University of Southern California Dissertations and Theses

Alkaline phosphatase levels in infants with spontaneous intestinal perforation vs. necrotizing enterocolitis with perforation

(USC Thesis Other)

Alkaline phosphatase levels in infants with spontaneous intestinal perforation vs. necrotizing enterocolitis with perforation

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1

Alkaline Phosphatase Levels in Infants with Spontaneous Intestinal Perforation vs.
Necrotizing Enterocolitis with Perforation.
Timur A. Azhibekov, M.D.
Department of Preventive Medicine
Master of Clinical, Biomedical and Translational Investigations
“University of Southern California”
December 13, 2018

2
Table of Contents

Abstract 3
Background and Significance 4
Methods 6
Study Design 6
Study Subjects 6
Definitions of Exposure Variables 7
Definitions of Secondary Outcomes 8
Data Transformation 8
Statistical Analysis 8
Results 9
Alkaline Phosphatase 10
Demographic and perinatal factors 11
Feedings 12
Disease presentation 12
Multivariate analysis and predictive monitoring 14
Other outcomes 15
Discussion 16
References 28
Tables and Figures 38

3
ABSTRACT
Objectives: To investigate changes in serum alkaline phosphatase levels in infants
with bowel perforation secondary to spontaneous intestinal perforation (SIP) or surgical
necrotizing enterocolitis (NEC). In conjunction with other risk factors, identify possible
role of serum alkaline phosphatase level in differentiating between these two conditions.
Methods: Retrospective case-control study of infants admitted with bowel
perforation to Children’s Hospital Los Angeles Newborn and Infant Critical Care unit from
2005 to 2015. Only infants diagnosed with SIP (cases) or surgical NEC (controls) were
included in the study. Demographic and prenatal data, information on postnatal exposures,
clinical, laboratory and radiographic findings, and outcomes were collected and analyzed.
Univariate logistic regression analysis was used to study associations between each risk
factor and case-control status, followed by bivariate analysis using adjustment for
gestational age (GA). Mixed linear models were used for repeat measurements. For
explanatory and predictive modeling, we utilized multivariate logistic regression with
subsequent testing for goodness of fit and receiver operating characteristic (ROC) curve
analysis.
Results: A total of 114 infants were included in the study: 48 infants with SIP (cases)
and 66 infants with NEC (controls). Cases had significantly lower GA and birth weight
compared to controls: 25.6 [IQR 24.9-27] vs. 28.2 [IQR 25.6-31.1] weeks (p=0.0002) and
785 [IQR 640-930] vs. 1,111 [IQR 790-1,758] grams (p=0.0001), respectively. In addition,
cases presented on an earlier median day of life (DOL) 8 vs. DOL 14 in controls
(p=0.0001). Exposures to antenatal steroids (83% vs. 52%, p=0.001) and postnatal
indomethacin (63% vs. 26%, p=0.0001) were significantly more frequent in cases. Infants
with combined exposure to steroids and indomethacin had even higher odds of having SIP
vs. NEC. Alkaline phosphatase levels on admission were significantly higher in cases:
median of 782 vs. 236 Units/L in controls (p<0.0001). Serum enzyme levels rapidly
decreased during the following 3 days, especially in infants with SIP, and remained
significantly higher in cases one week after the disease onset. In a multivariate logistic
regression analysis, GA, DOL at admission, combined exposure to steroids (both antenatal
and postnatal) and indomethacin, serum alkaline phosphatase level >500 Units/L, and
4
presence of pneumoperitoneum on initial radiograph remained significantly associated
with SIP. The final predictive model included GA, DOL at admission, combined exposure
to steroids and indomethacin, and serum alkaline phosphatase level >500 Units/L, with
overall good model fit as well as excellent discrimination with area under the curve (AUC)
0.93 (95%CI 0.88–0.98, p<0.0001). Cases had significantly lower mortality than controls
(6% vs. 42%, p<0.0001).
Conclusion: Following adjustment for other known risk factors, transient increase in
serum alkaline phosphatase level is significantly associated with SIP when compared to
NEC. Detection of such increase may assist in prompt diagnosis of SIP with potentially
significant clinical and prognostic implications.
BACKGROUND & SIGNIFICANCE
Spontaneous intestinal perforation (SIP) of the newborn is a common gastrointestinal
(GI) pathology in premature infants, and occurs in approximately 2-3% of very low birth
weight (VLBW, birth weight <1,500 g) infants (Alpan et al., 1985; Blakely et al., 2006;
Kawase, Ishii, Arai, & Uga, 2006). SIP is usually described as a single intestinal perforation
typically at the level of terminal ileum (Drewett & Burge, 2007; Holland, Shun, Martin,
Cooke-Yarborough, & Holland, 2003; Meyer, Payne, & Roback, 1991; Pumberger, Mayr,
Kohlhauser, & Weninger, 2002; Uceda, Laos, Kolni, & Klein, 1995). Histopathologic
features include focal hemorrhagic necrosis at the area of perforation with clearly defined
margins that is surrounded by intact bowel both proximally and distally. These
characteristics distinguish SIP from coagulative (ischemic) necrosis with severe intestinal
inflammation affecting the terminal ileum and colon as seen in patients with necrotizing
enterocolitis (NEC) (Ballance, Dahms, Shenker, & Kliegman, 1990), one of the most
common GI emergencies in the newborn. In addition to ischemic necrosis of the intestinal
mucosa with inflammation, NEC is also characterized by the invasion of enteric gas
forming organisms, and dissection of gas into the muscularis layer of the intestine and
portal venous system (Neu, 1996).
Multiple studies investigated the risk factors associated with these two disorders, and
a growing body of evidence supports the paradigm that SIP and NEC are two separate
5
clinical entities (Meyer et al., 1991; Pumberger et al., 2002; Adderson, Pappin, & Pavia,
1998; Cass et al., 2000) with different prognosis as well as short- and long-term outcomes.
