Abstract
Objective
Feeding intolerance (FI) is a common presentation of necrotizing enterocolitis (NEC) and sepsis. NEC and sepsis are associated with hematological changes, but these changes alone are not reliable biomarkers for early diagnosis. This study examined whether the combination of hematological indices and FI can be used as an early diagnostic tool for NEC or sepsis.
Study Design
This retrospective cohort study included infants born at <1,500 g or <30 weeks who had symptoms of FI. The exclusion criteria were congenital or chromosomal disorders, thrombocytopenia or platelet transfusion before the onset of FI, and history of bowel resection. We compared the hematological indices from infants with pathologic FI (due to NEC or sepsis) to infants with benign FI.
Results
During the study period, 211 infants developed FI; 185 met the inclusion criteria. Infants with pathologic FI (n = 90, 37 cases with NEC and 53 with sepsis) had lower birth gestational age and weight compared with 95 infants with benign FI (n = 95). Pathologic FI was associated with lower platelet count (median 152 × 103/μL vs. 285 × 103/μL, p < 0.001) and higher immature-to-total neutrophil (I/T) ratio (median 0.23 vs. 0.04, p < 0.001) at the onset of FI. Pathologic FI was also associated with a decrease in baseline platelets compared with an increase in benign FI. For diagnosis of pathologic FI, a decrease ≥10% in platelets from baseline had a sensitivity and specificity of 0.64 and 0.73, respectively, I/T ratio ≥0.1 had a sensitivity and specificity of 0.71 and 0.78, respectively, and the combination of both parameters had a sensitivity and specificity of 0.50 and 0.97, respectively.
Conclusion
FI caused by NEC or sepsis was associated with a decrease in platelets from baseline, and a lower platelet level and higher I/T ratio at the onset of FI. These findings can help clinicians in the management of preterm infants with FI.
Keywords: prematurity, necrotizing enterocolitis, sepsis, feeding intoleranc, platelets, neutrophils
Feeding intolerance (FI) is common in very low birth weight (VLBW) infants and the underlying causes range from benign to pathologic. It usually refers to a combination of symptoms such as abdominal distension, vomiting, increased prefeeding gastric residuals, and less frequently, bloody stools.1 Benign causes of FI include slow gut motility, exposure to maternal magnesium treatment, immature gut development, and intolerance to cow-milk-based products.1–4 Pathologic causes of FI include necrotizing enterocolitis (NEC) or sepsis, and are often a result of systemic and bowel wall inflammation and edema secondary to bacterial presence in the blood or in the intestinal wall.
NEC is a devastating inflammatory disease that affects 5 to 10% of VLBW infants, with an associated mortality of up to 50% and significant short- and long-term morbidities.5 Sepsis affects ~25% of VLBW infants, with mortality of up to 50% and significant morbidity.6 Currently, the diagnosis of NEC is based on the presence of characteristic radiological signs, such as pneumatosis intestinalis, portal venous gas, and free peritoneal air.7,8 These signs often present late in the course of NEC.8,9 Studies have attempted to identify biomarkers for early diagnosis of NEC and sepsis, but none has been shown to be reliable. Previous literature shows that platelet count and immature-to-total neutrophil (I/T) ratio are prognostic biomarkers for NEC, with lower platelet count and higher I/T ratio being associated with increased severity, mortality, and need for surgery.10–13 Multiple hematological abnormalities have also been associated with sepsis, including neutropenia or neutrophilia, increased I/T ratio, and thrombocytopenia.14–16 However, when used alone, these hematological changes are not reliable in the early diagnosis of NEC or sepsis.6,17,18
The use of hematological indices in combination with FI in preterm infants has not been explored. Thus, we sought to evaluate the value of hematological indices as an early indicator of NEC and sepsis in preterm infants who present with only signs of FI. We hypothesize that hematological values obtained from complete blood counts (CBCs) and differentials at the onset of FI enhance the early identification of NEC or sepsis.
Materials and Methods
Patient Population and Data Collection
A retrospective cohort study was performed on infants with a birth weight (BW) <1,500 g or birth gestational age (GA) <30 weeks who were admitted to the level 3 neonatal intensive care unit at Tampa General Hospital between October 2011 and December 2017. The study was approved by the University of South Florida Institutional Review Board. The inclusion criteria were the development of one or more signs of FI (increased abdominal girth from baseline, vomiting, prefeeding residuals >30% of feeding volume, or bloody stool) that resulted in evaluation with an abdominal X-ray and a CBC. We excluded infants with congenital anomalies or chromosomal disorders, infants who had persistent thrombocytopenia before the onset of FI, infants who received platelet transfusion within 5 days before the onset of FI, infants with incomplete CBC data, and infants who had a history of bowel resection. We divided these infants into two groups: “pathologic FI,” which includes infants with FI due to either NEC (stage 2a or greater by modified Bell’s criteria) or blood culture-positive sepsis and “benign FI,” which included infants who developed FI from other causes that was not due to NEC or sepsis and did not result in prolonged (>48 hours) administration of antibiotics or cessation of enteral feeding.
