Although PCT as a biomarker for sepsis has been described in previous studies. current knowledge of PCT levels to predict blood culture results is scarce, especially in children (12-13). A 2015 meta-analysis showed that PCT is highly accurate in differentiating bacterial and viral meningitis in children with 96% sensitivity (12). In our study included 311 consecutive patients, over a 7-year study period, PCT was effective to predict blood-culture results in cases that underwent a sepsis episode, particularly in immunocompromised children.
According to our data, mean PCT levels in cases with GNB infections were significantly higher than those with GPC and Candida spp infections. In our study, we found a 75% sensitivity and 53% specificity using PCT as a predictor of GNB infection in children. Previous studies in adults showed similar results (14-16). The median PCT levels in the adult population with GNB infection has been reported in 26.1 ng/ml, 25.1 ng/ml, and 7.47 ng/ml, comparable to our results with an 18.2 mcg/Lvalue (18-20). A previous report in the adult population of Shuhua et al (20), regarding sensitivity and specificity, with an optimal cut-off value of 3.11 ng/mL, led to 63.9% and 93.3% values, respectively. Yan et al (1) and Watanabe et al (17), described a similar sensitivity from GNB infection with 72.5% and 74.5% results, respectively. A subgroup analysis of our study demonstrates87.5% sensitivity using PCT for predicting GNB infection in immunocompromised children (AUC 0.906, 95% CI, 0.748 – 1), this finding has not been reported in children.
The role of PCT as a predictor of GPC in blood cultures, mainly in infections caused by Staphylococci was evaluated by Shomali et al, reporting higher mean PCT levels in infections by S. aureus compared to coagulase-negative Staphylococci (0.85 mcg/L versus 0.26 mcg/L, respectively) (21). In our study when we compare the PCT levels between Staphylococcus aureus and coagulase-negative Staphylococci, we found higher PCT levels in bloodstream infections by coagulase-negative Staphylococci (17.3 vs 0.8 ng/mL), a different result than Shomali et al study. This difference may be attributed to a higher isolation rate of coagulase-negative staphylococci in our hospital.
Studies that analyze PCT levels as a biomarker for invasive fungal infection by Candida spp are scarce and show conflicting data (22-25). According to median PCT levels in Candida spp infection, a median of 0.6 ng/ml, 0.5 ng/ml, 1 ng/ml and 0.5 ng/ml were reported by Shuhua et al (20), Miglietta et al (26), Oussalah et al (27) and Leli et al (29), respectively. In another work by Thomas-Rüddel et al (18) a higher median of 4.7 ng/ml was described. Consistent with previous studies in the adult population, we report lower PCT levels in fungal infections compared to bacterial events. In a previous study by Cartegiani et al (25), using PCT levels to predict a Candida spp bloodstream infection, 86.8% sensitivity was described. These results were similar to our work, with a 75% sensitivity. Although the identification of Candida species was not performed in our study, previous authors have not found any difference regarding PCT levels in infections by different Candida species.
Current knowledge of PCT role in immunocompromised children is scarce (30-34). Previous reports in children with cancer showed that PCT is an effective biomarker of sepsis during a fever and neutropenia episode; however, none of them evaluate the role of PCT to predict blood culture results (30,34). We report an 87.5% sensitivity of PCT for predicting blood culture-proven GNB infection, making PCT a useful resource in clinical practice. PCT levels were also increased in different types of immunosuppression. According to mean PCT levels, we found a statistically significant difference between immunocompromised (26.68 mcg/L) and immunocompetent (8.78 mcg/L) children with sepsis (p <0.05). A different result than Al-Nawas B in the adult population.
Our study has several limitations. A prospective design would aid in having better control of the variables and include a larger sample, to avoid heterogeneity of the cases. Likewise, PCT measurements were not serial, which would have allowed us to analyze PCT behavior concerning variables such as time, isolated microorganism, treatment, and outcome.