Therefore, early differentiation between SIP and NEC is important for both management
and parental counseling considerations. While survival of infants with SIP improved in the
last few decades, and survival rates vary between 64 and 90 percent (Adesanya et al., 2005;
Pumberger et al., 2002; Zamir et al., 1988), NEC is still associated with significantly higher
compared to SIP morbidity and mortality in premature neonates, particularly in VLBW
infants, thus early recognition and aggressive treatment are warranted (Adesanya et al.,
2005; Snyder et al., 1997).
In the clinical settings, lack of reliable markers that would be useful for differentiation
between the two disease processes makes it difficult to establish the correct diagnosis based
on the history, clinical signs and symptoms or radiographic findings. Ultimately, the
surgeon in the operating room usually makes the final diagnosis during exploratory
laparotomy. Earlier gestational and chronological age, lower birth weight, exposure to
steroids and non-steroidal anti-inflammatory medications are among the factors that have
been reported to be associated with SIP when compared to NEC. Data on laboratory
markers of SIP vs. NEC remains very sparse. In a small case series, Harms et al. reported
significant increase in serum alkaline phosphatase levels in premature infants with
spontaneous intestinal perforation (Harms, Lüdtke, Lepsien, & Speer, 1995). In an animal
model, bowel ischemia but not infarction led to increase in serum levels of intestinal
alkaline phosphatase (IAP) (Barnett, Davidson, & Bradley, 1976). In contrast, decrease in
IAP expression and activity was reported in the intestines of the newborn animals with
NEC compared to healthy controls (Whitehouse et al., 2010). Subsequently, experimental
use of IAP as a therapeutic agent has been shown to decrease systemic inflammatory
response associated with NEC in animal models (Rentea et al., 2012; Riggle et al., 2013;
Whitehouse et al., 2010).
The purpose of this study was to investigate the role of serum alkaline phosphatase
in the diagnosis of SIP compared to NEC, and in conjunction with other known and
potential factors develop a predictive model that may assist in differentiating these patients
from the ones with surgical NEC. We hypothesized that serum alkaline phosphatase levels
6
would be significantly higher in neonates with SIP compared to neonates with perforation
due to NEC after adjustment for other risk factors. In addition, we compared mortality
associated with both conditions and length of hospital stay of these infants.
METHODS
Study Design
A retrospective case-control study design was utilized using clinical data obtained
from inpatient medical records of eligible infants. The study was approved by the
Institutional Review Board at Children’s Hospital Los Angeles (CHLA).
Study Subjects
Infants admitted to the CHLA Newborn and Infant Critical Care unit (NICCU) from
March 2005 to March 2015 with the diagnosis of intestinal perforation were eligible to be
included into the study. Infants with any GI conditions that were determined as a cause of
intestinal perforation other than SIP or NEC were excluded. Baseline demographic data
such as birth weight (BW), gestational age (GA), sex, maternal race/ethnicity, and
gravidity/parity were collected. Information on maternal prenatal care and delivery
(presence of prenatal care, substance abuse, intrauterine growth restriction,
oligohydramnios, preeclampsia, eclampsia or pregnancy-induced hypertension, preterm
premature rupture of membranes (PPROM), clinical chorioamnionitis, placental abruption,
steroid or magnesium administration, method of delivery, Apgar scores at 1 and 5 minutes
of life), data on clinical findings, laboratory (alkaline phosphatase levels, blood cell counts,
serum transaminases and bilirubin levels, C-reactive protein (CRP), coagulation studies,
blood gas analysis, lactate, blood urea nitrogen (BUN) and creatinine levels) and
radiographic characteristics were collected as well. We also reviewed and extracted for
further analysis information on therapeutic interventions (mechanical ventilation,
surfactant administration, treatment with postnatal steroids, indomethacin, vasopressor-
inotropes, antibiotics and/or antifungal agents), patients’ length of stay and whether they
expired prior to discharge.
7
Included infants were subsequently divided into two groups/study arms based on the
primary outcome of interest. Subject assignment was made based on patient’s clinical
features and diagnosis at the time of illness and, whenever available, supported by the
findings during exploratory laparotomy and confirmed by the surgical pathology
examinations. Infants with SIP were assigned as cases. The control group included infants
diagnosed with surgical NEC with bowel perforation (Bell’s stage IIIB). Three subjects
that had contradicting diagnoses based on surgical findings vs. pathology examinations
were excluded.
Definitions of Exposure Variables
Demographic characteristics were obtained from the medical records. Prenatal care
was defined as more than one prenatal visit during the index pregnancy. Substance abuse
was defined as consumption of any drugs of abuse during the index pregnancy as
documented in medical records. Exposure to antenatal steroids was defined as one or more
doses of steroids (betamethasone or dexamethasone) given prior to delivery. Invasive
respiratory support at the time of admission was divided into 3 categories: no mechanical
ventilation, conventional mechanical ventilation, and high-frequency ventilation. Exposure
to postnatal steroids was defined as treatment with dexamethasone or hydrocortisone,
regardless of indications, that took place before the disease onset. Treatment with any of
the three vasopressors and/or inotropes (dopamine, dobutamine, epinephrine, or any
combination of thereof) following the onset of SIP or NEC constituted exposure to
vasopressor-inotropes as an indicator of disease severity.
Feeding data, when available, was categorized based on the amount and type of the
feedings prior to illness. Feeding amount was defined as a maximal amount of feeds that
infant had reached since birth and prior to the development of intestinal perforation. For
example, an infant that was not fed (“nil per os”, or NPO) on admission but had reached
partial or full feeds prior to illness was classified as partial or full feeds, respectively. As
for the feeding type, subjects were divided based on the feeds (breast milk, formula, or
both) they received during the time preceding the onset of the disease.
