prognostic value of cFGF23 changed over time to some extent, which might explain some discrepant findings between studies. Two additional novel aspects arise: the longitudinal changes of iFGF23 and cFGF23 were not linear and numerically different between HFrEF, HFmrEF and HFpEF patients. Overall, we could detect no trend in the course of the two FGF23 forms and their blood levels were globally stable during follow‐up without meaningful changes. Taking the minor importance of baseline FGF23 in this setting into account, the small absolute FGF23 changes over time most presumably do not induce clinical sequel.

Our data are in line with recent experimental findings also challenging FGF23's role in myocardial disease, e.g. primary elevations of FGF23 were found to be innocent and harmless to the rodent heart.[31] Such data call into question the direction of causality and underline speculations that actually heart disease per se contributes to elevated FGF23 levels (pointing towards reverse causality).[[29]] Indeed, while early research suggested FGF23 to be a primarily renally controlled mediator with unfavourable cardiovascular action,[43] more recent data have called this previously held dogma into question. Indeed, it now seems that the already intricate relationship between the cardiovascular system and FGF23 is further complicated by the heart ability to release FGF23.[43]

In this respect, we want to refer to the various significant correlations in TIME‐CHF between both iFGF23 and cFGF23 and several other relevant cardiac, renal, and inflammatory biomarkers and co‐morbidities established in clinical routine. These correlations point towards the importance of thorough adjustments in multivariable analyses. Powerful associations to other cardiac risk factors highlight the need to exclude residual confounding.[28] Our data point towards a different strength of association between cardiac risk factors and iFGF23 and cFGF23, respectively. One potential explanation for this discrepancy might emerge from important pre‐analytical aspects: iFGF23, but not cFGF23, showed significant diurnal variation. C‐terminal FGF23 had a significantly lower intra‐individual variation than iFGF23 (8.3% vs. 18.3%) but higher inter‐individual variation than iFGF23 (28.9% vs. 19.2%). Fourteen samples would be needed to estimate an individual's homeostatic set point (within 10%) for iFGF23 compared with only three samples for cFGF23.[37] The degradation of iFGF23 towards cFGF23 underlies external influences which might add to this variability.[41]

There are limitations to our study. First of all, TIME‐CHF was mostly an elderly collective of patients which, however, most likely represents the typical population of HF patients. Secondly, most patients were included shortly after an episode of acute cardiac decompensation but as the patients were followed up over time, the 3‐month time point represents a stage of full compensation and assuming this as the baseline condition did not alter the results in a meaningful way. Thirdly, there was active medical management during the time course of the original study, potentially confounding some of the results. However, patients in both groups received intensifying treatment as recommended by the guidelines, representing a 'real‐life' situation. Furthermore, owing to the real‐life situation of the original study, in which blood was drawn at different time points during the day, we are unable to account for potential diurnal variations of FGF23 levels. Finally, although this is not the first study on the role of FGF23 in determining prognosis for HF patients, it adds to the understanding in the field, based on several novel features previously not or insufficiently addressed. Thus, we included a population that is representative for patients seen in clinical practice. In addition, we investigated both iFGF23 and cFGF23, which has not been done in a sufficiently large and/or defined cohort. Moreover, we investigated the whole range of LVEF. Finally, we studied repeated time points during follow‐up, which is important to understand the role of a biomarker.

Conclusion

We investigated iFGF23 and cFGF23 as a biomarker and risk predictor in HF patients within the intensively characterized prospective HF cohort TIME‐CHF. The present data regarding FGF23, its changes over time and its role in assessing the prognosis in HF are unique in several ways and modify our understanding about FGF23 as a biomarker by several novel and important aspects. Analysing cFGF23 or iFGF23 adds limited information to risk prediction and outcome analysis in HF patients. Additionally, our data challenge the hypothesis about a causal link of FGF23 elevations with myocardial disease. Hence, the TIME‐CHF data do not support FGF23 (neither iFGF23 nor cFGF23) as a clinically relevant biomarker in HF patients.