The data were collected from the electronic medical records of the infants and from the Vermont Oxford Network database. A REDCap database was used for data collection and storage. We collected the following demographic and clinical information: BW, birth GA, gender, ethnicity, race, maternal age, prenatal care, intrauterine growth restriction (IUGR) or small for gestational age status, maternal comorbidities (chorioamnionitis and hypertension), antenatal steroids, multiple gestation, mode of delivery, 5-minute Apgar score, discharge day of life (DOL), discharge weight, discharge disposition, grade 3 or 4 intraventricular hemorrhage (IVH), periventricular leukomalacia (PVL), bronchopulmonary dysplasia (BPD) (defined as the need for oxygen or respiratory support at 36 weeks postmenstrual age [PMA]), and any stage of retinopathy of prematurity (ROP). We also collected the following information at the onset of FI: DOL, PMA, type of feeding, the presence of central line, the presence of a patent ductus arteriosus (PDA), PDA therapy (indomethacin or ibuprofen) prior to the FI diagnosis, and DOL of the most recent platelet transfusion prior to the onset FI. To compare the different values of the CBC, we obtained the absolute neutrophil count (ANC), absolute monocyte count (AMC), the platelet count, and the I/T ratio from the CBC at the time of diagnosis and the most recent CBC before the diagnosis (baseline CBC). All the CBCs were performed at the Tampa General Hospital laboratory.
Statistical Analysis
Mann–Whitney’s U test was used to compare nonparametric continuous variables and the Fischer’s exact test was used to compare categorical variables. All statistical tests were two sided and considered significant at p-value < 0.05. To evaluate the diagnostic power and identify a cutoff value of the percentage of change in the platelet count from baseline and of the I/T ratio at the onset of FI with the best diagnostic properties, we computed receiver-operating characteristics (ROC) curves of each parameter and of both parameters combined. The diagnostic power of the cutoff values was determined by computing the sensitivity, specificity, positive and negative predictive values (PPV and NPV), and positive and negative likelihood ratios from 2 × 2 tables. Statistical analysis was performed using IBM SPSS Statistics version 25.
Results
Demographic and Clinical Characteristics
We identified 211 infants who developed FI during the study period. Twenty-six infants were excluded by the exclusion criteria. Of the remaining 185 infants, 90 were diagnosed with pathologic FI (37 with NEC and 53 with sepsis) and 95 were diagnosed with benign FI (▶Fig. 1).
Fig. 1.
Flow diagram summarizing study groups and exclusions. CBC, complete blood count; FI, feeding intolerance; NEC, necrotizing enterocolitis.
The median birth GA and BW for infants with pathologic FI were 25 weeks (interquartile range [IQR] = 24–27 weeks) and 738 g (IQR = 609–928 g), and were both significantly lower from the median birth GA of 27 weeks (IQR = 26–29 weeks) and median BW of 930 g (IQR = 780–1,150 g) in infants with benign FI (▶Table 1). Other demographic and clinical characteristics at birth were similar in both groups, except for a higher incidence of chorioamnionitis and multiple gestation in infants with pathologic FI, a higher incidence of maternal hypertension in infants with benign FI, and a lower 5-minute Apgar score in infants with pathologic FI. Infants with pathologic FI had a lower PMA at the onset of FI compared with infants with benign FI. Other clinical characteristics at the onset of FI were similar between the two groups, except for the presence of a PDA. The length of stay and discharge weight were similar between the two groups, but the infants in the pathologic FI group had higher morbidity (IVH grade 3 or 4, PVL, ROP, and BPD) and higher mortality compared with the infants in the benign FI group (▶Table 1).
Table 1.
Demographic and clinical characteristics
| Pathologic FI N = 90 | Benign FI N = 95 | p-Value | ||
|---|---|---|---|---|
| Demographic and clinical characteristics at birth | ||||
| Birth weight (g); median (IQR) | 738 (609–928) | 930 (780–1,150) | <0.001 | |
| Birth GA (wk); median (IQR) | 25 (24–27) | 27 (26–29) | <0.001 | |
| Female gender; n (%) | 49 (54.4) | 43 (45.3) | 0.212 | |
| Hispanic; n (%) | 14 (15.6) | 20 (21.1) | 0.335 | |
| Race; n (%) | ||||
| White | 44 (48.9) | 58 (61) | 0.169 | |
| Black or African American | 42 (46.7) | 32 (33.7) | ||
| Asian | 2 (2.2) | 5 (5.3) | ||
| Other | 2 (2.2) | 0 | ||
| Maternal age (y); median (IQR) | 27 (23–32) | 28 (24–33) | 0.201 | |
| Prenatal care; n (%) | 87 (96.7) | 91 (95.8) | 0.755 | |
| Maternal chorioamnionitis; n (%) | 52 (57.8) | 31 (32.6) | 0.003 | |
| Maternal hypertension; n (%) | 20 (22.2) | 42 (44.2) | 0.002 | |
| Antenatal steroids; n (%) | 83 (92.2) | 88 (92.6) | 0.580 | |
| Multiple gestation; n (%) | 33 (36.7) | 21 (22.1) | 0.029 | |
| Cesarean section; n (%) | 66 (73.3) | 67 (70.5) | 0.671 | |
| 5-min Apgar <6; n (%) | 25 (27.8) | 14 (14.7) | 0.030 | |
| IUGR or SGA; n (%) | 16 (17.8) | 21 (22.1) | 0.462 | |
| Clinical characteristics at the onset of FI | ||||
| DOL; median (IQR) | 19 (10–35) | 17 (9–33) | 0.589 | |
| PMA (wk); median (IQR) | 29 (26–32) | 30 (28–33) | 0.005 | |
| Central line; n (%) | 50 (55.6) | 44 (46.3) | 0.209 | |
| PDA; n (%) | 52 (57.8) | 30 (31.6) | <0.001 | |
| PDA therapy; n (%) | 25 (27.8) | 20 (21.1) | 0.287 | |
| Any breast milk; n (%) | 64 (71.1) | 74 (77.9) | 0.289 | |
| Discharge outcomes | ||||
| Discharge DOL; median (IQR) | 100 (43–134) | 74 (64–103) | 0.134 | |
| Discharge weight (g); median (IQR) | 2,700 (1,511–3,799) | 2,800 (2,220–3,210) | 0.527 | |
| IVH grade 3 or 4; n (%) | 11 (12.2) | 4 (4.2) | 0.046 | |
| PVL; n (%) | 6 (6.7) | 0 | 0.038 | |
| BPD; n (%) | 46 (51.1) | 27 (28.4) | 0.002 | |
| Any ROP; n (%) | 49 (54.4) | 33 (34.7) | <0.001 | |
| Death; n (%) | 27 (30) | 1 (1.1) | <0.001 | |
Abbreviations: BPD, bronchopulmonary dysplasia; DOL, day of life; FI, feeding intolerance; GA, gestational age; IQR, interquartile range; IUGR, intrauterine growth restrictions; IVH, intraventricular hemorrhage; PDA, patent ductus arteriosus; PMA, postmenstrual age; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity; SGA, small for gestational age.