8
Day of illness on admission was defined as number of days between the onset of the
disease and the time of admission. Serum alkaline phosphatase levels on the day of
admission (Day 1) and, when available, on the day prior to admission, day two, three, four
and seven after admission to the NICCU were collected for analysis. For all other
laboratory indicators, we used values at the time of admission. Thrombocytopenia was
defined as platelet counts less than 100,000 per microliter. Coagulopathy was defined as
any abnormality of the coagulation panel.
Serum glucose levels of more than 150 mg/dL were defined as hyperglycemia.
Infants that had no urine output at the time of admission were classified as having anuria.
Definitions of Secondary Outcomes
Mortality was defined as death at any time during the current hospital stay in the
NICCU. Length of stay was defined as number of inpatient days from admission to the
NICCU till either death, transfer back to the referring facility, or discharge home.
Data Transformation
For regression analysis, a number of continuous variables representing mostly
laboratory data were transformed into arbitrary units by multiplying each value by
either 10
-3
(white blood cell (WBC) and absolute neutrophil counts (ANC)), or 10
-2
(BW
and admission weight, alkaline phosphatase levels), or 10
-1
(transaminases, blood urea
nitrogen (BUN)), or 10 (pH, creatinine). This allowed providing more meaningful and
clinically relevant interpretation of the associations between these factors and the
outcomes. In addition, continuous variables were centered on their mean values.
Statistical Analysis
SAS
®
University Edition statistical software (SAS Institute Inc., Cary, NC, USA) was
used for analysis. All figures were created using SAS
®
9.4 Output Delivery System (ODS)
Graphics procedures. Continuous exposure variables were presented as mean ± standard
deviation (SD) or median [interquartile range (IRQ)] based on the data distribution for each
variable. Categorical variables were presented as number of patients in the group
(percentage). Variables with missing data points are indicated in tables with an asterisk,
9
and denominators were adjusted accordingly to calculate percentages. All exposure
variables were compared between SIP and NEC using Student’s t-test or Mann-Whitney
test for continuous variables and Fisher’s exact test or Chi-square test for categorical
variables, as appropriate. Continuous variables of interest were dichotomized based on the
cutpoints that maximized sensitivity and specificity. Exposure variables that were found to
be statistically significant (p<0.05) in the univariate analysis were then adjusted for GA
using logistic regression to determine GA-adjusted odds ratios. To evaluate relationships
between various exposure variables, we used correlation analysis and linear regression.
Mixed effects linear modeling procedures were used to analyze serial changes in alkaline
phosphatase levels over time between the two groups accounting for repeat measurements
by using maximum likelihood estimation method. Case-control status and day of alkaline
phosphatase measurement were used as fixed effects, and individual patients as a random
effect. For explanatory and predictive modeling, we used multivariate logistic regression
to determine the independent impact of each covariate on the diagnosis of SIP vs NEC, and
final models were subsequently tested for goodness of fit and other model diagnostics.
Receiver operating characteristic (ROC) curve analysis was used to evaluate classification
accuracy of individual independent variables of interest and overall discrimination of the
various models. To analyze mortality and length of stay, Chi-square test and Mann-
Whitney test were utilized, respectively. In addition, linear regression analysis was applied
to analyze relationships between the length of stay and GA.
RESULTS
114 infants met both inclusion and exclusion criteria and were included in the study.
The SIP group was comprised of 48 infants (cases). In 34 SIP cases (71%), diagnosis was
confirmed by findings of exploratory laparotomy and pathology examinations. 14 SIP
infants (29%) had only a Penrose drain placed and did not require further surgical
interventions. Subgroup analysis comparing these infants with the rest of the infants with
SIP showed that there was no difference in demographic characteristics, prenatal or
intrapartum factors, and disease presentation, including radiographic and laboratory
findings (data not shown). Of note, infants with SIP that required drain only were not
exposed to postnatal steroids in contrast to the rest of the cases (0% vs. 38.2%, p=0.01). 66
10
infants with NEC (controls) were diagnosed with NEC based on clinical presentation,
findings during surgery and surgical pathology examinations.
1. Alkaline Phosphatase.
Infants with SIP presented with significantly higher levels of alkaline phosphatase on
admission: median of 782 units/L vs. median of 236 units/L in infants with NEC
(p<0.0001). Figure 1 demonstrates individual changes in alkaline phosphatase levels by
day of illness in each infant. Most of the patients with SIP showed a rapid decline in
alkaline phosphatase levels following admission, especially during the first 3 days. Figure
2 shows overall trends in alkaline phosphatase levels where solid lines represent locally
weighted (LOESS) smoothing of the alkaline phosphatase data for each group by day of
illness, and dotted lines are the regression lines for the same data (not accounting for repeat
measurements). There were multiple data points missing for subjects in each group,
however, subsequent analysis of available data using mixed modeling with adjustment for
repeat measurements supported significant differences in the mean alkaline phosphatase
levels on admission (1,239 units/L in infants with SIP vs. 396 units/L in NEC patients,
p<0.0001) as well as the rate of decline of these levels between the two groups (estimated
mean decrease of 248 units/L per day in SIP cases vs. 67 units/L per day of illness during
the first 3 days in NEC controls, p<0.0001). Interestingly, differences in serum alkaline
phosphatase levels remained significant by day 7 of illness: a median of 223 units/L [IQR
171–318] in 44 of 47 infants with SIP that survived the first week of hospitalization
compared to a median of 117 units/L [IQR 80–152] in 45 of 46 surviving NEC patients
(p<0.0001).
Next, we investigated the relationship between alkaline phosphatase level and PMA
on admission (Figure 3). There was no significant association between admission alkaline
phosphatase levels and PMA in infants with SIP: estimated mean decrease of -43 units/L
(95%CI [-211; 125], R
2
=0.006, p=0.61) per each week of PMA. In contrast, patients with
NEC demonstrated decline in admission alkaline phosphatase levels based on PMA:
estimated mean decrease of -51 units/L (95%CI [-69; -33], R
2
=0.35, p<0.0001) per week
of PMA increase.