Funding

The biochemical and statistical analysis of the present manuscript were supported by an unrestricted grant from Novartis to Vincent Brandenburg/Robert Stöhr. TIME‐CHF was sponsored by the Horten Research Foundation (Lugano, Switzerland), as well as by smaller unrestricted grants from AstraZeneca Pharma, Novartis Pharma, Menarini Pharma, Pfizer Pharma, Servier, Roche Diagnostics, Roche Pharma, and Merck Pharma.

Conflict of interest: none declared.

GRAPH: Table S1. Comparison of intact and C‐terminal FGF23 in different groups of heart failure patients at different time points (all measurements considered).Figure S1. Longitudinal development of log‐transformed mean C‐terminal FGF23 levels stratified according to subclasses of left ventricular ejection fraction.Figure S2. Longitudinal development of log‐transformed mean intact FGF23 levels stratified according to subclasses of heart failure.

1 Silver J, Naveh‐Many T. FGF‐23 and secondary hyperparathyroidism in chronic kidney disease. Nat Rev Nephrol 2013 ; 9 : 641 – 649. 2 Vervloet MG, Massy ZA, Brandenburg VM, Mazzaferro S, Cozzolino M, Urena‐Torres P, Bover J, Goldsmith D ; CKD‐MBD Working Group of ERA‐EDTA. Bone: a new endocrine organ at the heart of chronic kidney disease and mineral and bone disorders. Lancet Diabetes Endocrinol 2014 ; 2 : 427 – 436. 3 Brandenburg VM, Vervloet MG, Marx N. The role of vitamin D in cardiovascular disease: from present evidence to future perspectives. Atherosclerosis 2012 ; 225 : 253 – 263. 4 Wolf M. Update on fibroblast growth factor 23 in chronic kidney disease. Kidney Int 2012 ; 82 : 737 – 747. 5 Scialla JJ, Xie H, Rahman M, Anderson AH, Isakova T, Ojo A, Zhang X, Nessel L, Hamano T, Grunwald JE, Raj DS, Yang W, He J, Lash JP, Go AS, Kusek JW, Feldman H, Wolf M ; Chronic Renal Insufficiency Cohort (CRIC) Study Investigators. Fibroblast growth factor‐23 and cardiovascular events in CKD. J Am Soc Nephrol 2014 ; 25 : 349 – 360. 6 Plischke M, Neuhold S, Adlbrecht C, Bielesz B, Shayganfar S, Bieglmayer C, Szekeres T, Horl WH, Strunk G, Vavken P, Pacher R, Hulsmann M. Inorganic phosphate and FGF‐23 predict outcome in stable systolic heart failure. Eur J Clin Invest 2012 ; 42 : 649 – 656. 7 Ix JH, Katz R, Kestenbaum BR, de Boer IH, Chonchol M, Mukamal KJ, Rifkin D, Siscovick DS, Sarnak MJ, Shlipak MG. Fibroblast growth factor‐23 and death, heart failure, and cardiovascular events in community‐living individuals: CHS (Cardiovascular Health Study). J Am Coll Cardiol 2012 ; 60 : 200 – 207. 8 Parker BD, Schurgers LJ, Brandenburg VM, Christenson RH, Vermeer C, Ketteler M, Shlipak MG, Whooley MA, Ix JH. The associations of fibroblast growth factor 23 and uncarboxylated matrix Gla protein with mortality in coronary artery disease: the Heart and Soul Study. Ann Intern Med 2010 ; 152 : 640 – 648. 