Keywords
necrotizing enterocolitis - preterm Infants - BPD
Necrotizing enterocolitis (NEC) is one of the most common gastrointestinal conditions
requiring emergency surgery in the neonatal intensive care unit, affecting 5 to 10%
of premature infants with a birth weight (BW) ≤1,500 grams.[1]
[2] NEC remains a leading cause of morbidity and death among preterm infants and requires
increased hospital care and resource utilization.[3]
[4]
[5]
[6]
[7]
[8]
[9] NEC is associated with a severe systemic inflammatory response and septic shock-like
presentation that can contribute to multiorgan dysfunction due to intravascular volume
depletion, capillary leak syndrome, and hypotension.[10] Hemodynamic changes, surgical stresses, anesthetic agents, hypotension, and an inflammatory
surge affect multiple organs.[11]
[12] In addition, NEC has been associated with severe kidney injury and white matter
injury in neonates.
Preterm infants diagnosed with medical and surgical NEC are also more likely to be
diagnosed with BPD.[13] The high diagnosis rate of BPD in preterm term infants with NEC may be due to a
greater need for mechanical ventilation as a result of BPD's attendant volutrauma,
barotrauma, and hyperoxia effects on preterm lungs. Evolving evidence suggests that
interactions between the intestine and lungs are mediated by various pathways involving
the microbiome, immune system, and metabolites, as shown in animal models.[14] The intestinal dysbiosis, seen in preterm infants with NEC, modulates the gut immune
tolerance[15] and affects the premature lungs. The clinical impact of NEC on the underdeveloped
premature lung is not well studied. Currently, there is a lack of data on the pre-and
postoperative course, along with the surgical course, in infants who suffer from both
BPD and NEC, and how specific characteristics increase the risk for BPD.
In current literature, there is no study combining clinical, postoperative course,
and pathology findings to identify the subgroup of infants with surgical NEC at higher
risk of moderate/severe BPD. Therefore, we assessed moderate/severe BPD at 36 weeks
corrected gestational age (CGA) using the 2001 NIH consensus BPD definition.[16] This study aims to determine the clinical and intestinal pathological factors associated
with moderate/severe BPD at 36 weeks CGA in preterm infants before and following surgical
NEC.
Materials and Methods
Population and Study Design
The study was conducted in the level 4 neonatal intensive care unit at the University
of Mississippi Medical Center (UMMC), with approximately 1,000 annual admissions and
referrals from throughout the state. The UMMC Institutional Review Board had approved
this chart review study (2017–0127). All preterm infants admitted between January
2013 and December 31, 2018, with an NEC diagnosis (Bell's stage III) were included
in the study.[17] The Vermont Oxford Network (VON) database's 2019 report shows that UMMC has an NEC
rate of 12.8% (vs. 5.1% for all VON included hospitals) in infants with a BW less
than 1,500 grams. Infants diagnosed with medical NEC, spontaneous intestinal perforation,
congenital heart disease, intestinal atresia, and missing clinical data were excluded
from the analysis.
Demographic and Clinical Information
We collected demographic data, including gestational age (GA), BW, sex, appropriate
for gestational age (AGA) status, race, outborn status, mode of delivery, and Apgar's
score ≤6 at 5 minutes. We also collected maternal variables, including chorioamnionitis,
pregnancy-induced hypertension, and antenatal steroid usage. In addition, we collected
information on duration, FiO2 requirement, and mode of ventilation (invasive/noninvasive) before and following
NEC. Additional recorded clinical details included patent ductus arteriosus (PDA)
and indomethacin/ibuprofen treatment before NEC onset, frequency of PDA surgical ligation,
and inotrope (dopamine) use 24 hours after NEC onset. Sepsis-related variables included
blood culture-proven sepsis at the time of NEC onset and duration of antibiotics.
Blood culture-proven sepsis was defined as positive blood culture samples collected
at NEC onset. We also recorded the frequency of cholestasis (serum bilirubin >2 mg/dL)
after NEC onset. We also recorded growth data on weight, length, and weight for length
at 36 weeks corrected GA. Sex-specific Fenton growth charts were used for anthropometric
measurement. We also collected data on infants discharged on home oxygen.
Necrotizing Enterocolitis Information
NEC was defined using Bell's criteria.[17] The diagnosis of NEC was made by clinical signs/symptoms and based upon radiological
NEC findings, including pneumatosis, portal venous gas, and pneumoperitoneum on the
abdominal X-ray. The frequency of surgical NEC (Bell's stage III) was collected.[17] We recorded information on the age (in days) at the time of diagnosis of NEC. Neonates
who died 48 hours after NEC onset, or if massive bowel necrosis was found during laparotomy
or autopsy, were classified as having fulminant NEC.[18] Patients with NEC were managed surgically if they had clinical deterioration due
to intestinal perforation, hypotension, persistent electrolyte imbalances (e.g., hyperkalemia
and metabolic acidosis), increasing ventilator support, as well as anemia and thrombocytopenia
despite repeated red blood cell and platelet transfusions. The practice of surgical
management of NEC did not change substantially throughout the study period. At UMMC,
preterm infants with pneumoperitoneum who weigh less than 1 kg at the time of NEC
diagnosis and are hemodynamically unstable are treated first with a Penrose drain
at the bedside but may later receive laparotomy.
Histopathological Evaluation
Hematoxylin and eosin-stained surgical resected intestinal tissue sections were evaluated
for necrosis, inflammation, hemorrhage and reparative changes. A score of 0 was assigned
when the exam appeared normal, 1 for 1 to 25% necrosis/inflammation, 2 when 25 to
50% of the area was involved, 3 when 50 to 75% area was affected, and 4 when >75%
changes were seen.