11
Overall, alkaline phosphatase level on admission demonstrated very good
discrimination as evident by ROC curve analysis. Area under the curve (AUC) for alkaline
phosphatase alone was 0.846, p<0.0001. It further improved after addition of GA at birth:
AUC=0.867, p<0.0001 (Figure 4). Subsequent sensitivity and specificity analysis of the
existing dataset resulted in selection of the alkaline phosphatase level of 500 units/L as the
optimal threshold to identify patients with SIP with sensitivity of 80.8% and specificity of
81.5% (GA-adjusted OR 12.7, 95%CI [4.7; 34.3], p<0.0001).
2. Demographic and perinatal factors.
Maternal and infant demographic data as well as prenatal factors are presented in
Table 1. Infants with SIP were born at significantly lower GA and lower BW compared to
NEC patients: median of 25.6 vs. 28.2 weeks (p=0.0002) and median of 785 vs. 1,111
grams (p=0.0001), respectively. In infants with SIP, a significantly larger proportion of
mothers were pregnant for the first time (39.6% vs. 15.2% in NEC controls, p=0.003), and,
overall, had lower gravidity and parity compared to mothers of infants with NEC. 38/46
(82.6%) of infants with SIP were exposed to antenatal steroids (ANS) in contrast to only
33/63 (52.4%) of NEC controls, p=0.001; for 5 infants (2 cases and 3 controls) data on
exposure to antenatal steroids were not available.
Table 2 summarizes data on intrapartum and postnatal factors, including therapeutic
exposures and feeding data. While mode of delivery was similarly distributed between the
groups, infants with SIP had lower Apgar scores at both 1 and 5 min (5 vs. 6, p=0.0005 and
7 vs. 8, p<0.0001, respectively) with 29/46 (63%) of them having 5-minute Apgar score ≤
7 compared to only 13/61 (21.3%) of NEC patients (p<0.0001). Significantly higher
proportion of infants with SIP received surfactant therapy (89.4% vs. 60.6% in controls,
p=0.0006), and were treated with indomethacin more often (62.5% vs. 26.2% in controls,
p=0.0001) with the median number of doses equal to 3 in infants with SIP and 0 in controls,
p<0.0001. 23% of SIP cases received more than one course of indomethacin compared to
only 2/65 (3%) of NEC controls, p=0.002. When combined exposure to steroids (antenatal
and postnatal) and indomethacin was analyzed, 26/46 (55.3%) of cases received both prior
to illness as compared to 10/62 (16.1%) of controls, p<0.0001. The proportion of patients
exposed to steroids only was similar between the groups (31.9% of infants with SIP vs.
12
40.3% of NEC patients), as well as proportion of patients exposed to indomethacin only
(6.4% vs. 8.1%, respectively). Only 6.4% of SIP cases were not exposed to any steroids
and/or Indomethacin, compared to 35.5% of controls (p<0.0001).
3. Feedings.
A significantly higher proportion of infants with SIP compared to NEC were never
fed prior to developing intestinal perforation: 41.7% vs. 3% of NEC patients (p<0.0001).
The SIP cases were also more likely to receive trophic feeds (52.1%) compared to controls
(10.6%). None of the cases achieved full feeds prior to admission in contrast to 49/66
(74.2%) of NEC controls, p<0.0001 (Figure 5). Figure 6 demonstrates the distribution of
infants with regards to maximal feeding amount prior to illness based on postmenstrual age
(PMA) at admission, separately for each group. Infants with SIP remained NPO or received
only trophic feeds essentially regardless of the PMA of up to 30 weeks, while the majority
of NEC patients were on full feeds at a PMA of 28 weeks and higher. Similar trends were
observed when maximal amount of feeds was analyzed against gestational age at birth for
each group (not shown).
Limited data were available on the type of feedings received by the infants prior to
illness; only 21/48 (44%) of cases and 44/66 (67%) of controls had feeding type identified
in medical records. Of these infants, a significantly larger proportion of SIP cases received
maternal (71.4%) or donor (9.5%) breast milk compared to NEC controls that received
only 25% and 4.5%, respectively (p=0.002). Only 3/21 (14.3%) of cases were given
formula, while 22/44 (50%) of controls were fed exclusively with formula prior to illness
(Figure 7).
4. Disease presentation.
Infants with SIP were both younger and smaller at the time of admission to NICCU,
parallel to their GA at birth and BW (Table 3). These infants were admitted on a median
day of life (DOL) 8 compared to DOL 14 in NEC controls (p=0.0001) with admission
weight of 800 grams vs. 1,310 grams in controls (p<0.0001). Accordingly, median PMA
of the SIP cases was only 27 weeks, while NEC controls were at 31.2 weeks of PMA on
admission (p<0.0001). More so, there was a negative linear relationship between GA and
13
DOL at admission (Figure 8) with weaker correlation in infants with SIP (r= –0.37,
p=0.009) than in NEC patients (r= –0.58, p<0.0001). Estimated mean decrease in
chronological age at admission was 1.1 day per week increase in GA at birth for cases and
1.9 day for controls. Infants with SIP also presented one day earlier since the onset of the
disease.
Only 1 patient with SIP (2.1%) presented with bloody stool, while 55% of NEC
patients had this symptom (p<0.0001). The majority of patients in both groups had
abdominal distention, and a significantly larger proportion of infants with SIP developed
bluish discoloration of the abdominal wall (60.4% vs. 39.4% in NEC patients, p=0.037).