9 Seiler S, Cremers B, Rebling NM, Hornof F, Jeken J, Kersting S, Steimle C, Ege P, Fehrenz M, Rogacev KS, Scheller B, Bohm M, Fliser D, Heine GH. The phosphatonin fibroblast growth factor 23 links calcium‐phosphate metabolism with left‐ventricular dysfunction and atrial fibrillation. Eur Heart J 2011 ; 32 : 2688 – 2696. Arnlov J, Carlsson AC, Sundstrom J, Ingelsson E, Larsson A, Lind L, Larsson TE. Serum FGF23 and risk of cardiovascular events in relation to mineral metabolism and cardiovascular pathology. Clin J Am Soc Nephrol 2013 ; 8 : 781 – 786. Taylor EN, Rimm EB, Stampfer MJ, Curhan GC. Plasma fibroblast growth factor 23, parathyroid hormone, phosphorus, and risk of coronary heart disease. Am Heart J 2011 ; 161 : 956 – 962. Ter Maaten JM, Voors AA, Damman K, van der Meer P, Anker SD, Cleland JG, Dickstein K, Filippatos G, van der Harst P, Hillege HL, Lang CC, Metra M, Navis G, Ng L, Ouwerkerk W, Ponikowski P, Samani NJ, van Veldhuisen DJ, Zannad F, Zwinderman AH, de Borst MH. Fibroblast growth factor 23 is related to profiles indicating volume overload, poor therapy optimization and prognosis in patients with new‐onset and worsening heart failure. Int J Cardiol 2018 ; 253 : 84 – 90. Udell JA, Morrow DA, Jarolim P, Sloan S, Hoffman EB, O'Donnell TF, Vora AN, Omland T, Solomon SD, Pfeffer MA, Braunwald E, Sabatine MS. Fibroblast growth factor‐23, cardiovascular prognosis, and benefit of angiotensin‐converting enzyme inhibition in stable ischemic heart disease. J Am Coll Cardiol 2014 ; 63 : 2421 – 2428. Gruson D, Ferracin B, Ahn SA, Rousseau MF. Head to head comparison of intact and C‐terminal fibroblast growth factor 23 in heart failure patients with reduced ejection fraction. Int J Cardiol 2017 ; 248 : 270 – 273. Poelzl G, Trenkler C, Kliebhan J, Wuertinger P, Seger C, Kaser S, Mayer G, Pirklbauer M, Ulmer H, Griesmacher A. FGF23 is associated with disease severity and prognosis in chronic heart failure. Eur J Clin Invest 2014 ; 44 : 1150 – 1158. Wohlfahrt P, Melenovsky V, Kotrc M, Benes J, Jabor A, Franekova J, Lemaire S, Kautzner J, Jarolim P. Association of fibroblast growth factor‐23 levels and angiotensin‐converting enzyme inhibition in chronic systolic heart failure. JACC Heart Fail 2015 ; 3 : 829 – 839. Leaf DE, Siew ED, Eisenga MF, Singh K, Mc Causland FR, Srivastava A, Ikizler TA, Ware LB, Ginde AA, Kellum JA, Palevsky PM, Wolf M, Waikar SS. Fibroblast growth factor 23 associates with death in critically ill patients. Clin J Am Soc Nephrol 2018 ; 13 : 531 – 541. Olauson H, Vervloet MG, Cozzolino M, Massy ZA, Urena Torres P, Larsson TE. New insights into the FGF23‐Klotho axis. Semin Nephrol 2014 ; 34 : 586 – 597. Seiler S, Rogacev KS, Roth HJ, Shafein P, Emrich I, Neuhaus S, Floege J, Fliser D, Heine GH. Associations of FGF‐23 and sKlotho with cardiovascular outcomes among patients with CKD stages 2‐4. Clin J Am Soc Nephrol 2014 ; 9 : 1049 – 1058. Panwar B, Judd SE, Wadley VG, Jenny NS, Howard VJ, Safford MM, Gutierrez OM. Association of fibroblast growth factor 23 with risk of incident coronary heart disease in community‐living adults. JAMA Cardiol 2018 ; 3 : 318 – 325. Bergmark BA, Udell JA, Morrow DA, Jarolim P, Kuder JF, Solomon SD, Pfeffer MA, Braunwald E, Sabatine MS. Klotho, fibroblast growth factor‐23, and the renin‐angiotensin system – an analysis from the PEACE trial. Eur J Heart Fail 2019 ; 21 : 462 – 470. Grabner A, Schramm K, Silswal N, Hendrix M, Yanucil C, Czaya B, Singh S, Wolf M, Hermann S, Stypmann J, Di Marco GS, Brand M, Wacker MJ, Faul C. FGF23/FGFR4‐mediated left ventricular hypertrophy is reversible. Sci Rep 2017 ; 7 : 1993. Grabner A, Amaral AP, Schramm K, Singh