This retrospective study was conducted at the level 4 neonatal intensive care unit (NICU) at the University of Mississippi Medical Center, a regional referral center, after approval by the Institutional Review Board (2017 − 0127). A detailed review of the electronic medical records identified 243 patients with medical and surgical NEC (NEC Bell stage II and above)(16) who underwent NEC management in the period between January 2013 and December 2018. We identified 65 infants with surgical NEC qualifying for the study (see Fig. 1).
Clinical information
We recorded demographic characteristics including birth weight, gestational age, sex, race (African American, Caucasian, or Latino), and mode of delivery (C-section / Vaginal delivery), APGAR scores at 5 minutes, out born status, and small for gestational age status. We collected information regarding maternal factors, including pregnancy-induced hypertension, chorioamnionitis, and antenatal steroids.
NEC information
We noted the NEC features such as the age of onset and clinical presentation (abdominal distension, feeding intolerance, and bloody stools). The NEC diagnosis was made on abdominal X-ray findings such as pneumatosis, pneumoperitoneum, and portal venous gas. We recorded information on Penrose drain, time to laparotomy, length and region of bowel resected, types of stoma creation following NEC surgery.
Histopathological Evaluation
Hematoxylin & eosin-stained surgical resected intestinal tissue sections were evaluated for necrosis, inflammation, hemorrhage, and reparative changes by a pediatric pathologist. A score of 0 was assigned when the exam appeared normal, 1 for 1–25% necrosis/ inflammation, 2 when 25–50% area involved, 3 when 50–75% area was affected, and 4 when > 75% changes were seen (17).
Postoperative Morbidity
To assess postoperative morbidity, we recorded the duration of postoperative ileus, days of parenteral nutrition (PN) days, intestinal failure (PN > 90days), and time to achieve full feeds. Short bowel syndrome was defined as infants who were still requiring TPN at discharge or more than 90 days after NEC onset. Days of parenteral nutrition were defined as the interval between postoperative day 1 until full enteral feedings were achieved (defined as 120 ml/kg/day). Surgical morbidity was classified as surgical site infections (including dehiscence and abscesses), strictures, fistulas, adhesions, and perforations.
We recorded information on the length of stay and mortality. The length of stay was defined as the total hospitalization duration from the day of admission until discharge or death. Mortality was defined as death due to any cause prior to hospital discharge.
We also collected data on bronchopulmonary dysplasia status at 36 weeks based on the oxygen requirement at the time of assessment(18).
Hematology information
We recorded complete blood cell count results from the electronic chart before the NEC onset (last available CBC inpatient record before NEC onset), on the day of NEC onset, 24 hours, and 48 hours after onset. We collected data on relative (presented as percentages) as well as on the absolute values. If we had multiple CBC on the same day, we recorded data from what we judged to be the most abnormal. We also collected data on platelet and RBC transfusion before and after the NEC onset.
Renal function data: We captured all serum creatinine (SCr) measurements and daily urine output (UOP) before and five days after NEC onset. After NEC onset, the incidence of AKI was determined using the modified neonatal staging criteria as previously described in the kidney disease: Improving Global Outcomes (KDIGO) Clinical Practice Guideline for AKI (19–23).
Neonatal MRI data
All MRI brain scans (without contrast) were scored independently by two pediatric neuroradiologists unaware of the infants' clinical course. Our NICU standard of care is to obtain a brain MRI at 36 weeks corrected age or before discharge whenever clinically feasible in neonates with birthweight less than 1500 grams. We used a standardized scoring system as used by Woodward et al. and consisting of eight 3-point scales (9) to evaluate the white matter brain injury and the gray matter brain injury.
Cerebellar Injury
We also assessed cerebellar lesions on brain MRI. We scored the scans on a binary scale with 0 being no injury and 1 indicating the presence of cerebellar injury. Cerebellar Injury patterns that we identified on MRI brain were cerebellar hemorrhage, siderosis and/or cerebellar volume loss. Hemorrhage (superficial siderosis and parenchymal hemorrhage) detection varied between susceptibility weighted imaging (SWI) and gradient recalled echo (GRE) techniques, performed on a 1.5T or 3T strength MRI, utilizing between 1.5-2mm slices for SWI and 4-5mm slices for GRE sequences. No slice gap was used for most scans, but a more remote exam utilized a 1.5mm slice gap. Volume loss was assessed with coronal and axial T2 weighted sequences on either 1.5T or 3T strength MRI, utilizing 4-5mm slices, and no slice gap, except for two scans, which utilized fast shunt protocol technique with 2.5mm slice gap and an older MRI utilizing a 1.5mm slice gap. Two of the scans were unable to assess for hemorrhage due to motion or utilization of T2-only shunt protocol technique. Most brain MRIs were performed using the GRE sequence, which is less sensitive for the detection of hemorrhage, and a minority were performed with the more sensitive SWI technique. Asymmetric volume loss in the cerebellum was contralateral to the germinal matrix hemorrhage in all cases, a known phenomenon associated with damage to crossing white matter tracts/transsynaptic degeneration. Hemorrhages varied between location in the vermis and cerebellar hemispheres. Superficial siderosis along the cerebellum and brainstem were also noted.
Neurodevelopment assessment at two years of age
At our center, infants underwent a neurodevelopmental comprehensive evaluation conducted by child development specialists using Bayley Scales of Infant Development (BSID-III) during the study period who were well aware of the MRI findings and the clinical course. We recorded cognitive and psychomotor development assessment scores. The Mental Development Index (MDI) assesses environmental responsiveness and sensory and perceptual abilities, memory, learning, and early language and communication abilities; the Psychomotor Development Index (PDI) assesses gross and fine motor skills.
Statistical Methods
Normally distributed continuous variables are summarized as means and standard deviations (± SD). Comparisons between normally distributed continuous measures for those with and without cerebellar injury were performed using Student's t-test for equal variance cases and Welch’s unequal variances t-test for unequal variances. For continuous data exhibiting non-normal distributions medians with interquartile range (IQR) [1st quartile; 3rd quartile] are presented, and differences were tested using the Kruskal-Walli’s test. Categorical data were summarized as counts with relative frequencies as percentages, and differences in the groups were analyzed using the Chi-squared test (χ² test) or Fisher's exact test.
Univariate logistic regression analyses examined the unadjusted association between each of the risk factors and cerebellar injury. Logistic regression analyses compared clinical and pathological findings among neonates with cerebellar injury to those without cerebellar injury. Multivariate logistic regression models were used to evaluate the adjusted associations between cerebellar injury and clinical-histological factors, using absence of cerebellar injury as the reference All tests were two-sided and a p-value < 0.05 was considered statistically significant. The statistical analyses were performed in SAS 9.4 statistical software.