None of the SIP cases had pneumatosis intestinalis reported on abdominal radiographs
compared to 80.3% of NEC patients having documented pneumatosis (p<0.0001), and only
1 patient with SIP (2.1%) had history of portal air in contrast to 34.9% of controls
(p<0.0001). On the other hand, the majority of the SIP cases (95.8%) had free air reported
on the abdominal radiograph at disease onset compared to 45.5% of NEC controls
(p<0.0001).
As reported previously, infants with SIP presented with significantly higher levels of
alkaline phosphatase on admission (782 vs. 236 Units/L in controls, p<0.0001). In contrast,
patients with NEC presented with significantly higher levels of CRP, lower white blood
cell counts with leukopenia and lower absolute neutrophil counts, a significant left shift
with increased immature to total neutrophils (I:T) ratio, and a trend toward
thrombocytopenia (Table 3). Liver enzymes were also significantly higher in NEC patients;
however, coagulation abnormalities were not significantly different between the two
groups. Patients with NEC were more likely to present with metabolic acidosis, primarily
lactic acidosis, although many subjects had missing values of lactate in both groups. Blood
urea nitrogen (BUN) and creatinine levels were both lower in NEC controls compared to
SIP infants.
33.3% of infants with SIP presented with hyperglycemia, not significantly different
from patients with NEC (24.2%, p=0.29). Incidence of culture-positive sepsis, both
bacterial and fungal, was also similar between the groups (18.8% vs. 30.3%, p=0.16, and
6.3% vs. 6.1%, p=1.0, respectively). None of the infants with SIP developed anuria, while
14
16.9% of controls presented with no urine output following the onset of the disease prior
to admission (p=0.002).
All but one infant with SIP that underwent exploratory laparotomy (33/34 patients,
or 97%) had intestinal perforation located in the ileum. One patient had both ileum and
colon involved. Among NEC patients, the majority of infants (59.1%) had ileum affected
by the NEC, followed by 19.7% of infants with pan-necrotizing enterocolitis (“NEC
totalis”) involving the entire intestine, 12.1% with colonic involvement, and 9.1% of
infants that had ileum and colon affected simultaneously.
5. Multivariate analysis and predictive modeling.
Significant differences in GA and BW between the two groups resulted in
confounding of the associations between the majority of independent variables and primary
outcomes. There was strong positive linear relationship between GA and BW with
correlation coefficient r =0.898 (p<0.0001). In regression analysis, estimated mean
increase in BW was 146 grams per one-week increase in GA (p<0.0001). While
intrauterine growth restriction was similarly distributed in both SIP cases and NEC
controls, GA was selected as a primary confounder for bivariate analysis as it may reflect
degree of immaturity and its effects on other variables more accurately than BW.
Results of the bivariate analysis are presented in Table 4 containing odds ratios for
development of SIP compared to NEC before and after adjustment for GA. Adjustment for
GA had effect on most associations between independent variables and SIP/NEC status as
indicated by changes in the estimated odds ratios. For parity, Apgar score at 1 minute,
surfactant therapy, bluish discoloration of the abdominal wall, and creatinine level,
association with the SIP vs NEC grouping became no longer statistically significant. In
contrast, exposure to high-frequency ventilation became significant in bivariate analysis
with odds of having SIP being one-third of the odds of presenting with NEC (OR=0.33,
p=0.02).
In multivariate logistic regression, only GA, DOL on admission, simultaneous
exposure to steroids and indomethacin, alkaline phosphatase level on admission >500
units/L, and presence of free abdominal air remained statistically significant (Table 5).
15
Increase in GA and DOL on admission decreased the odds of having SIP, while
simultaneous exposure to steroids and indomethacin, alkaline phosphatase level on
admission >500 units/L, and presence of free abdominal air were associated with higher
odds of SIP.
To develop parsimonious predictive model, only 4 of the above factors were included
in logistic regression: GA at birth ( β1 –0.328, 95%CI [–0.555, –0.101], p=0.005), DOL on
admission ( β2 –0.2, 95%CI [–0.3, –0.101], p<0.0001), combined exposure to steroids and
indomethacin ( β3 1.863, 95%CI [0.266, 3.461], p=0.022), and alkaline phosphatase level
>500 units/L ( β4 1.64, 95%CI [0.328, 2.953], p=0.014). Based on our dataset, the estimated
probability cutoff point of ≥0.486 has sensitivity of 86.9% and specificity of 90.2% in
predicting diagnosis of SIP versus NEC.
This model demonstrated overall good fit: Hosmer-Lemeshow goodness of fit test
showed very similar frequencies of observed and expected covariate patterns ( Χ
2
=7.4, df
=8, p=0.49). Comparative ROC curve analysis (Figure 9) also showed excellent
discrimination of the model with AUC = 0.93 (p<0.0001). Below is the final model:
logit [ π( Χ )] = –2.1131 + (–0.328)(GA – 27) + (–0.2)(DOL – 14) + 1.863*X3 + 1.64*X4
,
with GA = gestational age at birth in full weeks, DOL = day of life on admission, X3
= combined exposure to steroids and indomethacin (No=0, Yes=1), and X4 = alkaline
phosphatase level >500 Units/L on admission (No=0, Yes=1).
6. Other outcomes.
Table 6 summarizes survival and length of stay (LOS) data. Mortality was
significantly lower in infants with SIP: only 3(6.3%) infants died during the hospital stay
compared to 28(42.4%) infants with NEC (p<0.0001). Median LOS before death was 8
days in these patients with SIP; while median LOS before death for NEC patients that died
during the hospital stay was 2 days, the difference did not reach statistical significance
(p=0.09). 17/28 (~61%) of NEC patients died within the first 2 days following admission
(Figure 10). There was no significant association between mortality and GA or BW in
16
either group; however, mortality was the highest among infants with NEC in ELBW
category: 14 of 25 infants, or 56%.