CBC Analysis
The median platelet count at the onset of pathologic FI was significantly lower than the median platelet count at the onset of benign FI (152 × 103/μL [IQR = 98–248 × 103/μL] and 285 × 103/μL [IQR = 226–388 × 103/μL], respectively, p < 0.001). The median I/T ratio at the onset of pathologic FI was significantly higher than the median I/T ratio at the onset of benign FI (0.23 [IQR = 0.08–0.39] and 0.04 [IQR = 0–0.1], respectively, p < 0.001).
Infants with pathologic FI had a significant decrease in platelet count at the onset of FI from baseline (median platelet count of 152 × 103/μL [IQR = 98–248 × 103/μL] at the onset of FI vs. 218 × 103/μL [IQR = 149–271 × 103/μL] at baseline, p < 0.001). On the contrary, infants with benign FI had a higher platelet count at the onset of FI compared with baseline (median platelet count of 285 × 103/μL [IQR = 226–388 × 103/μL] at the onset of FI vs. 266 × 103/μL [IQR = 178–338 × 103/μL] at baseline, p = 0.029). When we compared the percentage change in the platelet count at the onset of FI from baseline, infants with pathologic FI had a median decrease of 23%, compared with a median increase of 7% seen in infants with benign FI (p < 0.001). Notably, the two groups had similar ANC and AMC at the onset of FI, and there was no significant change in either one at the onset of FI compared with baseline in the two groups.
The ROC curves for the percentage change in the platelet count at the onset of FI from baseline and for the I/T ratio at the onset of FI are illustrated in ▶Fig. 2A, B. For the percentage change in platelet count, the area under the curve was 0.74 (95% confidence interval [CI] = 0.67–0.81; p < 0.001), indicating fair-to-good diagnostic accuracy. Using this curve, we determined that a decrease of ≥10% in the platelet count from baseline has a sensitivity of 0.64 and specificity of 0.73 for pathologic FI (▶Table 2). For the I/T ratio at the onset of FI, the area under the curve was 0.80 (95% CI = 0.73–0.86; p < 0.001), indicating a relatively good diagnostic accuracy. Using this curve, we determined that an I/T ratio of ≥0.1 has a sensitivity of 0.71 and a specificity of 0.78, and an I/T ratio of ≥0.2 has a sensitivity of only 0.53 but a high specificity of 0.93 for diagnosis of pathologic FI (▶Table 2).
Fig. 2.
ROC curves: (A) ROC curve for the percentage change in the platelet count from baseline. Area under the curve is 0.74, indicating “fair-to-good” diagnostic accuracy; (B) ROC curve for the I/T ratio at the onset of FI. Area under the curve is 0.8, indicating good diagnostic accuracy; (C) ROC curve of the predicted probability of the combined change in platelet count and I/T ratio. Area under the curve is 0.81, indicating good diagnostic accuracy. I/T, immature-to-total neutrophil; ROC, receiver-operating characteristics.
Table 2.
Diagnostic properties of different hematological cutoff values for diagnosis of pathologic FI
| Sensitivity | Specificity | PPV | NPV | LR+ | LR− | |
|---|---|---|---|---|---|---|
| Platelet decrease ≥10% | 0.64 | 0.73 | 0.69 | 0.68 | 2.37 | 0.49 |
| I/T ratio ≥0.1 | 0.71 | 0.78 | 0.75 | 0.74 | 3.23 | 0.37 |
| I/T ratio ≥0.2 | 0.53 | 0.93 | 0.68 | 0.87 | 1.98 | 0.14 |
Abbreviations: FI, feeding intolerance; I/T, immature-to-total neutrophil; LT+, positive likelihood ratio; LR−, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value.
We also performed a binary logistic regression analysis of the combination of the presence of a change in platelet count of ≥10% and an I/T ratio of ≥0.1 at the onset of FI, and we computed the ROC curve for the predicted probability that resulted from the analysis (▶Fig. 2C). The ROC curve showed that a predicted probability of 0.74, which corresponds to an I/T ratio of ≥0.1 and a platelet decrease ≥10%, has a sensitivity of only 0.50 but a specificity of 0.97.
