There is an urgent need for better diagnostic tests for early detection of necrotizing enterocolitis (NEC). While IFABP and calprotectin are promising biomarkers, understanding the normal variations across gestational ages and during the early post-natal period is paramount to accurately interpreting results in a clinical setting. In a study comparing IFABP levels among premature sheep, differences were seen between extremely and moderately premature sheep, with the lowest levels in the most premature sheep. Similar findings occur in other intestinal biomarkers. For example, calprotectin, a fecal marker of inflammation in the intestine, is routinely used clinically in older children and adults, but has had mixed results in the neonatal population. Very low calprotectin levels are seen in extremely premature infants, but the levels progressively increase and remain higher than normal childhood levels until after 3 months of age, even without any obvious gastrointestinal pathology.[6, 14, 15] This range of levels is in part explained by the complex interactions between with evolving intestinal microbiota and the development of immune tolerance and/or inflammation during the first few months of life.
In our study, most infants had minimally detectable IFABP levels during the first week of life, regardless of gestational age. Over 75% of subjects had an IFABP level <1 ng/ml, with 50% having a level <0.03 ng/ml. There was a statistical difference between IFABP levels across the gestational ages, but this is unlikely to be clinically significant as the median values were <1 ng/ml for each gestational age group. By comparison, most proposed cut-offs for NEC have ranged from 2 to 9 ng/ml for Bell Stage 2 and 3 NEC, with even higher levels (>19 ng/ml) reported in one study of surgical infants with a high mortality.[10, 11, 13, 16] In our cohort, there were two outliers with very high levels who did not have identifiable gastrointestinal disease.
Among the cohort of premature infants, the population most at risk of developing NEC, there were 76 repeat measures after the first week of life. Of the 76 measures, 31 (41%) were >2.5 ng/ml and 4 (5%) were >9 ng/ml, suggesting there would be a high rate of false-positives for the lower cut-off values. Only some of the variability in IFABP levels was explained by day of life. There was not an obvious association with demographic or birth factors, such as gender, race, birth weight, Apgar score, or chorioamnionitis. We have limited feeding data (only 50% of longitudinal subjects), with the vast majority being fed breast milk. It is possible that feeding type, and therefore intestinal microbiome, impacts the IFABP level. We also were reliant on culture results to exclude sepsis, so it is possible that culture-negative infection could have affected the results.
In contrast, calprotectin levels were more variable during the first week of life, partially explained by gestational age. The proposed cut-off level for NEC is 3,000 ng/ml. Twenty-nine subjects (26%) would have had false-positive results. Calprotectin is associated with sepsis. While all of our subjects had negative cultures, it is possible that other etiologies of infection were contributing to the higher levels seen during the first week. After the first week of life, only 5 (7%) of the levels were above the cut-off, suggesting a lower risk for false-positive results.
There were limitations to the study, including the relatively small sample size; however, subjects were from two centers and represent the spectrum of gestational ages seen in the NICU. Furthermore, our longitudinal subjects represent the population most likely to develop NEC. Our sub-analyses were limited by missing data, particularly type of feeding. Future studies will be needed to more formally evaluate feeding type and potential associations with the microbiome.
In summary, gestational age and day of life may influence biomarker levels in premature infants. IFABP was relatively low during the first week of life across all gestational ages. However, among premature infants, a substantial proportion had detectable levels with repeat measures as day of life increased, which may ultimately affect the sensitivity and specificity of IFABP for detecting NEC. Calprotectin levels were quite variable during the first week of life, which was associated with gestational age. However, levels quickly fell during subsequent measures in the preterm infants. Understanding the timing of testing in relation to post-menstrual age will be critical in interpreting results for clinical management.