Of 45 survivors with SIP, 23 (51.1%) were transferred back to the referring hospitals
after a median LOS of 42 days. A similar proportion of surviving infants with NEC and
their LOS were observed: 16/38 (42.1%) of infants with NEC with a median LOS of 36
days before their return transfer (p=0.41 and p=0.33, respectively). Patients that survived
to discharge home also had LOS that was not statistically significantly different between
the groups; however, further analysis showed that in infants with SIP LOS to discharge
home was strongly associated with GA: for each one week increase in GA LOS decreased
by an average of 14.2 days (95%CI [7.5; 20.9], R
2
=0.49, p=0.0003). NEC patients, in
contrast, showed no association between GA and LOS to discharge ( β=-0.34, 95%CI [-6.2;
5.6], p=0.91) or LOS before return transfer ( β=-3.4, 95%CI [-10.2; 3.4], p=0.31). Although
there was a trend toward decreasing LOS with increase in GA at birth in infants with SIP
that were transferred (Figure 11, left upper panel), regression analysis did not reveal a
statistically significant association ( β=-12.7, 95%CI [-27.7; 2.1], p=0.089).
All patients in both groups were treated with broad-spectrum antibiotics. 22.9% of
infants with SIP and 31.8% of infants with NEC were also treated with antifungal agents
(p=0.4). In SIP group, 14 patients (29.2%) underwent drain placement only, 9 patients
(18.8%) had drain placement first but required subsequent laparotomy, and 25 patients
(52.1%) had exploratory laparotomy as initial intervention. All NEC patients underwent
exploratory laparotomy, with 4 patients (6.1%) having drain placement initially. 11 patients
with NEC (16.7%) required silo placement, with only one of these 11 patients surviving
till discharge.
DISCUSSION
In this retrospective case-control study we identified that serum alkaline
phosphatase levels were significantly higher in infants with SIP compared to NEC patients.
These findings are consistent with observations in a small set of patients previously
reported (Harms et al., 1995). In 4 of 5 premature infants with SIP that had alkaline
phosphatase levels available, those levels were notably elevated, with 3 infants having
17
levels above 2,000 units/L. Alkaline phosphatase levels returned to a normal range within
1-2 days following surgery. In our study, 20 of 47 infants with SIP (43%) had alkaline
phosphatase levels above 1,000 units/L with 6 of them (13%) having levels above 2,000
units/L and a maximal level of 4,818 units/L. When compared to NEC patients, only 3 of
66 infants with NEC (5%) had alkaline phosphatase level on admission above 1,000 units/L
with a maximal level of 1,745 units/L. Our cohort of SIP patients had similar GA range
and clinical presentation but smaller BW compared to above infants. The increase in serum
alkaline phosphatase level was also transient with a rapid decline during the first three days
following intestinal perforation. The difference in alkaline phosphatase levels between the
two groups remained significant on day 7 of illness; however, its magnitude was
significantly smaller.
Alkaline phosphatase represents a group of enzymes that can be found in different
locations throughout the body. Multiple isoenzyme forms of alkaline phosphatase have
been identified, and the main function of these enzymes is catalyzing hydrolysis of a
number of organic phosphate esters in various tissues. Serum alkaline phosphatase has
three major sources, namely liver, bone, and intestine, with the first two contributing the
most to normal serum levels (Kaplan, 1972). Accordingly, the highest levels have been
observed in bone disorders with rapid bone turnover and increased osteoblastic activity
such as metabolic bone disease of prematurity, and in patients with hepatobiliary disorders,
particularly with cholestasis. In premature infants with metabolic bone disease, a gradual
increase in bone-specific alkaline phosphatase is typically observed after 4-6 weeks of age
(Abrams & Committee on Nutrition, 2013). Abdallah et al. studied serum alkaline
phosphatase as a biomarker of osteopenia of prematurity (Abdallah et al., 2016). Serial
alkaline phosphatase levels were obtained starting at 8 weeks of age. The authors reported
a strong negative correlation of alkaline phosphatase and GA as well as significantly higher
enzyme levels in infants with osteopenia compared to nonosteopenic infants. More so, 2
consecutive samples taken 1 week apart from the same patient showed an average 6.2%
increase in alkaline phosphatase level. Interestingly, they reported the optimal cutoff level
of 500 units/L to identify infants with osteopenia with 100% sensitivity and ~81%
specificity.
18
Hepatic alkaline phosphatase plays an important role in regulation of secretory
activities of intrahepatic biliary epithelium (Alvaro et al., 2000), and its levels increase
along with gamma-glytamyl transpeptidase (GGT) when extra- or intrahepatic cholestasis
develops; however, the exact timing of such increase remains unclear. In premature infants,
intrahepatic cholestasis commonly results from prolonged exposure to parenteral nutrition,
especially in infants that underwent GI surgery, particularly for NEC (Duro et al., 2011).
Early signs of parenteral nutrition-associated liver disease (PNALD) can be detected after
7 days of receiving parenteral nutrition, first as a gradual increase in direct bilirubin
followed by an increase in serum transaminases after 3 weeks (Koseesirikul,
Chotinaruemol, & Ukarapol, 2012). Ongoing exposure to parenteral nutrition results in
further increase in direct bilirubin and serum transaminases as well as alkaline phosphatase,
and resolution of these biochemical abnormalities occurs over the course of weeks to
months depending on the underlying pathology.
