Platelet transfusions in a murine model of neonatal polymicrobial sepsis: Divergent effects on inflammation and mortality

Abstract

Background: Platelet transfusions (PTxs) are often given to septic preterm neonates at high platelet count thresholds in an attempt to reduce bleeding risk. However, the largest randomized controlled trial (RCT) of neonatal transfusion thresholds found higher mortality and/or major bleeding in infants transfused at higher thresholds. Using a murine model, we investigated the effects of adult PTx on neonatal sepsis-induced mortality, systemic inflammation, and platelet consumption.

Study design and methods: Polymicrobial sepsis was induced via intraperitoneal injection of cecal slurry preparations (CS1, 2, 3) into P10 pups. Two hours after infection, pups were transfused with washed adult Green Flourescent Protein (GFP+) platelets or control. Weights, platelet counts, and GFP% were measured before 4 and 24 h post-infection. At 24 h, blood was collected for quantification of plasma cytokines.

Results: The CS batches varied in 24 h mortality (11%, 73%, and 30% in CS1, 2, and 3, respectively), due to differences in bacterial composition. PTx had differential effects on sepsis-induced mortality and systemic inflammatory cytokines, increasing both in mice infected with CS1 (low mortality) and decreasing both in mice infected with CS2 and 3. In a mathematical model of platelet kinetics, the consumption of transfused adult platelets was higher than that of endogenous neonatal platelets, regardless of CS batch.

Discussion: Our findings support the hypothesis that transfused adult platelets are consumed faster than endogenous neonatal platelets in sepsis and demonstrate that PTx can enhance or attenuate neonatal inflammation and mortality in a model of murine polymicrobial sepsis, depending on the composition of the inoculum and/or the severity of sepsis.

Keywords: Cecal slurry; mathematical modeling; neonatal platelet transfusion; neonatal sepsis; neonatal thrombocytopenia; platelet consumption.

Figure 1.. Heatmap showing the relative abundance of specific bacterial species in CS2 and CS3 (high mortality) compared to CS1 (low mortality).

This analysis showed an up-regulation of pathogenic bacteria such as Legionella spp and Sutterella spp, and a downregulation of beneficial enteric bacteria like Bifidobacterium choerinum, in CS2 and 3 compared to CS1.

Figure 2.. Plasma cytokine concentrations in non-transfused mice infected with CS, 24 hours after infection.

Shown are the cytokines that were statistically different between CS1 (n=8) and CS2/3 (n=5).

Figure 3.. Response of 31 cytokines to platelet transfusion (PLT) in P10 mice after infection with different CS batches, compared to infusion of tyrodes buffer (TY, controls).

For nearly all cytokines, platelets increased plasma concentrations (PLT:TY ratio >1) after infection with CS1 (●, n=16) but reduced the concentrations (ratio < 1) after infection with CS2–3 (×, n=13). The mean difference for overall cytokine change between CS1 and CS2–3 was highly significant (bottom panel), even though the difference was non-significant for most individual cytokines, with the exception of LIX (CXCL5) (p-values at left). Ratio error limits of ±1 SE are shown when the batch difference was significant or near-significant (top cytokines), otherwise omitted for clarity.

Figure 4.. Comparison of predicted and observed total platelet counts after infection.

Simulated platelet counts (white bars) were in good agreement with observed values (gray bars) for (A) low mortality (n = 7) and (B) high mortality (n = 4) groups at 24 hours before infection and 4 hours and 24 hours after infection. Bars of observed total platelet count represent the mean ± 95% confidence intervals (CI). All model-predicted values were within the 95% CI of the observed values, indicating good agreement.

Figure 5.. Comparison of predicted and observed total platelet counts after infection and platelet transfusion.

Simulated endogenous and transfused platelets were in good agreement with observed platelets for (A) low mortality (n = 8) and (B) high mortality (n = 7) groups at 24 hours before infection and 4 and 24 hours after infection. Endo refers to endogenous platelets denoted by bars with pattern, and Trans refers to transfused platelets denoted by bars without pattern. Simulated results (Sim) are indicated with white color, and observed platelet counts (Obs) are denoted by gray color. Bars of observed total platelet count represent the mean ± 95% confidence intervals (CI). All model predicted values were within the 95% CI of the observed values. Also consistent with the model, the decrease in platelet counts between 4 and 24 hours was significantly more pronounced for the transfused compared to the endogenous platelets in both the low-mortality (p=0.005) and the high-mortality (p=0.0004) sepsis models.