Postoperative Information
To assess the impact of postoperative clinical factors of infants with surgical NEC
on BPD, data such as postoperative ileus days (defined as infants being NPO after
laparotomy), time to reach full feeds (≥120 mL/kg/d), total parenteral nutrition days,
length of stay, and mortality were measured. The length of hospitalization was defined
as the total hospital duration from the day of admission until discharge or death.
Mortality was defined as death due to any cause before hospital discharge.
We also recorded information on intestinal failure (parenteral nutrition >90 days)
and surgical morbidity which was classified as surgical site infections (including
dehiscence and abscesses), strictures, fistulas, adhesions, and perforations.
Bronchopulmonary Dysplasia Data
We collected data on bronchopulmonary dysplasia status at 36 weeks of corrected GA
and the type of steroid (hydrocortisone/dexamethasone) used during the clinical course.
Bronchopulmonary dysplasia was classified as mild, moderate, or severe based on the
oxygen requirement at assessment.[16] To determine the assessment time, patients were separated based on the GA of less
than 32 weeks and greater than or equal to 32 weeks. Patients born with a CGA of less
than 32 weeks were assessed at 36 weeks postmenstrual age or at discharge, whichever
came first, and patients born with a GA of greater than or equal to 32 weeks were
evaluated at 56 days after birth or discharge, whichever came first. Mild BPD for
each CGA group was classified as requiring >21% oxygen for at least 28 days and breathing
room air at the assessment time. Moderate BPD needed less than 30% oxygen at the assessment
time. Severe BPD was classified as requiring greater than or equal to 30% oxygen or
positive pressure ventilation at the assessment time. Since 2016, UMMC has used volume
ventilation and continuous positive airway pressure (CPAP) more frequently compared
to previous years. This study compared infants with moderate to severe BPD criteria
and infants with mild to none BPD.
Renal Function Data
The incidence of acute kidney injury (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]
[20]
[21]
[22]
[23] We examined serum creatinine (SCr) measurements and daily UOP the day before NEC
diagnosis, at NEC onset, at 24/48/72/96 hours after onset, and 1 week after NEC diagnosis.
The maximum AKI stage reported was the highest SCr or UOP AKI stage 7 days after NEC
onset. Our hospital laboratories measured SCr using the isotope dilution mass spectrometry
method. The UOP was estimated using weighed diapers recorded by bedside nurses in
the electronic medical charts. We stratified neonates as without severe (no or stage
1) and with severe (stage 2 and 3) AKIs as had been done in previous neonatal and
pediatric studies.[20]
[24]
Statistical Methods
We summarized the data as mean and standard deviation (± SD) for normally distributed
continuous variables. Comparisons between those with non/mild BPD and moderate to
severe BPD were performed using Student's t-test, Welch's t-test, and ANOVA with or without Welch–Satterthwaite estimation of degrees of freedom
depending on the equality of the variances. For continuous data exhibiting skewed
distribution, the median with interquartile range (IQR; first quartile; third quartile)
is presented, and differences in the data were tested using the Mann–Whitney U-test or Kruskal–Wallis test. Categorical data were summarized as counts with relative
frequencies as percentages, and differences in the groups were analyzed using the
Chi-square test (χ
2 test) or Fisher's exact tests when appropriate.[25] The multivariable effect of clinical and histopathological variables on the presence
of moderate to severe BPD in preterm infants with surgical NEC was further evaluated
with multiple logistic regression. We chose multivariable model candidate variables
based on a priori selection for clinical importance and required univariate p-values ≤0.15 for association tests with moderate to severe BPD. Evaluations for significant
multicollinearity led to the length of stay, necrosis, and GA being eliminated from
the multivariable modeling process. A p-value of <0.05 was considered statistically significant for all the analyses, and
all tests were two-sided. All the statistical analyses were performed in SAS (version
9.4).
Results
Ninety-two infants were included in the analysis. Out of 92, 77.1% (71/92) of infants
had moderate to severe BPD, and 22.8% (21/92) of infants had non/mild BPD at 36 weeks
CGA. The cohort had a median GA of 26.2 weeks (IQR: 24.3–29.3) and a median BW of
767.5 g (IQR: 640–1,125). The study had 34.8% (32/92) females and predominantly African–American
(77.2% [71/92]) infants. Fifty-six infants (60.9%) were outborn. 69.5% (64/92) infants
were born by Cesarean section, and 72.2% (65/90) received antenatal steroids. Additional
full demographic, maternal, and clinical information has been summarized in [Table 1].