Group Stratification by Birth Weight
We further stratified infants by BW into two subgroups, infants with BW ≤750 g and infants with BW >750 g. Analysis of the demographic and CBC data in both subgroups showed no differences in the BW, birth GA, DOL at diagnosis, or PMA at diagnosis between infants with benign FI and infants with pathologic FI. The subgroups had similar differences between platelet count at the onset of FI, change in the platelet count prior to the onset of FI from baseline, and I/T ratio at the onset of FI (▶Table 3). ROC curves computed for the two subgroups were similar to those for the main groups. In infants >750 g, the ROC curve for the I/T ratio showed improved diagnostic power with the area under the curve of 0.84 (95% CI = 0.76–-0.92; p < 0.001). This curve showed that an I/T ratio of ≥0.1 has a sensitivity of 0.76 and a specificity of 0.78.
Table 3.
Clinical characteristics and CBC parameters in infants with pathologic versus benign FI, stratified by BW
| Clinical characteristics | ≤750 g | >750 g | ||||
|---|---|---|---|---|---|---|
| Pathologic FI N = 48 | Benign FI N = 19 | p-Value | Pathologic FI N = 42 | Benign FI N = 76 | p-Value | |
| BW (g); median (IQR) | 617 (552–696) | 630 (595–650) | 0.651 | 943 (808–1,242) | 1,000 (850–1,270) | 0.066 |
|
| ||||||
| Birth GA (wk); median (IQR) | 24 (23–25) | 25 (23–27) | 0.282 | 27 (26–29) | 27.5 (26–29) | 0.132 |
|
| ||||||
| DOL at the onset of FI; median (IQR) | 20 (9–35) | 21 (12–33) | 0.928 | 17 (10–38) | 16 (8–33) | 0.645 |
|
| ||||||
| PMA at the onset of FI (wk); median (IQR) | 27 (25–30) | 28 (27–30) | 0.143 | 30 (28–33) | 30 (28–34) | 0.553 |
|
| ||||||
| Discharge DOL; median (IQR) | 104 (44–155) | 116 (101–130) | 0.366 | 40 (23–77) | 54 (39–71) | 0.285 |
|
| ||||||
| Discharge weight (g); median (IQR) | 2,700 (1,207–4,033) | 3,125 (2,850–3,580) | 0.098 | 1,838 (1,140–2,730) | 2,193 (2,006–2,645) | 0.844 |
|
| ||||||
| Maternal chorioamnionitis; n (%) | 32 (66.7) | 6 (31.6) | 0.020 | 20 (47.6) | 25 (32.9) | 0.284 |
|
| ||||||
| IUGR or SGA; n (%) | 12 (25) | 10 (52.6) | 0.044 | 4 (9.5) | 11 (14.5) | 0.569 |
|
| ||||||
| Central line; n (%) | 33 (68.8) | 13 (68.4) | 1.000 | 17 (40.5) | 31 (40.8) | 1.000 |
|
| ||||||
| PDA; n (%) | 31 (64.6) | 7 (36.8) | 0.056 | 21 (50) | 23 (30.3) | 0.050 |
|
| ||||||
| PDA therapy; n (%) | 13 (27.1) | 7 (36.8) | 0.555 | 12 (28.6) | 13 (17.1) | 0.163 |
|
| ||||||
| Any breast milk; n (%) | 31 (64.6) | 17 (89.5) | 0.069 | 33 (78.6) | 57 (75) | 0.822 |
|
| ||||||
| IVH grade 3 or 4; n (%) | 7 (14.6) | 1 (5.3) | 0.424 | 4 (9.5) | 3 (3.9) | 0.245 |
|
| ||||||
| PVL; n (%) | 2 (4.2) | 0 | 0.537 | 4 (9.5) | 0 | 0.018 |
|
| ||||||
| Any ROP; n (%) | 32 (66.7) | 14 (73.7) | 0.280 | 17 (40.5) | 19 (25) | <0.001 |
|
| ||||||
| BPD; n (%) | 31 (64.6) | 10 (52.6) | 0.413 | 15 (35.7) | 17 (22.4) | 0.134 |
|
| ||||||
| Death; n (%) | 17 (35.4) | 0 | 0.002 | 10 (23.8) | 1 (1.3) | <0.001 |
|
| ||||||
| CBC parameters | ||||||
|
| ||||||
| Platelet count at the onset of FI × 103/μL; median (IQR) | 149 (93–230) | 237 (162–343) | 0.001 | 182 (104–258) | 299 (232–400) | <0.001 |
|
| ||||||
| Percentage change in platelet count from baseline; median (IQR) | −20% (−48 to +14%) | +12% (−4 to +44%) | 0.001 | −25% (−47 to 0%) | +3% (−16 to +35%) | <0.001 |
|
| ||||||
| I/T ratio at the onset of FI; median (IQR) | 0.19 (0.07–0.34) | 0.05 (0.02–0.10) | 0.004 | 0.28 (0.10–0.44) | 0.04 (0–0.91) | <0.001 |
Abbreviations: BPD, bronchopulmonary dysplasia; BW, birth weight; CBC, complete blood count; DOL, day of life; FI, feeding intolerance; GA, gestational age; IQR, interquartile range; IUGR, intrauterine growth restrictions; IVH, intraventricular hemorrhage; PDA, patent ductus arteriosus; PMA, postmenstrual age; PVL, periventricular leukomalacia; ROP, retinopathy of prematurity; SGA, small for gestational age.