S, Sloan A, Yanucil C, Li J, Shehadeh LA, Hare JM, David V, Martin A, Fornoni A, Di Marco GS, Kentrup D, Reuter S, Mayer AB, Pavenstadt H, Stypmann J, Kuhn C, Hille S, Frey N, Leifheit‐Nestler M, Richter B, Haffner D, Abraham R, Bange J, Sperl B, Ullrich A, Brand M, Wolf M, Faul C. Activation of cardiac fibroblast growth factor receptor 4 causes left ventricular hypertrophy. Cell Metab 2015 ; 22 : 1020 – 1032. Gutierrez OM, Mannstadt M, Isakova T, Rauh‐Hain JA, Tamez H, Shah A, Smith K, Lee H, Thadhani R, Juppner H, Wolf M. Fibroblast growth factor 23 and mortality among patients undergoing hemodialysis. N Engl J Med 2008 ; 359 : 584 – 592. Goetz R, Nakada Y, Hu MC, Kurosu H, Wang L, Nakatani T, Shi M, Eliseenkova AV, Razzaque MS, Moe OW, Kuro‐o M, Mohammadi M. Isolated C‐terminal tail of FGF23 alleviates hypophosphatemia by inhibiting FGF23‐FGFR‐klotho complex formation. Proc Natl Acad Sci U S A 2010 ; 107 : 407 – 412. Koller L, Kleber ME, Brandenburg VM, Goliasch G, Richter B, Sulzgruber P, Scharnagl H, Silbernagel G, Grammer TB, Delgado G, Tomaschitz A, Pilz S, Berger R, Mortl D, Hulsmann M, Pacher R, Marz W, Niessner A. Fibroblast growth factor 23 is an independent and specific predictor of mortality in patients with heart failure and reduced ejection fraction. Circ Heart Fail 2015 ; 8 : 1059 – 1067. Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JG, Coats AJ, Falk V, Gonzalez‐Juanatey JR, Harjola VP, Jankowska EA, Jessup M, Linde C, Nihoyannopoulos P, Parissis JT, Pieske B, Riley JP, Rosano GM, Ruilope LM, Ruschitzka F, Rutten FH, van der Meer P. 2016 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure: The Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC). Developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur J Heart Fail 2016 ; 18 : 891 – 975. Marthi A, Donovan K, Haynes R, Wheeler DC, Baigent C, Rooney CM, Landray MJ, Moe SM, Yang J, Holland L, di Giuseppe R, Bouma‐de Krijger A, Mihaylova B, Herrington WG. Fibroblast growth factor‐23 and risks of cardiovascular and noncardiovascular diseases: a meta‐analysis. J Am Soc Nephrol 2018 ; 29 : 2015 – 2027. Matsui I, Oka T, Kusunoki Y, Mori D, Hashimoto N, Matsumoto A, Shimada K, Yamaguchi S, Kubota K, Yonemoto S, Higo T, Sakaguchi Y, Takabatake Y, Hamano T, Isaka Y. Cardiac hypertrophy elevates serum levels of fibroblast growth factor 23. Kidney Int 2018 ; 94 : 60 – 71. Takahashi H, Ozeki M, Fujisaka T, Morita H, Fujita SI, Takeda Y, Shibata K, Sohmiya K, Hoshiga M, Tamaki J, Ishizaka N. Changes in serum fibroblast growth factor 23 in patients with acute myocardial infarction. Circ J 2018 ; 82 : 767 – 774. Pastor‐Arroyo EM, Gehring N, Krudewig C, Costantino S, Bettoni C, Knopfel T, Sabrautzki S, Lorenz‐Depiereux B, Pastor J, Strom TM, Hrabe de Angelis M, Camici GG, Paneni F, Wagner CA, Rubio‐Aliaga I. The elevation of circulating fibroblast growth factor 23 without kidney disease does not increase cardiovascular disease risk. Kidney Int 2018 ; 94 : 49 – 59. Brunner‐La Rocca HP, Buser PT, Schindler R, Bernheim A, Rickenbacher P, Pfisterer M ; TIME‐CHF Investigators. Management of elderly patients with congestive heart failure – design of the Trial of Intensified versus standard Medical therapy in Elderly patients