Table 1
Demographic and clinical features of neonates with surgical NEC stratified by BPD
status
|
N
|
Overall
n (%)/median (IQR)
|
Non/mild BPD
n (%)/median (IQR)
|
Moderate to severe BPD
n (%)/median (IQR)
|
p-Value
|
|
|
N = 92
|
N = 21
|
N = 71
|
|
|
Pregnancy-induced hypertension
|
92
|
27 (29.3)
|
8 (38.1)
|
19 (26.8)
|
0.31
|
|
Maternal chronic hypertension
|
79
|
14 (17.7)
|
3 (16.7)
|
11 (18.0)
|
1
|
|
Chorioamnionitis
|
91
|
9 (9.9)
|
0 (0.00)
|
9 (12.9)
|
0.11
|
|
Antenatal steroids
|
90
|
65 (72.2)
|
9 (45.0)
|
56 (80.0)
|
0.002
|
|
Gestational age (wk)
|
92
|
26.2 (24.3, 29.3)
|
31.6 (28.4, 32.6)
|
25.2 (24, 27)
|
<0.001
|
|
Birth weight (g)
|
92
|
767.5 (640, 1,125)
|
1,600.0 (1,030.0, 2,095.0)
|
710.0 (620.0, 911.0)
|
<0.001
|
|
Small for gestational age
|
92
|
30 (32.6)
|
7 (33.3)
|
23 (32.4)
|
0.93
|
|
Gender (female)
|
92
|
32 (34.8)
|
5 (23.8)
|
27 (38.0)
|
0.22
|
|
Ethnicity
|
92
|
|
|
|
0.34
|
|
Caucasian
|
|
17 (18.5)
|
2 (9.5)
|
15 (21.1)
|
|
|
African American
|
|
71 (77.2)
|
18 (85.7)
|
53 (74.6)
|
|
|
Latino
|
|
2 (2.2)
|
1 (4.8)
|
1 (1.4)
|
|
|
Other
|
|
2 (2.2)
|
0 (0.00)
|
2 (2.8)
|
|
|
Mode of delivery
|
92
|
|
|
|
0.38
|
|
C-section
|
|
64 (69.6)
|
13 (61.9)
|
51 (71.8)
|
|
|
Vaginal
|
|
28 (30.4)
|
8 (38.1)
|
20 (28.2)
|
|
|
Apgar's score <6 at 5 min
|
91
|
28 (30.8)
|
4 (19.0)
|
24 (34.3)
|
0.18
|
|
Outborn
|
92
|
56 (60.9)
|
12 (57.1)
|
44 (62.0)
|
0.69
|
|
Patent ductus arteriosus
|
92
|
54 (58.7)
|
6 (28.6)
|
48 (67.6)
|
0.001
|
|
Patent ductus arteriosus, indomethacin treated
|
92
|
12 (13.0)
|
0 (0.00)
|
12 (16.9)
|
0.06
|
|
Patent ductus arteriosus, surgically ligated
|
92
|
5 (5.4)
|
0 (0.00)
|
5 (7.0)
|
0.58
|
|
Central line present (d)
|
87
|
52 (30, 80)
|
30.5 (14.5, 60)
|
59.0 (36.0, 93.0)
|
0.010
|
|
Positive blood culture sepsis
|
92
|
31 (33.7)
|
8 (38.1)
|
23 (32.4)
|
0.62
|
|
CRP on day of NEC onset (mg/L)
|
79
|
3.2 (1.2, 8)
|
2.6 (0.9, 4.6)
|
4.7 (1.4, 8.5)
|
0.08
|
|
CRP at 24 h after NEC onset (mg/L)
|
72
|
8 (3, 19.4)
|
5.4 (1.6, 17.0)
|
8.5 (3.0, 19.8)
|
0.16
|
|
CRP at 48 h after NEC onset (mg/L)
|
60
|
14.9 (3.3, 21.9)
|
7.8 (3.3, 20.3)
|
15.5 (3.9, 22.0)
|
0.39
|
|
CRP at 96 h after NEC onset (mg/L)
|
64
|
7.2 (4.1, 15.6)
|
5.4 (4.3, 24.5)
|
7.6 (4.1, 15.5)
|
0.93
|
|
CRP at 1 wk after NEC onset (mg/L)
|
64
|
5.3 (2.9, 8.2)
|
7.4 (6.8, 10.4)
|
4.8 (2.5, 8.0)
|
0.07
|
|
CRP at 2 wk after NEC onset (mg/L)
|
61
|
3.4 (1.9, 5.7)
|
8.1 (3.1, 14.7)
|
3.4 (1.7, 5.3)
|
0.03
|
|
Cholestasis at NEC onset
|
90
|
59 (65.6)
|
8 (42.1)
|
51 (71.8)
|
0.015
|
|
AKI by serum creatinine
|
92
|
|
|
|
0.24
|
|
No AKI
|
|
46 (50.0)
|
14 (66.7)
|
32 (45.1)
|
|
|
Stage 1
|
|
19 (20.7)
|
3 (14.3)
|
16 (22.5)
|
|
|
Stage 2
|
|
12 (13.0)
|
3 (14.3)
|
27 (38.0)
|
|
|
Stage 3
|
|
15 (16.3)
|
1 (4.8)
|
8 (11.3)
|
|
|
AKI by urine output
|
92
|
|
|
|
0.011
|
|
No AKI
|
|
47 (51.1)
|
14 (66.7)
|
33 (46.5)
|
|
|
Stage 1
|
|
4 (4.3)
|
1 (4.8)
|
3 (4.2)
|
|
|
Stage 2
|
|
28 (30.4)
|
1 (4.8)
|
27 (38.0)
|
|
|
Stage 3
|
|
13 (14.1)
|
5 (23.8)
|
8 (11.3)
|
|
|
Severe AKI vs. non-severe AKI
Stage 2/3 vs. 0/1
|
92
|
54 (58.7)
|
6 (28.6)
|
48 (67.6)
|
0.001
|
Abbreviations: AKI, acute kidney injury; BPD, bronchopulmonary dysplasia; CRP, C-reactive
protein (mg/L); IQR, interquartile range; NEC, necrotizing enterocolitis.
Note: Categorical variables are presented as count (percentage). Continuous variables
are presented as mean (standard deviation) and, if not normally distributed, as median
with interquartile range. Continuous measures' statistical associations with lung
injury were evaluated with the Kruskal–Wallis test. Differences in categorical measures'
associations were tested using the Chi-square test when expected cell counts were
adequate, otherwise Fisher-Freeman-Halton Exact test was used with low expected cell
counts. Bold values are statistically significant.
The infants with moderate to severe BPD had a later age of NEC onset (13 [7, 23] vs.