NEC versus Sepsis
We then compared the platelet count, I/T ratio, and other CBC parameters (ANC or AMC) between infants with NEC and sepsis. Demographic and clinical characteristics of both groups were similar except for lower postnatal age at diagnosis for infants who developed sepsis, a higher incidence of IUGR in NEC, a more presence of central lines at diagnosis in sepsis, and a higher mortality in NEC (▶Table 4). Analysis of the CBC parameters showed no differences between NEC and sepsis in the platelet count at the onset of FI, the change in platelet count from baseline, or the I/T ratio at the onset of FI. There was also no difference in the ANC or the AMC at the onset of FI.
Table 4.
Clinical characteristics and CBC parameters in infants with NEC and sepsis
| NEC N = 37 | Sepsis N = 53 | p-Value | |
|---|---|---|---|
| Birth weight (g); median (IQR) | 750 (585–943) | 735 (620–918) | 0.658 |
| Birth GA (wk); median (IQR) | 26 (24–29) | 25 (24–27) | 0.284 |
| DOL at diagnosis; median (IQR) | 13 (9–23) | 7 (6–17) | 0.034 |
| PMA at diagnosis (wk); median (IQR) | 27 (25–30) | 26 (25–27) | 0.082 |
| Discharge DOL; median (IQR) | 64 (37–164) | 102 (72–133) | 0.419 |
| Discharge weight (g); median (IQR) | 2,189 (1,116–4,293) | 2,900 (2,025–3,688) | 0.329 |
| Maternal chorioamnionitis; n (%) | 23 (62.2) | 29 (54.7) | 0.522 |
| IUGR or SGA; n (%) | 12 (32.4) | 4 (7.5) | 0.004 |
| Central line; n (%) | 12 (32.4) | 38 (71.7) | <0.001 |
| PDA; n (%) | 19 (51.4) | 33 (62.3) | 0.208 |
| PDA therapy; n (%) | 6 (16.2) | 19 (35.8) | 0.056 |
| Any breast milk; n (%) | 23 (62.2) | 41 (77.4) | 0.092 |
| Death; n (%) | 20 (54.1) | 7 (13.2) | <0.001 |
| CBC parameters | |||
| Platelet count at the onset of FI × 103/μL; median (IQR) | 143 (91–267) | 153 (101–237) | 0.899 |
| Percentage change in platelet count from baseline; median (IQR) | −34% (−48 to −9%) | −12% (−46 to +13%) | 0.120 |
| I/T ratio at the onset of FI; median (IQR) | 0.26 (0.11–0.41) | 0.2 (0.06–0.58) | 0.217 |
Abbreviations: CBC, complete blood count; DOL, day of life; GA, gestational age; IQR, interquartile range; IUGR, intrauterine growth restrictions; NEC, necrotizing enterocolitis; PDA, patent ductus arteriosus; PMA, postmenstrual age; SGA, small for gestational age.
Discussion
FI is a common problem in preterm infants,19,20 and often leads to disruption of feeding plans, prolonged use of parenteral nutrition, and prolonged hospitalization.1 Diagnosing the etiology of FI can be challenging, and while it frequently results from benign etiologies, it is sometimes the first sign of a serious pathologic process, such as NEC and sepsis. Due to the high mortality and morbidity associated, and to the lack of reliable markers for both conditions,21,22 clinicians frequently stop feeds and initiate antibiotic therapy in the preterm population, which can further increase their risk for long-term complications.
The diagnosis of NEC and sepsis remains largely subjective due to the lack of reliable biomarkers with high NPV and PPV.23,24 The ideal diagnostic biomarkers need to identify these conditions early and should accurately differentiate these conditions from other diseases. Inflammatory biomarkers, such as C-reactive protein, cytokines, and procalcitonin, and hematological indices such as I/T ratio and platelets, lack specificity,21,25 which limits their application in preterm infants with multiple comorbidities. Daily surveillance with biomarkers is not cost-effective and increases blood loss from phlebotomy in small preterm infants.26,27 Polymerase chain reaction technologies produce faster results and require less blood volume, but they are not as sensitive in comparison to blood culture in the diagnosis of sepsis.28 A study by Luo et al showed that the combination of resistin-like molecule β and thrombocytopenia had a sensitivity of 83% and specificity of 93% in early diagnosis of NEC.29 A mathematical algorithm of preterm infant vital signs, including heart rate, respiratory rate, temperature, and bradycardia and desaturation events, was found to detect sepsis or NEC with 84% sensitivity and 80% specificity, and potentially identify these conditions earlier than clinical suspicion.30 Ongoing studies explore the options of rapid tests and multiomic approaches. Rapid tests to detect multiple cytokines/chemokines and acute-phase proteins have been found to improve accuracy and help in early diagnosis of sepsis and NEC.31,32 With the recent advancements in genomics technologies, a signature combination of proteins and/or inflammatory mediators using multiomic approaches, that is, metabolomics, metagenomics, and proteomics, can predict the disease onset and development.33,34 All these early findings require further validation before clinical application.
