Raised FGF23 Correlates to Increased Mortality in Critical Illness, Independent of Vitamin D

Abstract

Background: Fibroblast Growth Factor (FGF23) is an endocrine hormone classically associated with the homeostasis of vitamin D, phosphate, and calcium. Elevated serum FGF23 is a known independent risk factor for mortality in chronic kidney disease (CKD) patients. We aimed to determine if there was a similar relationship between FGF23 levels and mortality in critically ill patients.

Methods: Plasma FGF23 levels were measured by ELISA in two separate cohorts of patients receiving vitamin D supplementation: critical illness patients (VITdAL-ICU trial, n = 475) and elective oesophagectomy patients (VINDALOO trial, n = 76). Mortality data were recorded at 30 and 180 days or at two years, respectively. FGF23 levels in a healthy control cohort were also measured (n = 27).

Results: Elevated FGF23 (quartile 4 vs. quartiles 1-3) was associated with increased short-term (30 and 180 day) mortality in critical illness patients (p < 0.001) and long-term (two-year) mortality in oesophagectomy patients (p = 0.0149). Patients who died had significantly higher FGF23 levels than those who survived: In the critical illness cohort, those who died had 1194.6 pg/mL (range 0-14,000), while those who survived had 120.4 pg/mL (range = 15-14,000) (p = 0.0462). In the oesophagectomy cohort, those who died had 1304 pg/mL (range = 154-77,800), while those who survived had 644 pg/mL (range = 179-54,894) (p < 0.001). This was found to be independent of vitamin D or CKD status (critical illness p = 0.3507; oesophagectomy p = 0.3800). FGF23 levels in healthy controls were similar to those seen in oesophagectomy patients (p = 0.4802).

Conclusions: Elevated baseline serum FGF23 is correlated with increased mortality in both the post-oesophagectomy cohort and the cohort of patients with critical illness requiring intensive care admission. This was independent of vitamin D status, supplementation, or CKD status, which suggests the presence of vitamin D-independent mechanisms of FGF23 action during the acute and convalescent stages of critical illness, warranting further investigation.

Keywords: FGF23; critical illness; intensive care.

Figure 1

Vitamin D metabolism and the role of FGF23 (highlighted in turquoise). Green arrows indicate stimulus effect and may represent several enzymatic sub-steps, blue arrows indicate enzyme conversion to a functional 1,25(OH)2D3, orange arrows indicate degradation, and red arrows indicate downregulation/enzymatic blockade.

Figure 2

FGF23 levels in healthy controls vs. oesophagectomy patients. The median FGF23 (IQR) was 2057 pg/mL (307.1–4663.0) in healthy controls and 835.0 pg/mL (526.0–1929.0) in oesophagectomy patients (p = 0.4802). Critical illness patients were not included due to differences in FGF23 measurement protocols.

Figure 3

(A). Mortality of critical illness patients (VITdAL-ICU) at 6 months, based on their baseline (day 0) FGF23 levels. High baseline FGF23 levels resulted in 56% mortality, while low baseline FGF23 levels resulted in 26% mortality (chi-squared test with Yates correction, p < 0.0001). FGF23 cut-off was at 334 pg/mL (Q4). The grey bars represent the % of patients deceased after 6 months, while the black bars represent the % of patients still alive. (B). Mortality of oesophagectomy patients (VINDALOO) at two years, based on their pre-operative FGF23 levels. High FGF23 levels resulted in 64% mortality, while low FGF23 levels resulted in 41% mortality (chi-squared test with Yates correction, p = 0.0149). FGF23 cut-off was at 1104 pg/mL (Q4). The grey bars represent the % of patients deceased after two years, while the black bars represent the % of patients still alive. (C). baseline FGF23 levels of deceased vs. surviving patients in the critical illness (VITdAL-ICU) cohort (alive: median 136.2 pg/mL, IQR 61.3–450.3; deceased: median 1195 pg/mL, IQR 174.0–5295; p = 0.0462) and the oesophagectomy (VINDALOO) cohort (alive: median 641 pg/mL, IQR 332.0–1444; deceased: median 999 pg/mL, IQR 603.5–4784; p < 0.0001). Log-converted FGF23 levels were used to calculate significance. Data are represented as median and IQR. (D). Mean 25(OH)2D3 levels in alive vs. deceased groups at day 0 for both cohorts. No significant difference was observed between the alive and deceased groups for either cohort (critical illness p = 0.8274; oesophagectomy p = 0.6056). Data are represented as median and IQR, * p < 0.05, **** (p < 0.001), ns = not significant.