5 days [2, 1]; p = 0.001), less fulminant NEC (5.6 [4/71] vs. 33.3% [7/21]; p = 0.002), significantly higher frequency of PDA (67.6 [48/71] vs. 28.6% [6/21]; p = 0.001) before NEC, had more central line days (59 [IQR: 36, 93] vs. 30.5 days [IQR:
14.5, 60]; p = 0.01), higher rate of cholestasis (71.8% [51/71] vs. 42.1% [8/21]; p = 0.015), had more antenatal steroids (80% [56/71] vs. 45% [9/21]; p = 0.002), younger GA (25.2 weeks [IQR 24, 27] vs. 31.6 weeks [IQR 28.4, 32.6]; p <0.001), had lower BW (710 [IQR: 620, 911] vs. 1,600 grams [IQR: 1030, 2095]; p <0.001), and had more severe AKI (67.6% [48/71] vs. 28.6% [6/21]; p = 0.001) compared to infants with non/mild BPD.
Infants with moderate to severe BPD were more likely to require invasive ventilation
(76.3 [54/71] vs. 42.9% [9/21]; p = 0.020) at day 7 after birth and for a longer duration (7.5 [IQR: 4, 15] vs. 1.5
days [IQR: 1, 4]; p = 0.003) compared to infants with non/mild BPD. In addition, the infants with moderate
to severe BPD received significantly longer invasive support (40.5 [IQR: 12, 59] vs.
6 days [IQR: 2, 13]) and noninvasive respiratory support (51 [IQR: 25, 93] vs. 5 days
[IQR: 4, 8.5]; p ≤ 0.001) and needed higher FiO2 following surgical NEC compared to the non/mild BPD infants. In addition, those infants
with moderate to severe BPD following NEC were discharged on home oxygen more frequently
than infants with non/mild BPD (38.1 [16/42] vs. 0% [0/11]; p = 0.023). Additional details of this data are summarized in [Tables 2] and [3].
Table 2
NEC features and surgical morbidity in preterm infants with of surgical NEC stratified
by BPD status
|
N
|
Overall
n (%)/median (IQR)
|
Non/mild BPD
n (%)/median (IQR)
|
Moderate-to-severe BPD
n (%)/median (IQR)
|
p-Value
|
|
|
N = 92
|
N = 21
|
N = 71
|
|
|
Clinical presentation
|
92
|
|
|
|
0.80
|
|
Abdominal distension
|
|
83 (90.2)
|
19 (90.5)
|
64 (90.1)
|
|
|
Bloody stools
|
|
7 (7.6)
|
2 (9.5)
|
5 (7.0)
|
|
|
Feeding intolerance
|
|
2 (2.2)
|
0 (0.00)
|
2 (2.8)
|
|
|
Pneumatosis
|
92
|
41 (44.6)
|
13 (61.9)
|
28 (39.4)
|
0.06
|
|
Pneumoperitoneum
|
92
|
51 (55.4)
|
12 (57.1)
|
39 (54.9)
|
0.85
|
|
Portal venous gas
|
92
|
6 (6.5)
|
2 (9.5)
|
4 (5.6)
|
0.61
|
|
Age of NEC onset (d)
|
92
|
10 (5, 23)
|
5 (2, 10)
|
13.0 (7.0, 23.0)
|
0.001
|
|
Fulminant NEC
|
92
|
11 (12.0)
|
7 (33.3)
|
4 (5.6)
|
0.002
|
|
Present of Penrose drain
|
89
|
38 (42.7)
|
5 (25.0)
|
33 (47.8)
|
0.06
|
|
Surgery <48 h
|
92
|
62 (67.4)
|
17 (81.0)
|
45 (63.4)
|
0.13
|
|
Time to surgery (h)
|
|
|
21 (4.5, 48)
|
51.5 (24, 232)
|
0.001
|
|
Length of bowel resected (cm)
|
89
|
12.7 (4.3, 28.8)
|
27.5 (6.15, 44.95)
|
12.0 (4.3, 24.9)
|
0.05
|
|
Region of bowel resected
|
84
|
|
|
|
0.82
|
|
Small bowel resected
|
|
53 (63.1)
|
11 (57.9)
|
42 (64.6)
|
|
|
Large bowel resected
|
|
4 (4.8)
|
1 (5.3)
|
3 (4.6)
|
|
|
Combined large and small bowel resected
|
|
27 (32.1)
|
7 (36.8)
|
20 (30.8)
|
|
|
Presence of ileocecal valve
|
91
|
65 (71.4)
|
13 (61.9)
|
52 (74.3)
|
0.27
|
|
Surgical morbidity (infection, adhesions, strictures, and dehiscence)
|
92
|
36 (39.1)
|
3 (14.3)
|
33 (46.5)
|
0.008
|
|
More than one surgical morbidity (infection, adhesions, strictures, dehiscence)
|
92
|
12 (13.0)
|
1 (4.8)
|
11 (15.5)
|
0.28
|
|
Adhesions
|
92
|
15 (16.3)
|
1 (4.8)
|
14 (19.7)
|
0.17
|
|
Wound dehiscence
|
92
|
14 (15.2)
|
0 (0.00)
|
14 (19.7)
|
0.034
|
|
Wound infection
|
92
|
6 (6.5)
|
1 (4.8)
|
5 (7.0)
|
1
|
|
Stricture
|
92
|
8 (8.7)
|
0 (0.00)
|
8 (11.3)
|
0.19
|
|
Fistula
|
92
|
6 (6.5)
|
1 (4.8)
|
5 (7.0)
|
1
|
|
Compartment syndrome
|
92
|
2 (2.2)
|
1 (4.8)
|
1 (1.4)
|
0.40
|
|
Intestinal failure
|
79
|
42 (53.2)
|
2 (13.3)
|
40 (62.5)
|
<0.001
|
Abbreviations: BPD, bronchopulmonary dysplasia; NEC, necrotizing enterocolitis; IQR,
interquartile range.
