Platelet Transfusions Can Increase or Decrease the Inflammatory Response and Mortality in a Murine Model of Neonatal Polymicrobial Sepsis

Thrombocytopenia affects 18-35% of all neonates in the Neonatal Intensive Care Unit and ~70% of those born extremely prematurely, with sepsis being a frequent cause. Platelet transfusions (PTx) are frequently given to septic preterm neonates at higher platelet count (PC) thresholds than those used in adults, in an attempt to reduce their bleeding risk. However, in the largest randomized trial of neonatal PTx thresholds, infants transfused at a higher PC threshold had a significantly higher mortality and/or major bleeding compared to infants transfused at a lower threshold. We hypothesized that the deleterious effects of PTx would be related to a potential "developmental mismatch" resulting from adult platelets being transfused into a neonate. Among other developmental differences, adult platelets (human and murine) exhibit significantly higher surface P-selectin expression following activation than neonatal platelets. P-selectin is essential for the interaction of platelets with immune cells. Thus, we hypothesized that adult platelets transfused into septic neonates would be consumed faster than endogenous neonatal platelets (due to higher potential for immune interaction), and would increase inflammation and mortality. To test these hypotheses, we used a published murine model of neonatal sepsis, consisting of injecting cecal slurry (CS) into C57BL/6 pups. CS batches were prepared by isolating the cecal content of adult C57BL/6 mice, which was weighted, aliquoted and frozen until use. Three different CS batches were prepared and injected IP into post-natal day 10 pups at a dose of 1.1 (CS1) or 1.0 mg/g (CS2 and 3). Two hours after infection, pups were transfused with washed platelets from adult GFP mice (5x10 ⁷ platelets/g) or Tyrode's buffer (control). Weights, PCs and GFP platelet % were measured before, 4h and 24h post-infection. Blood was collected via terminal bleed at 24h, and plasma separated for quantification of 31 cytokines by multiplex. Despite identical preparation, CS batches varied greatly in their 24h mortality (11% vs 73% vs. 30% for CS1, 2 and 3, respectively). Moreover, PTx had different effects on the mortality of pups infected with different CS batches, increasing the 24h mortality of pups infected with CS1 (30% in transfused vs 11% in non-transfused, RR 2.70, 95% CI 1.02-7.15) but decreasing the mortality of pups infected with CS2 (46% vs. 73%) or CS3 (9% vs. 30%), with a combined RR of 0.52; 95% CI 0.30-0.91. Bacterial counts differed between CS batches, but did not correlate with mortality. Comparison of the microbiome composition using deep sequencing revealed an increased presence of pathogenic bacterial species (Legionella, Sutterella, and Helicobacter species) in CS2 and 3 compared to CS1, and a relative abundance of beneficial bacterial (Actinobacteria and Proteobacteria) in CS1. Different CS batches also elicited different cytokine responses, with significant differences noted in G-CSF, IL-1α, IL-1β, IL-3, IL-7, IL-12p70, and IL-15 levels (p<0.05). For all of these cytokines, except G-CSF, levels were lower in mice infected with CS1 compared to CS2 or 3. Next, we investigated the effects of PTx on the plasma cytokine profile of mice infected with CS1 or CS2/3 (combined), compared to their infected, non-transfused littermates. For nearly all cytokines, PTx increased the response after infection with CS1, but decreased it after infection with CS2/3, with a significant difference in mean global cytokine effect (p<0.0001). For individual cytokines, however, these differences only reached statistical significance for LIX (CXCL5, p=0.04) and approached significance for IL15 and IL17 (p=0.06). Finally, we developed a mathematical model to compare the consumption of endogenous neonatal platelets (GFP-) to that of transfused adult platelets (GFP+) in pups infected with CS1 vs. CS2. In both, the calculated percent consumption was higher for adult platelets than for neonatal platelets (54.8% vs. 32.6% for CS1 and 56.5% vs. 40.4% for CS2). In conclusion, our findings support the hypothesis that adult transfused platelets are consumed faster than endogenous platelets in early neonatal sepsis, and demonstrate that platelet transfusions can either enhance or attenuate the neonatal inflammatory response and the mortality in a model of murine polymicrobial sepsis, depending on the bacterial composition of the inoculum and/or the severity of the sepsis.