In this study, we evaluated the diagnostic value of hematological indices for early detection of NEC or sepsis in preterm infants with signs of FI. We showed that infants who developed FI due to NEC or sepsis had a significant decrease in platelet count from baseline and a higher I/T ratio at the onset of FI compared with those who had FI from benign causes. Importantly, the change in platelet count and the value of I/T ratio provided a fair-to-good diagnostic accuracy for NEC or sepsis. Specifically, our study showed that an I/T ratio of ≥0.1 at the onset of FI had a good sensitivity and specificity which makes it useful as a screening test. When an I/T ratio of ≥0.1 is in concurrence with a platelet decrease of ≥10%, the sensitivity decreased but the specificity significantly increased, which indicates that the combination of both tests can be useful as a confirmatory tool. Similarly, an I/T ratio ≥0.2 had a high specificity in our study, so it should raise the suspicion for pathologic FI. To our knowledge, this is the first study, to date, that shows the utility of the platelet count change and the I/T ratio as early screening and diagnostic markers for NEC and sepsis in infants with FI.
Our findings confirmed the previously observed associations of CBC indices with NEC and sepsis. Low platelet count is often present in NEC and can reflect poor prognosis.10,12,13,35 High I/T ratio is also associated with both sepsis and NEC.12–14 In this study, we showed that in infants who present with nonspecific symptoms of FI, a decrease in platelet count or high I/T ratio suggests a more pathologic underlying cause. This could ultimately help clinicians make decisions on the initial course of management of these infants. Our data did not specifically determine if those parameters have any prognostic value in both diseases. Since our study is retrospective from a single center and with a relatively small sample size, the results require further validations prior to clinical application.
The decrease in the platelet count from baseline and the lower platelet count at the onset of FI in our study can be explained by local consumption that results from platelet activation and aggregation in the microvasculature at the site of NEC.13 In infants with sepsis, the pathophysiology of decreased platelets is not completely understood and may include endothelial dysfunction, microangiopathic thrombosis, platelet activation and consumption, and antibody-mediated platelet destruction.36–39 The increase in the I/T ratio in both NEC and sepsis likely results from the systemic inflammatory response that typically accompanies both conditions, as systemic inflammation leads to the release of immature neutrophils from the bone marrow, resulting in increased I/T ratio. While the exact timing of these hematological changes in the pathophysiology of NEC and sepsis is not completely defined, our study showed that in infants with signs of FI, these changes can be used as early markers for diagnosis.
Our data confirmed the previously reported associations between lower GA at birth, lower BW, chorioamnionitis, and IUGR as risk factors for sepsis and NEC.13,40–44 Preterm infants have lower platelet function and less effective neutrophils,45–47 but it is unclear how GA affects platelet and immature neutrophil production at the quantitative level. Maternal chorioamnionitis and IUGR status are also known to influence many hematological indices, such as white blood cell count, neutrophil count, I/T ratio, and platelet count.48–51 These changes tend to resolve with increase in postnatal age.51–53 In an attempt to control for these confounders, we excluded infants with known thrombocytopenia prior to the onset of FI. We also used stratification by BW (as a surrogate for prematurity) and our findings persisted after stratification.
Conclusion
In conclusion, our study shows that pathologic FI caused by NEC and sepsis is associated with a decrease in the platelet count from baseline, a lower platelet count, and a higher I/T ratio at the onset of FI compared with benign FI. These findings can be beneficial to guide clinical decisions in the initiation and continuation of antibiotics and cessation of enteral feeding in preterm infants who develop FI. Prospective studies with larger sample size are needed to confirm our results and to further delineate the relationship between the different components of the CBC in NEC and sepsis.
Key Points.
FI is a common presentation of NEC and sepsis in preterm infants.
FI due to NEC or sepsis is associated with changes in platelets and I/T ratio.
These changes could be useful as early markers for diagnosis.
Footnotes
Conflict of Interest
None declared.
References
- 1.Fanaro S. Feeding intolerance in the preterm infant. Early Hum Dev 2013;89(Suppl 2):S13–S20 [DOI] [PubMed] [Google Scholar]
- 2.Manuyakorn W, Benjaponpitak S, Siripool K, et al. Cow milk protein allergy presenting as feeding intolerance and eosinophilia: case reports of three preterm neonates. Paediatr Int Child Health 2015;35(04):337–341 [DOI] [PubMed] [Google Scholar]