Note: Categorical variables are presented as count (percentage). Continuous variables
are presented as mean (standard deviation) and, if not normally distributed, as median
with interquartile range. Continuous measures' statistical associations with lung
injury were evaluated with the Kruskal–Wallis test. Differences in categorical measures'
associations were tested using the Chi-square test when expected cell counts were
adequate; otherwise, Fisher–Freeman–Halton exact test was used with low expected cell
counts. Bold values are statistically significant.
Table 3
Ventilation, oxygenation, and anthropometric data in infants with surgical NEC classified
by BPD
|
Variable
|
N
|
Overall
n (%)/median (IQR)
|
Non/mild BPD
n (%)/median (IQR)
|
Moderate-to-severe BPD
n (%)/median (IQR)
|
p-Value
|
|
|
N = 92
|
N = 21
|
N = 71
|
|
|
Day 7 FiO2
|
51
|
28 (21, 40)
|
23.0 (21.0, 46.0)
|
29.0 (22.0, 36.0)
|
0.58
|
|
Day 7 ventilation mode
|
52
|
|
|
|
0.02
|
|
Room air
|
|
4 (7.7)
|
3 (21.4)
|
1 (2.6)
|
|
|
Intubated
|
|
35 (67.3)
|
6 (42.9)
|
29 (76.3)
|
|
|
CPAP
|
|
5 (9.6)
|
3 (21.4)
|
2 (5.3)
|
|
|
HFNC
|
|
5 (9.6)
|
2 (14.3)
|
3 (7.9)
|
|
|
Noninvasive (NIMV)
|
|
3 (5.8)
|
0 (0.00)
|
3 (7.9)
|
|
|
Mode of ventilation outborn
|
36
|
|
|
|
1
|
|
Intubated
|
|
35 (97.2)
|
4 (100.0)
|
31 (96.9)
|
|
|
NIMV
|
|
1 (2.8)
|
0 (0.00)
|
1 (3.1)
|
|
|
FiO2 admission outborn
|
37
|
44 (29, 70)
|
47.5 (36, 77.5)
|
44.0 (28.0, 70.0)
|
0.47
|
|
Invasive ventilation duration before NEC
|
60
|
7 (3, 13.5)
|
1.5 (1, 4)
|
7.5 (4, 15)
|
0.003
|
|
Non-invasive duration before NEC
|
32
|
8 (3.5, 15)
|
5.5 (2, 8)
|
9.5 (4, 17)
|
0.10
|
|
FiO2 7 d before NEC
|
31
|
25 (21, 40)
|
25.5 (21, 50)
|
25.0 (21.0, 38.0)
|
0.97
|
|
Ventilation mode 2 weeks after NEC
|
78
|
|
|
|
<0.001
|
|
Room air
|
|
12 (15.4)
|
9 (64.3)
|
3 (4.7)
|
|
|
Intubated
|
|
54 (69.2)
|
2 (14.3)
|
52 (81.3)
|
|
|
CPAP
|
|
5 (6.4)
|
1 (7.1)
|
4 (6.3)
|
|
|
HFNC
|
|
7 (9.0)
|
2 (14.3)
|
5 (7.8)
|
|
|
Invasive vent duration after NEC (d)
|
88
|
22 (8, 56)
|
6 (2, 13)
|
40.5 (12, 59)
|
<0.001
|
|
Noninvasive duration after NEC
|
73
|
41 (11, 71)
|
5 (4, 8.5)
|
51.0 (25.0, 93.0)
|
<0.001
|
|
FiO2 after 2 weeks of NEC
|
79
|
28 (21, 36)
|
21.0 (21.0, 21.0)
|
30.0 (24.0, 36.0)
|
0.001
|
|
Home O2
|
53
|
16 (30.2)
|
0 (0.00)
|
16 (38.1)
|
0.02
|
|
Weight at 36 week (g)
|
82
|
2,085 (1,800, 2,335)
|
2,332.0 (2,200.0, 2,445.0)
|
2,030.0 (1,800.0, 2,250.0)
|
0.03
|
|
Length at 36 weeks (cm)
|
81
|
42 (39, 44)
|
45.0 (43.0, 48.4)
|
41.3 (39.0, 43.8)
|
0.002
|
|
Weight for length
|
81
|
5 (4.5, 5.5)
|
5.1 (4.5, 5.4)
|
4.9 (4.5, 5.5)
|
0.68
|
|
Weight centile at 36 weeks
|
82
|
8 (2, 19)
|
18.0 (11.0, 26.0)
|
6 (2, 18)
|
0.06
|
|
Length centile at 36 weeks
|
81
|
2 (0, 10)
|
19.0 (4.0, 69.0)
|
2 (0, 9.5)
|
0.02
|
Abbreviations: CPAP, continuous positive airway pressure; fiO2, fraction of inspired oxygen; HFNC, high flow nasal cannula; IQR, interquartile range.
Note: Categorical variables are presented as count (percentage). Continuous variables
are presented as mean (standard deviation) and, if not normally distributed, as median
with interquartile range. Continuous measures' statistical associations with lung
injury were evaluated with the Kruskal–Wallis test. Differences in categorical measures'
associations were tested using the Chi-square test when expected cell counts were
adequate, otherwise Fisher–Freeman–Halton exact tests were used with low expected
cell counts. Fenton chart used for anthropometric measurement. Bold values are statistically
significant.
