Elevated free hemoglobin and decreased haptoglobin levels are associated with adverse clinical outcomes, unfavorable physiologic measures, and altered inflammatory markers in pediatric cardiac surgery patients

Abstract

Background: There are data suggesting that free hemoglobin (Hb), heme, and iron contribute to infection, thrombosis, multiorgan failure, and death in critically ill patients. These outcomes may be mitigated by haptoglobin.

Study design and methods: 164 consecutively treated children undergoing surgery for congenital heart disease were evaluated for associations between free Hb and haptoglobin and clinical outcomes, physiologic metrics, and biomarkers of inflammation RESULTS: Higher perioperative free Hb levels (and lower haptoglobin levels) were associated with mortality, nosocomial infection, thrombosis, hours of intubation and inotropes, increased interleukin-6, peak serum lactate levels, and lower nadir mean arterial pressures. The median free Hb in patients without infection (30 mg/dL; 29 interquartile range [IQR], 24-52 mg/dL) was lower than in those who became infected (39 mg/dL; IQR, 33-88 mg/ 31 dL; p = 0.0046). The median mechanical ventilation requirements were 19 (IQR, 7-72) hours in patients with higher levels of haptoglobin versus 48 (IQR, 18-144) hours in patients with lower levels (p = 0.0047). Transfusion dose, bypass duration, and complexity of surgery were all significantly correlated with Hb levels and haptoglobin levels. Multivariate analyses demonstrated that these variables were independently and significantly associated with outcomes.

Conclusions: Elevated pre- and postoperative levels of free Hb and decreased levels of haptoglobin were associated with adverse clinical outcomes, inflammation, and unfavorable physiologic metrics. Transfusion, RACHS score, and duration of bypass were associated with increased free Hb and decreased haptoglobin. Further investigation of the role of hemolysis and haptoglobin as potential mediators or markers of outcomes is warranted.

Figure 1

Free hemoglobin at 12 hours post-operatively is shown as a boxplot (median and 10^th, 25^th, 75^th, 90^th percentiles) for patients who died (and for whom samples were available) (n=4) versus patients who survived (n=154). Two patients had no samples available or died before 12 hours. The graphs on the right are histograms displaying the relative number of patients in each range of free hemoglobins. The median free hemoglobin in survivors (31; interquartile range 24–63) was less than in those who died (93; IQR 43–170) (p=0.036 by Wilcoxon test).

Figure 2

Free hemoglobin at 12 hours post-operatively are shown as a boxplot (median and 10^th, 25^th, 75^th, 90^th percentiles) for patients who developed infections (n=27) versus patients who remained uninfected during their hospital stay (n=131). Two patients had no sample or died prior to 12 hours. The graphs on the right are histograms displaying the relative number of infected patients in that range of free hemoglobin levels. The median free hemoglobin in patients without infection (30; interquartile range 24–52) was lower than in those who became infected (39; IQR 33–88) (p=0.0046 by Wilcoxon test).

Figure 3

Free hemoglobin at 12 hours post-operatively are shown for patients who developed thrombosis (n=15) versus patients who had no thrombosis during their hospital stay (n=143) as a boxplot (median and 10^th, 25^th, 75^th, 90^th percentiles). Two patients had no sample or died prior to 12 hours. The graphs on the right are histograms displaying the relative number of patients with thrombosis at the specified range of free hemoglobin. The median free hemoglobin in patients without thrombosis (31; interquartile range 24–50) was less than in those who developed a thrombosis. (60; IQR 29–125) (p=0.057 by Wilcoxon test).

Figure 4

Hours of intubation are shown as a boxplot (median and 10^th, 25^th, 75^th, 90^th percentiles) in patients with low pre-operative haptoglobins (n=67) versus patients with adult normal levels (60–200 mg/dl) (n=91). Two patients were missing data on pre-operative haptoglobin. The graphs on the right are histograms displaying the relative number of patients in each range of haptoglobin levels. The median hours of intubation were 19 (interquartile range 7 to 72) in patients with adult normal levels vs. 48 (IQR 18 to 144) hours in patients with low levels (p=0.0047 by Wilcoxon test).

Figure 5

Hours of inotropes are shown as a boxplot (median and 10^th, 25^th, 75^th, 90^th percentiles) in patients with low pre-operative haptoglobins (n=67) versus patients with adult normal levels (60–200 mg/dl) (n=91). Two patients were missing data on pre-operative haptoglobin. The graphs on the right are histograms displaying the relative number of patients in each range of haptoglobin levels. The median hours of inotropes were 48 (interquartile range 7 to 96) in patients with adult normal levels vs. 72 (IQR 24 to 168) hours in patients with low levels (p=0.01 by Wilcoxon test).