with Congestive Heart Failure (TIME‐CHF). Am Heart J 2006 ; 151 : 949 – 955. Pfisterer M, Buser P, Rickli H, Gutmann M, Erne P, Rickenbacher P, Vuillomenet A, Jeker U, Dubach P, Beer H, Yoon SI, Suter T, Osterhues HH, Schieber MM, Hilti P, Schindler R, Brunner‐La Rocca HP ; TIME‐CHF Investigators. BNP‐guided vs symptom‐guided heart failure therapy: the Trial of Intensified vs standard Medical therapy in Elderly patients with Congestive Heart Failure (TIME‐CHF) randomized trial. JAMA 2009 ; 301 : 383 – 392. Zoppellaro G, Faggin E, Puato M, Pauletto P, Rattazzi M. Fibroblast growth factor 23 and the bone‐vascular axis: lessons learned from animal studies. Am J Kidney Dis 2012 ; 59 : 135 – 144. Sanders‐van Wijk S, Maeder MT, Nietlispach F, Rickli H, Estlinbaum W, Erne P, Rickenbacher P, Peter M, Pfisterer MP, Brunner‐La Rocca HP ; TIME‐CHF Investigators. Long‐term results of intensified, N‐terminal‐pro‐B‐type natriuretic peptide‐guided versus symptom‐guided treatment in elderly patients with heart failure: five‐year follow‐up from TIME‐CHF. Circ Heart Fail 2014 ; 7 : 131 – 139. Gutierrez O, Isakova T, Rhee E, Shah A, Holmes J, Collerone G, Juppner H, Wolf M. Fibroblast growth factor‐23 mitigates hyperphosphatemia but accentuates calcitriol deficiency in chronic kidney disease. J Am Soc Nephrol 2005 ; 16 : 2205 – 2215. Smith ER, Cai MM, McMahon LP, Holt SG. Biological variability of plasma intact and C‐terminal FGF23 measurements. J Clin Endocrinol Metab 2012 ; 97 : 3357 – 3365. Sanders‐van Wijk S, van Empel V, Davarzani N, Maeder MT, Handschin R, Pfisterer ME, Brunner‐La Rocca HP ; TIME‐CHF Investigators. Circulating biomarkers of distinct pathophysiological pathways in heart failure with preserved vs. reduced left ventricular ejection fraction. Eur J Heart Fail 2015 ; 17 : 1006 – 1014. David V, Martin A, Isakova T, Spaulding C, Qi L, Ramirez V, Zumbrennen‐Bullough KB, Sun CC, Lin HY, Babitt JL, Wolf M. Inflammation and functional iron deficiency regulate fibroblast growth factor 23 production. Kidney Int 2016 ; 89 : 135 – 146. Almahmoud MF, Soliman EZ, Bertoni AG, Kestenbaum B, Katz R, Lima JAC, Ouyang P, Miller PE, Michos ED, Herrington DM. Fibroblast growth factor‐23 and heart failure with reduced versus preserved ejection fraction: MESA. J Am Heart Assoc 2018 ; 7 : e008334. Stohr R, Sandstede L, Heine GH, Marx N, Brandenburg V. High‐dose ferric carboxymaltose in patients with HFrEF induces significant hypophosphatemia. J Am Coll Cardiol 2018 ; 71 : 2270 – 2271. Emrich IE, Brandenburg V, Sellier AB, Schauerte J, Wiedenroth J, Untersteller K, Lennartz CS, Seiler‐Mussler S, Wagenpfeil S, Fliser D, Heine GH. Strength of fibroblast growth factor 23 as a cardiovascular risk predictor in chronic kidney disease weaken by proBNP adjustment. Am J Nephrol 2019 ; 49 : 203 – 211. Stohr R, Schuh A, Heine GH, Brandenburg V. Corrigendum: FGF23 in cardiovascular disease: innocent bystander or active mediator? Front Endocrinol (Lausanne) 2018 ; 9 : 422.

By Robert Stöhr; Vincent M. Brandenburg; Gunnar H. Heine; Micha T. Maeder; Gregor Leibundgut; Alexander Schuh; Urs Jeker; Matthias Pfisterer; Sandra Sanders‐van Wijk and Hans‐Peter Brunner‐la Rocca

Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author