Postoperatively, those with moderate to severe BPD took significantly longer median
time to reach full feeds (73.5 [IQR: 32,106] vs. 58 days [IQR: 42, 65]; p = 0.024) and needed parenteral nutrition for a longer period (109 [IQR: 77, 147]
vs. 55 days [IQR: 19, 70]; p <0.001). Those with moderate to severe BPD had significantly higher surgical morbidity
(46.5% [33/71] vs. 14.3% [3/21]; p = 0.008) and a higher rate of intestinal failure (62.5 [40/79] vs. 13.3% [2/79];
p <0.001) compared to those with non/mild BPD. The preterm infants with moderate to
severe BPD had significantly lower median length at postnatal 36 weeks GA (41.2 [IQR:
39, 43.6] vs. 44.3 cm [IQR: 42.5, 48.4]; p = 0.003) compared to infants with non/mild BPD. Those infants with moderate to severe
BPD had significantly more postnatal use of steroids (67.6 [48/71] vs. 38.1% [8/21];
p = 0.015) and a longer median length of hospitalization (161 [IQR: 112, 186] vs. 64
days [IQR: 20, 91]; p< 0.001) than those with non/mild BPD ([Table 4]).
Table 4
Postoperative outcomes in preterm infants with surgical NEC stratified based on BPD
status
|
Variable
|
N
|
Overall
n (%)/median (IQR)
|
Non/mild BPD
n (%)/median (IQR)
|
Moderate-to-severe BPD
n (%)/median (IQR)
|
p-Value
|
|
|
N = 92
|
N = 21
|
N = 71
|
|
|
Postoperative Ileus days (d)
|
81
|
13 (9, 17)
|
9.5 (7, 14)
|
14.0 (10.0, 20.0)
|
0.07
|
|
Postoperative day at starting enteral feedings (d)
|
80
|
14 (10, 18.5)
|
10.0 (10.0, 14.0)
|
14.0 (11.0, 21.0)
|
0.11
|
|
Day of attainment of full enteral feedings (120 mL/kg)
|
67
|
67 (32, 89)
|
58.0 (42.0, 65.0)
|
73.5 (32, 106)
|
0.024
|
|
Duration of parenteral nutrition (d)
|
92
|
95 (57.5, 140)
|
55.0 (19.0, 70.0)
|
109.0 (77.0, 147.0)
|
<0.001
|
|
Breast milk
|
92
|
16 (17.4)
|
1 (4.8)
|
15 (21.1)
|
0.107
|
|
Donor milk
|
92
|
21 (22.8)
|
1 (4.8)
|
20 (28.2)
|
0.035
|
|
Formula feeds
|
92
|
51 (55.4)
|
9 (42.9)
|
42 (59.2)
|
0.187
|
|
Breast milk and formula feeds
|
92
|
17 (18.5)
|
3 (14.3)
|
14 (19.7)
|
0.753
|
|
Assisted ventilation (intubated)
|
90
|
|
|
|
0.009
|
|
Intubated
|
|
78 (86.7)
|
13 (68.4)
|
65 (91.5)
|
|
|
High flow nasal cannula
|
|
7 (7.8)
|
2 (10.5)
|
5 (7.0)
|
|
|
CPAP
|
|
3 (3.3)
|
2 (10.5)
|
1 (1.4)
|
|
|
Room air
|
|
2 (2.2)
|
2 (10.5)
|
0 (0.00)
|
|
|
24 h presser support
|
92
|
70 (76.1)
|
14 (66.7)
|
56 (78.9)
|
0.249
|
|
Postnatal use of steroids
|
92
|
56 (60.9)
|
8 (38.1)
|
48 (67.6)
|
0.015
|
|
Length of stay (d)
|
92
|
133 (85, 178)
|
64.0 (20.0, 91.0)
|
161.0 (112.0, 186.0)
|
<0.001
|
|
Death
|
92
|
22 (23.9)
|
8 (38.1)
|
14 (19.7)
|
0.083
|
Abbreviations: BPD, bronchopulmonary dysplasia; IQR, interquartile range; NEC, necrotizing
enterocolitis.
Note: Categorical variables are presented as count (percentage). Continuous variables
are presented as mean (standard deviation) and, if not normally distributed, as median
with interquartile range. Continuous measures' statistical associations with lung
injury were evaluated with the Kruskal–Wallis test. Differences in categorical measures'
associations were tested using the Chi-square test when expected cell counts were
adequate, otherwise Fisher–Freeman–Halton exact test were used with low expected cell
counts. Bold values are statistically significant.
The preterm infants with moderate to severe BPD had greater percent inflammation (p = 0.038) and numerically lower mean necrosis score (1.5 ± SD = 1.3 vs. 2.2 ± SD = 1.3;
p =0.04) on intestine histopathology compared to the non/mild BPD group following surgical
NEC ([Table 5]).