- 3.Belden MK, Gnadt S, Ebert A. Effects of maternal magnesium sulfate treatment on neonatal feeding tolerance. J Pediatr Pharmacol Ther 2017;22(02):112–117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Indrio F, Riezzo G, Cavallo L, Di Mauro A, Francavilla R. Physiological basis of food intolerance in VLBW. J Matern Fetal Neonatal Med 2011;24(Suppl 1):64–66 [DOI] [PubMed] [Google Scholar]
- 5.Chu A, Hageman JR, Caplan MS. Necrotizing enterocolitis. Neoreviews 2013;14(03):e113 [Google Scholar]
- 6.Gonzalez BE, Mercado CK, Johnson L, Brodsky NL, Bhandari V. Early markers of late-onset sepsis in premature neonates: clinical, hematological and cytokine profile. J Perinat Med 2003;31(01):60–68 [DOI] [PubMed] [Google Scholar]
- 7.Walsh MC, Kliegman RM. Necrotizing enterocolitis: treatment based on staging criteria. Pediatr Clin North Am 1986;33(01):179–201 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Buonomo C The radiology of necrotizing enterocolitis. Radiol Clin North Am 1999;37(06):1187–1198, vii [DOI] [PubMed] [Google Scholar]
- 9.Coursey CA, Hollingsworth CL, Wriston C, Beam C, Rice H, Bisset G III. Radiographic predictors of disease severity in neonates and infants with necrotizing enterocolitis. AJR Am J Roentgenol 2009;193(05):1408–1413 [DOI] [PubMed] [Google Scholar]
- 10.Kenton AB, O’Donovan D, Cass DL, et al. Severe thrombocytopenia predicts outcome in neonates with necrotizing enterocolitis. J Perinatol 2005;25(01):14–20 [DOI] [PubMed] [Google Scholar]
- 11.Sho S, Neal MD, Sperry J, Hackam DJ. A novel scoring system to predict the development of necrotizing enterocolitis totalis in premature infants. J Pediatr Surg 2014;49(07):1053–1056 [DOI] [PubMed] [Google Scholar]
- 12.Song R, Subbarao GC, Maheshwari A. Haematological abnormalities in neonatal necrotizing enterocolitis. J Matern Fetal Neonatal Med 2012;25(Suppl 4):22–25 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Maheshwari A. Immunologic and hematological abnormalities in necrotizing enterocolitis. Clin Perinatol 2015;42(03):567–585 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hornik CP, Benjamin DK, Becker KC, et al. Use of the complete blood cell count in late-onset neonatal sepsis. Pediatr Infect Dis J 2012;31(08):803–807 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lu Q, Duan H, Yu J, Yao Y. Are global coagulation and platelet parameters useful markers for predicting late-onset neonatal sepsis? Clin Lab 2016;62(1-2):73–79 [DOI] [PubMed] [Google Scholar]
- 16.Varsha RU, Rusia U, Sikka M, Faridi MM, Madan N. Validity of hematologic parameters in identification of early and late onset neonatal infection. Indian J Pathol Microbiol 2003;46(04):565–568 [PubMed] [Google Scholar]
- 17.Krediet T, Gerards L, Fleer A, van Stekelenburg G. The predictive value of CRP and I/T-ratio in neonatal infection. J Perinat Med 1992;20(06):479–485 [DOI] [PubMed] [Google Scholar]
- 18.Rosenfeld CR, Shafer G, Scheid LM, Brown LS. Screening and serial neutrophil counts do not contribute to the recognition or diagnosis of late-onset neonatal sepsis. J Pediatr 2019;205:105–111.e2 [DOI] [PubMed] [Google Scholar]
- 19.Kaban RK, Wardhana, Hegar B, et al. Lactobacillus reuteri DSM 17938 improves feeding intolerance in preterm infants. Pediatr Gastroenterol Hepatol Nutr 2019;22(06):545–553 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lucchini R, Bizzarri B, Giampietro S, De Curtis M. Feeding intolerance in preterm infants. How to understand the warning signs. J Matern Fetal Neonatal Med 2011;24(Suppl 1):72–74 [DOI] [PubMed] [Google Scholar]
- 21.Sharma D, Farahbakhsh N, Shastri S, Sharma P. Biomarkers for diagnosis of neonatal sepsis: a literature review. J Matern Fetal Neonatal Med 2018;31(12):1646–1659 [DOI] [PubMed] [Google Scholar]
- 22.Terrin G, Stronati L, Cucchiara S, De Curtis M. Serum markers of necrotizing enterocolitis: a systematic review. J Pediatr Gastroenterol Nutr 2017;65(06):e120–e132 [DOI] [PubMed] [Google Scholar]
- 23.Adams M, Bassler D. Practice variations and rates of late onset sepsis and necrotizing enterocolitis in very preterm born infants, a review. Transl Pediatr 2019;8(03):212–226 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rhee C, Kadri SS, Danner RL, et al. Diagnosing sepsis is subjective and highly variable: a survey of intensivists using case vignettes. Crit Care 2016;20:89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Goldstein GP, Sylvester KG. Biomarker discovery and utility in necrotizing enterocolitis. Clin Perinatol 2019;46(01):1–17 [DOI] [PubMed] [Google Scholar]
- 26.Lam HS, Cheung HM, Poon TC, et al. Neutrophil CD64 for daily surveillance of systemic infection and necrotizing enterocolitis in preterm infants. Clin Chem 2013;59(12):1753–1760 [DOI] [PubMed] [Google Scholar]
- 27.Däbritz J, Jenke A, Wirth S, Foell D. Fecal phagocyte-specific S100A12 for diagnosing necrotizing enterocolitis. J Pediatr 2012;161(06):1059–1064 [DOI] [PubMed] [Google Scholar]
- 28.Pammi M, Flores A, Leeflang M, Versalovic J. Molecular assays in the diagnosis of neonatal sepsis: a systematic review and metaanalysis. Pediatrics 2011;128(04):e973–e985 [DOI] [PubMed] [Google Scholar]
- 29.Luo J, Li HP, Xu F, Wu BQ, Lin HC. Early diagnosis of necrotizing enterocolitis by plasma RELMβ and thrombocytopenia in preterm infants: a pilot study. Pediatr Neonatol 2019;60(04):447–452 [DOI] [PubMed] [Google Scholar]