Intestinal alkaline phosphatase (IAP) regulates gut microbiome and promotes
growth of commensal microorganisms, participates in absorption of long chain fatty acids,
detoxifies lipopolysaccharide as well as decreases intestinal and systemic inflammation
and bacterial translocation (Fawley & Gourlay, 2016). It has been best studied in animal
models, where it was found to be expressed in the intestinal epithelial cells throughout the
intestine with the highest levels in the duodenum and terminal ileum (Shields, Bair, Bates,
Yedlin, & Alpers, 1982; Van Dongen, Kooyman, Visser, Holt, & Galjaard, 1977). In an
experimental dog model, appearance of serum IAP was detected following mesenteric
occlusion within 6 hours, primarily in animals that had ischemia but not infarction (Barnett
et al., 1976), suggesting that alterations in mucosal blood supply could result in release of
IAP into the systemic circulation. In rat pups, on the other hand, both expression and
activity of alkaline phosphatase were significantly decreased in animals with NEC
compared to controls (Whitehouse et al., 2010). When supplemental exogenous IAP was
given prior to development of NEC in this model, alkaline phosphatase activity in the
terminal ileum was very similar to controls, and these animals demonstrated significantly
lower degrees of intestinal injury than pups with NEC and no IAP supplementation. These
effects were found to be dose-dependent while decrease in NEC-related intestinal
permeability was similar regardless of the dose (Rentea et al., 2012). In addition, absence
19
of IAP has been shown to result in exacerbated intestinal inflammation that can be modified
with oral replacement of the enzyme leading to inactivation of lipopolysaccharide and
decrease or even prevention of intestinal inflammation and endotoxemia (Beumer et al.,
2003; Koyama, Matsunaga, Harada, Hokari, & Komoda, 2002; Riggle et al., 2013; van
Veen et al., 2005).
Due to the retrospective nature of our study, we did not have data on what isoform
of alkaline phosphatase was primarily responsible for the increase in total serum level of
the enzyme. However, the facts that increase in serum alkaline phosphatase in infants with
SIP was observed at a median age of 7 days with subsequent rapid decline in the next 3
days, and that this transient increase was not accompanied by increase in bilirubin or serum
transaminases, with the latter being, in fact, significantly lower in these patients, point
against metabolic bone disease or liver dysfunction as possible explanations of elevated
alkaline phosphatase levels at disease presentation. It appears much more plausible that
localized bowel ischemia with subsequent release of IAP is responsible for the transient
increase in serum alkaline phosphatase levels in patients with SIP. This hypothesis needs
to be tested and confirmed in a prospective study with serial measurements of alkaline
phosphatase in patients at risk, and, at a minimum, identification of the isoforms of the
enzyme once rapid increase in detected.
Other independent risk factors identified in our study are consistent with available
literature. Earlier GA, smaller birth weight and predominance of male gender in infants
with SIP have been documented previously: GA of 25 – 27 weeks, BW of 670 – 780 grams,
and up to 71% male infants in the cohort have been reported by other authors (Attridge,
Herman, et al., 2006; Cass et al., 2000; Drewett & Burge, 2007; Meyer et al., 1991;
Vongbhavit & Underwood, 2017; Rajan Wadhawan et al., 2013). Holland et al. reported a
higher median BW of 937 grams in infants with SIP; however, 65% of affected infants
were males and median age at diagnosis was 7 days (Holland et al., 2003), similar to our
findings. In addition, Attridge et al. compared infants with SIP with two groups of infants:
infants with NEC and control infants matched by GA, BW and gender, using large clinical
database of Pediatrix Medical Group (Attridge, Clark, Walker, & Gordon, 2006a). The
authors found that infants with SIP were less mature, had smaller BW, 65% were males
20
and median age at the onset of the disease was 7 days compared to 15 days in infants with
NEC. They also reported earlier age at onset of SIP in larger infants (Attridge, Clark,
Walker, & Gordon, 2006b).
In infants with NEC, later age at onset of the disease has been reported with
negative correlation between GA and age at onset of NEC (Uauy et al., 1991; Yee et al.,
2012). Uauy et al. reported median age at onset of NEC of 23 days in infants with GA less
than 26 weeks, and 11 days in infants with GA more than 31 weeks (Uauy et al., 1991).
These data support our findings as demonstrated in Figure 8 that, additionally, shows
greater variability in age at NEC onset in infants with GA less than 30 weeks.
When antenatal risk factors were analyzed, we found that only exposure to antenatal
steroids (ANS) was significantly higher in infants with SIP. In general, current literature
does not support an association of ANS exposure with SIP when infants with SIP are
compared to healthy controls. Attridge et al. initially reported ANS exposure rate of 58.1%
in infants with SIP vs. 65% of control infants without SIP that were matched based on
gender and BW (Attridge, Clark, Walker, et al., 2006a). However, after excluding infants
that developed SIP during the first 3 days of life, authors reported similar ANS exposure
rates: 62.9% in infants with SIP vs. 64.4% of their matched controls (Attridge, Clark, &
Gordon, 2006). Other authors reported even lower exposure rate of 44.6% in infants with
SIP that was similar to their controls (54.6%) (Rajan Wadhawan et al., 2013). Our infants
with SIP had significantly higher exposure rate to ANS of 82.6% while only 52.4% of NEC
patients were exposed to ANS, similar to exposure rates of the above-mentioned studies. It
is unclear why NEC patients received ANS less frequently, while, following adjustment
for GA, exposure to ANS showed a significant reduction in the strength of association with
SIP. More so, when exposure to ANS was analyzed together with postnatal steroids (PNS)
and postnatal indomethacin treatments, the proportion of infants exposed to ANS alone
was similar between SIP and NEC patients (see further discussion below).
Ragouilliaux et al. reported significantly higher incidence of chorioamnionitis in
infants with SIP compared to matched controls without SIP (40% vs. 12%) (Ragouilliaux,
Keeney, Hawkins, & Rowen, 2007). Incidence of chorioamnionitis in our study was similar
between the two groups, with 16.5% of infants with SIP having chorioamnionitis
21
documented in medical records, thus, significantly lower than previously reported.
Preeclampsia was recently reported as an independent risk factor for SIP (Yılmaz et al.,
2014) when compared to controls without SIP. In a multivariate analysis, these authors also
found that ANS and intrauterine growth restriction were associated with SIP. They did not
specify the proportion of infants with SIP that were exposed to preeclampsia precluding
such comparison with our patients; nevertheless, incidence of preeclampsia was similar
between the groups in our study (see Table 1).