Table 5
Histopathological changes in infants with surgical NEC classified by BPD status
|
Overall
|
Non/mild BPD
|
Moderate to severe BPD
|
p-Value
|
|
N = 92
|
N = 21
|
N = 71
|
|
|
Necrosis, mean (± SD)
Median (IQR)
|
1.6 (± 1.3)
2 (0, 3)
|
2.2 (± 1.3)
3 (1, 3)
|
1.5 (± 1.3)
1.5 (0, 2)
|
0.04
|
|
Inflammation, mean (± SD)
median (IQR)
|
1.9 (± 1.0)
2 (1, 2.5)
|
1.9 (± 1.3)
2 (1, 3)
|
1.9 (± 0.9)
2 (1, 2)
|
0.64
|
|
Hemorrhage, mean (± SD)
Median (IQR)
|
2.3 (± 1.2)
2 (2, 3)
|
2.5 (± 1.5)
3 (2, 4)
|
2.2 (± 1.1)
2 (2, 3)
|
0.34
|
|
Reparative change, mean (± SD)
Median (IQR)
|
0.4 (± 0.5)
0 (0, 1)
|
0.3 (± 0.5)
0 (0, 1)
|
0.5 (± 0.5)
0 (0, 1)
|
0.24
|
|
Necrosis, n (%)
|
0 (0, 1)
|
0 (0, 1)
|
0 (0, 0)
|
0.08
|
|
0
|
23 (29.1)
|
3 (17.6)
|
20 (32.3)
|
|
|
< 25%
|
13 (16.5)
|
2 (11.8)
|
11 (17.7)
|
|
|
25–50%
|
18 (22.8)
|
2 (11.8)
|
16 (25.8)
|
|
|
50–75%
|
20 (25.3)
|
9 (52.9)
|
11 (17.7)
|
|
|
> 75%
|
5 (6.3)
|
1 (5.9)
|
4 (6.5)
|
|
|
Inflammation, n (%)
|
|
|
|
0.03
|
|
0
|
5 (6.3)
|
3 (17.6)
|
2 (3.2)
|
|
|
< 25%
|
23 (29.1)
|
4 (23.5)
|
19 (30.6)
|
|
|
25–50%
|
31 (39.2)
|
3 (17.6)
|
28 (45.2)
|
|
|
50–75%
|
14 (17.7)
|
5 (29.4)
|
9 (14.5)
|
|
|
> 75%
|
6 (7.6)
|
2 (11.8)
|
4 (6.5)
|
|
|
Hemorrhage, n (%)
|
|
|
|
0.08
|
|
0
|
6 (7.6)
|
3 (17.6)
|
3 (4.8)
|
|
|
< 25%
|
13 (16.5)
|
1 (5.9)
|
12 (19.4)
|
|
|
25–50%
|
28 (35.4)
|
4 (23.5)
|
24 (38.7)
|
|
|
50–75%
|
17 (21.5)
|
3 (17.6)
|
14 (22.6)
|
|
|
> 75%
|
15 (19.0)
|
6 (35.3)
|
9 (14.5)
|
|
|
Reparative change, n (%)
|
33 (41.8)
|
5 (29.4)
|
28 (45.2)
|
0.24
|
Abbreviations: BPD, bronchopulmonary dysplasia; IQR, interquartile range; NEC, necrotizing
enterocolitis; SD, standard deviation.
Note: Categorical variables are presented as count (percentage). Continuous variables
are presented as mean (standard deviation) and, if not normally distributed, as median
with interquartile range. Continuous measures' statistical associations with lung
injury were evaluated with the Kruskal–Wallis test. Differences in categorical measures'
associations were tested using the Chi-square test when expected cell counts were
adequate, otherwise Fisher-Freeman-Halton exact tests were used with low expected
cell counts. Bold values are statistically significant.
On multivariable logistic regression analysis, lower BW (adjusted OR = 0.3, [95% confidence
interval (CI): 0.1–0.5]; p = 0.001), the presence of PDA (adjusted OR = 10.3, [95% CI: 1.6–65.4]; p = 0.0136), and longer parenteral nutritional days (adjusted OR = 8.8, [95% CI: 2.0–43.0];
p = 0.005) were associated with higher odds of moderate to severe BPD. The multivariable
logistic regression model results for factors associated with moderate to severe BPD
are summarized in [Table 6]. Surgical morbidity was an independent risk factor for moderate to severe BPD after
removing BW from the regression model (adjusted OR = 6, [95% CI: 1.2–38.7]; p = 0.034).
Table 6
Factors associated with moderate-to-severe BPD with multiple logistic regression analysis
|
Multiple logistic regression (moderate-to-severe BPD vs. non-/mild BPD)
|
|
Effect
|
OR
|
95% CI
|
p-Value
|
|
|
Lower CL
|
Upper CL
|
|
|
Birth weight (grams per standard deviation)
|
0.281
|
0.078
|
0.530
|
0.0013
|
|
Patent ductus arteriosus (present vs. absent)
|
10.276
|
1.614
|
65.412
|
0.0136
|
|
Any surgical morbidity (present vs. none)
|
4.593
|
0.748
|
28.185
|
0.0996
|
|
TPN days (per standard deviation of days)
|
8.822
|
1.961
|
42.992
|
0.005
|
|
Severe AKI vs. nonsevere AKI
|
4.769
|
0.768
|
29.609
|
0.0936
|
Abbreviations: AKI, acute kidney injury; BPD, bronchopulmonary dysplasia; CI, confidence
interval; NEC, necrotizing enterocolitis; OR, odds ratio; TPN, total parenteral nutrition.
Notes: Bold values are statistically significant.
Multiple logistic regression with stepwise selection (entry univariate significance
level p≤ 0.15) was performed, the retained variables with birth weight are listed above after
selection. Candidate variables into the model include:
• Patent ductus arteriosus (present vs. absent).
• NEC age onset (per cohort standard deviation in days).
• Any surgical morbidity (present vs. none).
• TPN days (per cohort standard deviation in days).
• Severe AKI vs. nonsevere AKI.
• Higher degree of inflammation (present vs. absent).
• Time to surgery from NEC onset (per cohort standard deviation in days).
• Combined FiO2.
Combined FiO2 is defined as the combination of FiO2 admission outborn and day 7 FiO2. Higher degree of inflammation was defined as inflammation involving >25% of the
specimen area. The OR and its 95% CI for the continuous variables were expressed as
per standard deviation of the variable in the cohort.
Discussion
Approximately three-fourths (77%) of infants following surgical NEC had moderate to
severe BPD in this cohort. Those with moderate to severe BPD had significantly lower
GA, BW, and a higher rate of PDA. In addition, these infants had late NEC onset and
severe clinical course postoperatively compared to infants with non/mild BPD. Infants
with moderate to severe BPD following NEC needed higher respiratory support for a
longer duration, received parental nutrition and central line support for longer periods
of time, had more AKI, intestinal failure, cholestasis, achieved enteral feeds >120 mL/kg
later, and more often went home on oxygen support. These findings indicate a higher
degree of lung compromise by NEC with a more severe inflammatory process and associated
multisystem morbidity. We also noticed infants with moderate to severe BPD had poor
linear growth at 36 weeks following NEC, as shown in a few reports.[26]
[27]
[28]
In this cohort, preterm infants with moderate to severe BPD had more severe AKI (67.6
vs. 28.6%) following NEC. AKI appears not to be an isolated event but instead reflects
remote multiorgan dysfunction involving the lungs, heart, liver, intestines, and brain
through an inflammatory mechanism involving neutrophil migration and cytokine expression
increased oxidative stress as shown in animal models.[29] A recent study has also shown that AKI was associated with higher odds of BPD in
preterm infants born between 29 and 32 weeks of gestation.[30] The fluid overload in an oliguric state or decreased renal function caused by AKI
may affect the lung mechanics and exacerbate the BPD in preterm infants with surgical
NEC.