- 30.Mithal LB, Yogev R, Palac HL, Kaminsky D, Gur I, Mestan KK. Vital signs analysis algorithm detects inflammatory response in premature infants with late onset sepsis and necrotizing enterocolitis. Early Hum Dev 2018;117:83–89 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Batfalsky A, Lohr A, Heussen N, Neunhoeffer F, Orlikowsky TW. Diagnostic value of an interleukin-6 bedside test in term and preterm neonates at the time of clinical suspicion of early- and late-onset bacterial infection. Neonatology 2012;102(01):37–44 [DOI] [PubMed] [Google Scholar]
- 32.Buchegger P, Sauer U, Toth-Székély H, Preininger C. Miniaturized protein microarray with internal calibration as point-of-care device for diagnosis of neonatal sepsis. Sensors (Basel) 2012;12(02):1494–1508 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Stewart CJ, Nelson A, Treumann A, et al. Metabolomic and proteomic analysis of serum from preterm infants with necrotising entercolitis and late-onset sepsis. Pediatr Res 2016;79(03):425–431 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Giuliani S, Tan YW, Zheng D, et al. Coagulation gene expression profiling in infants with necrotizing enterocolitis. J Pediatr Gastroenterol Nutr 2016;63(06):e169–e175 [DOI] [PubMed] [Google Scholar]
- 35.Ververidis M, Kiely EM, Spitz L, Drake DP, Eaton S, Pierro A. The clinical significance of thrombocytopenia in neonates with necrotizing enterocolitis. J Pediatr Surg 2001;36(05):799–803 [DOI] [PubMed] [Google Scholar]
- 36.Assinger A, Schrottmaier WC, Salzmann M, Rayes J. Platelets in sepsis: an update on experimental models and clinical data. Front Immunol 2019;10:1687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Bedet A, Razazi K, Boissier F, et al. Mechanisms of thrombocytopenia during septic shock: a multiplex cluster analysis of endogenous sepsis mediators. Shock 2018;49(06):641–648 [DOI] [PubMed] [Google Scholar]
- 38.Wang YQ, Wang B, Liang Y, Cao SH, Liu L, Xu XN. Role of platelet TLR4 expression in pathogensis of septic thrombocytopenia. World J Emerg Med 2011;2(01):13–17 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Tate DY, Carlton GT, Johnson D, et al. Immune thrombocytopenia in severe neonatal infections. J Pediatr 1981;98(03):449–453 [DOI] [PubMed] [Google Scholar]
- 40.Fanaroff AA, Korones SB, Wright LL, et al. ; The National Institute of Child Health and Human Development Neonatal Research Network. Incidence, presenting features, risk factors and significance of late onset septicemia in very low birth weight infants. Pediatr Infect Dis J 1998;17(07):593–598 [DOI] [PubMed] [Google Scholar]
- 41.Samuels N, van de Graaf RA, de Jonge RCJ, Reiss IKM, Vermeulen MJ. Risk factors for necrotizing enterocolitis in neonates: a systematic review of prognostic studies. BMC Pediatr 2017;17(01):105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Gowda H, Norton R, White A, Kandasamy Y. Late-onset neonatal sepsis-a 10-year review from North Queensland, Australia. Pediatr Infect Dis J 2017;36(09):883–888 [DOI] [PubMed] [Google Scholar]
- 43.Sorsa A Epidemiology of neonatal sepsis and associated factors implicated: observational study at neonatal intensive care unit of Arsi University Teaching and Referral Hospital, South East Ethiopia. Ethiop J Health Sci 2019;29(03):333–342 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Temming LA, Dicke JM, Stout MJ, et al. Early second-trimester fetal growth restriction and adverse perinatal outcomes. Obstet Gynecol 2017;130(04):865–869 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Carr R Neutrophil production and function in newborn infants. Br J Haematol 2000;110(01):18–28 [DOI] [PubMed] [Google Scholar]
- 46.Esiaba I, Angeles DM, Milford TM, et al. Platelet-neutrophil interactions are lower in cord blood of premature newborns. Neonatology 2019;115(02):149–155 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Rajasekhar D, Barnard MR, Bednarek FJ, Michelson AD. Platelet hyporeactivity in very low birth weight neonates. Thromb Hae most 1997;77(05):1002–1007 [PubMed] [Google Scholar]
- 48.Maconi M, Rolfo A, Cardaropoli S, Brini M, Danise P. Hematologic values in healthy and small for gestational age newborns. Lab Hematol 2005;11(02):152–156 [PubMed] [Google Scholar]
- 49.Dahmane Ayadi I, Ben Hamida E, Youssef A, Sdiri Y, Marrakchi Z. Prevalence and outcomes of thrombocytopenia in a neonatal intensive care unit. Tunis Med 2016;94(04):305–308 [PubMed] [Google Scholar]
- 50.Dubruc E, Lebreton F, Giannoli C, et al. Placental histological lesions in fetal and neonatal alloimmune thrombocytopenia: a retrospective cohort study of 21 cases. Placenta 2016;48:104–109 [DOI] [PubMed] [Google Scholar]
- 51.Sahni M, Franco-Fuenmayor ME, Shattuck K. Management of late preterm and term neonates exposed to maternal chorioamnionitis. BMC Pediatr 2019;19(01):282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Del Vecchio A Evaluation and management of thrombocytopenic neonates in the intensive care unit. Early Hum Dev 2014;90(Suppl 2):S51–S55 [DOI] [PubMed] [Google Scholar]
- 53.Kazmi SH, Bailey SM, Mally PV, Verma S, Borkowsky W, Howell HB. The effect of complete blood count timing on lumbar puncture rates in asymptomatic infants born to mothers with chorioamnionitis. Cureus 2018;10(12):e3737. [DOI] [PMC free article] [PubMed] [Google Scholar]