In our cohort, surgical morbidity, especially wound dehiscence, was significantly
and independently associated with moderate to severe BPD, but only after removing
BW from the multivariable regression model (OR = 6.7, [95% CI: 1.2–38.7]; p = 0.034). This exploratory analysis, was not prespecified. It suggests that a greater
severity of underlying intestinal pathology and requirement for a second surgery may
be associated with higher odds of moderate to severe BPD. We hypothesize that the
stresses of the second laparotomy, including anesthesia, sedation, paralytic agents,
blood products, and assisted ventilation, expose the underlying preterm lung to additional
oxidative, hypoxic, and inflammatory injury in preterm infants with surgical NEC.
The consequences of a second laparotomy need further study, as well as the contribution
of BW in understanding these associations.
Those with moderate to severe BPD had higher inflammatory percent involvement but
less necrosis on intestinal pathology. We previously reported that incomplete resection
of inflamed intestine was associated with longer hospital stay and mortality than
complete resection.[31] The reasons and mechanisms leading to greater severity of BPD in preterm infants
are still not fully understood. Toll-like receptor 4 activation by LPS has been associated
with NEC in animal model studies.[32]
[33]
[34] Jia et al, have shown that TLR4 expression in the lung gradually increases during
postnatal development and its association with NEC-associated BPD.[35] They have demonstrated that mice and humans with NEC-associated lung inflammation
express higher pulmonary TLR4 than age-matched controls. The intestinal epithelial
TLR4 activation induced high-mobility group box 1 release from the intestine, which
activated pulmonary epithelial TLR4, leading to the induction of the neutrophil recruiting
CXCL5 and the influx of proinflammatory neutrophils to the lung. The aerosolized administration
of a carbohydrate TLR4 inhibitor prevented CXCL5 upregulation and blocked NEC-induced
BPD in mice model.[35]
The adult models of intestinal injury have shown that there is cross-talk and homeostasis
between the intestinal and respiratory systems via various pathways involving the
microbiome, immune system, and metabolites.[14] The studies have shown that intestinal dysbiosis modulates the gut immune tolerance
and response by affecting the alteration of dendritic cell priming of the T cells,
involving various cytokines.[15] In addition, short-chain fatty acids, such as propionate produced by intestinal
bacteria, can modulate lung inflammation in a mice model.[36] A multicenter clinical study has demonstrated that infants who developed NEC had
elevated interleukin (IL)-1β, IL-6, IL-8, and IL-10, monocyte chemoattractant protein-1/CC-motif
ligand-2, macrophage inflammatory protein-1β/CC-motif ligand-3 and C-reactive protein,
creating a pro-inflammatory state.[37] In an extension of that study, the overall cytokine pattern generated suggests that
BPD might be associated with impairment in the transition from innate immune response
mediated by neutrophils to the adaptive immune response mediated by T-lymphocytes.[38]
Limitations
Limitations include that this study is a single-center retrospective albeit consecutive
experience, perhaps reducing the study's generalizability. UMMC had an NEC rate of
12.8%, higher than other centers in the VON 2019 data. In our cohort, most neonates
with surgical NEC are African American, likely due to the race distribution in Mississippi
and the patient population of the UMMC. Between 2018 and 2020, 43% of infants born
in Mississippi were African American, and approximately 60% of UMMC patients self-identify
as African American. Since 2017, UMMC has used donor breast milk for infants less
than 34 weeks when the mother's breast milk was not available. As per VON 2020 data,
in infants with less than 1,500 grams, UMMC's exclusive breast milk usage rate is
low (18 vs. 58%) compared to all VON network hospitals. In this cohort, sample size
limits our ability to detect associations between clinical factors, NEC, BPD, and
outcomes. Multiple comparisons yield a higher probability of Type I errors. Our study
is observational, retrospective, and descriptive. Thus, causality cannot be established
with or without formal adjustment for multiple comparisons. Given the limited availability
of prior information on factors correlating BPD status in neonates with surgical NEC,
our study is exploratory. It should be considered hypothesis generating with regard
to the level of evidence provided.
Conclusion
In conclusion, three-fourths of preterm infants had moderate to severe BPD following
surgical NEC. Infants developing moderate to severe BPD were significantly more likely
to have a PDA. Surgical NEC infants with moderate to severe BPD were more likely to
experience longer postoperative morbidity, acute kidney injury, intestinal failure,
greater surgical complications, poor linear growth, require greater respiratory support,
and longer length of hospitalization. Further research is required to identify and
develop lung protection strategies in preterm infants with surgical NEC. Improved
risk prediction for the development of BPD in surgical NEC infants from the postoperative
time until discharge might help target lung protection strategies. Given the relative
infrequency of surgical NEC, larger multicenter prospective and translational studies
with adequate sample size are required to develop sound evidence-based risk stratification
strategies, including inflammatory, metabolomics, nutritional, intestinal/lung microbiome,
and genomic data in preterm infants with surgical NEC. Our study suggests the need
to understand the inflammatory surge better and develop immunomodulatory strategies
in preterm infants with surgical NEC.