Exposure to indomethacin during the first two weeks of life, particularly during
days of life 0–3, has been previously identified as a risk factor for development of SIP
(Attridge, Clark, Walker, et al., 2006a; Shorter, Liu, Mooney, & Harmon, 1999). Other
studies failed to detect this association. In a large multicenter randomized controlled trial
(RCT) investigating short- and long-term effects of prophylactic indomethacin in
extremely low birth weight (ELBW) infants (the TIPP trial), prophylactic use of
indomethacin was not associated with increased risk of gastrointestinal perforation
(Schmidt et al., 2001). However, gastrointestinal perforation was a secondary outcome in
this study, and this finding needs to be interpreted with caution. Kelleher et al. also did not
find increased risk of SIP in ELBW infants treated with prophylactic indomethacin during
the first 24 hours of life in a large retrospective cohort study using Eunice Kennedy Shriver
National Institute of Child Health and Human Development Neonatal Research Network
Generic Database (Kelleher et al., 2014). Instead, these authors found that initiation of
enteral feeds during the first 3 days of life was associated with a significant decrease of the
risk of SIP in these infants, whether or not they were treated with prophylactic
indomethacin. This finding indirectly supports our observation that enteral feeds were not
initiated prior to intestinal perforation in more than 40% of infants with SIP compared to
only 3% of NEC patients. On the other hand, using a similar study design and inclusion
criteria applied to the Canadian Neonatal Network database, Stavel et al. showed no effect
of early enteral feeds on odds of SIP, while prophylactic indomethacin use was associated
with significantly higher odds of SIP in ELBW infants regardless of enteral feeds (Stavel
et al., 2017).
Wadhawan et al. found that after adjustment for other risk factors medical treatment
of patent ductus arteriosus (PDA) with indomethacin but not prophylactic indomethacin
22
(used in the first 24 hours of life) was significantly associated with higher odds of SIP
(Rajan Wadhawan et al., 2013). In our study, we found that after adjustment for GA infants
with SIP were 2.7 times more likely to be exposed to indomethacin compared to infants
who were diagnosed with NEC. Indomethacin was administered at any time prior to disease
onset; however, data about the exact timing or indication for indomethacin treatment was
not available. In the majority of subjects when indication was documented in transfer
records, indomethacin was given for treatment of patent ductus arteriosus.
Several studies reported that early administration of postnatal steroids (PNS), i.e.
within the first 24 hours after birth, for prophylaxis of bronchopulmonary dysplasia is also
associated with increased risk of SIP. Gordon et al. reported higher incidence of SIP in
ELBW infants with early dexamethasone treatment compared to controls (P. Gordon,
Rutledge, Sawin, Thomas, & Woodrum, 1999). In that study, prophylactic indomethacin
alone was not associated with increased risk of SIP, but there was a trend toward increased
risk of SIP when infants were exposed to both indomethacin and dexamethasone. Paquette
et al. found that combined exposure to early dexamethasone and indomethacin was
associated with significantly higher odds of SIP after adjustment for other factors
(Paquette, Friedlich, Ramanathan, & Seri, 2006). In a multicenter RCT Stark et al.
identified significantly higher incidence of SIP during the first 14 days after birth in ELBW
infants in the dexamethasone group compared to placebo-treated controls (Stark et al.,
2001). Authors also reported higher incidence of SIP in infants exposed to prophylactic
indomethacin as well as greater effect of early dexamethasone prophylaxis (EDP) on risk
of SIP when it was combined with exposure to indomethacin prophylaxis. These findings
were subsequently included along with 3 other RCTs in a meta-analysis with a pooled
sample of 1,383 infants (P. V. Gordon, Young, & Marshall, 2001). Authors confirmed that
EDP was associated with significantly higher odds of SIP.
Exposure to postnatal steroids, particularly to dexamethasone, was relatively
infrequent in our study, and it was not significantly different between the two groups.
Hydrocortisone was the most frequently used glucocorticoid, and vasopressor-resistant
hypotension appeared to be the primary indication for steroids in our population, although
the data on exact timing or indication for such treatment was also not available.
23
As mentioned earlier, a number of studies reported combined exposure to
indomethacin and steroids, antenatal or postnatal, as an even stronger risk factor for SIP
than exposure to either of these therapeutic agents alone (P. Gordon et al., 1999; Paquette
et al., 2006; Stark et al., 2001). Watterberg et al. found that infants randomized to
hydrocortisone group and treated with indomethacin were more likely to develop
spontaneous perforation compared to placebo-treated infants that were exposed to
indomethacin alone (Watterberg et al., 2004). The authors found no difference in incidence
of SIP between hydrocortisone-treated infants without indomethacin exposure and infants
in the placebo group. Similarly, no association between indomethacin alone and SIP was
detected. Attridge et al. reported the use of indomethacin or hydrocortisone during the first
3 days of life was independently associated with SIP after adjustment for other variables
when compared to their matched controls without SIP (Attridge, Clark, Walker, et al.,
2006a; Attridge, Clark, & Gordon, 2006). Furthermore, combined exposure to both
treatments during the first 3 days of life was associated with higher incidence of SIP in
these infants. This difference was no longer significant when the treatment window with
both medications was extended to 14 days since birth. Subsequently, the authors analyzed
the effects of combined exposure to ANS and indomethacin on SIP in infants that
developed spontaneous perforation between days 4 and 14 of life and their matched
controls (Attridge, Clark, & Gordon, 2006). They found that infants with SIP had
significantly higher proportion of infants exposed to indomethacin alone or ANS and
indomethacin combined, with nearly twice as many infants in each of the two subgroups
compared to controls. When exposure to ANS with and without indomethacin treatment
was analyzed in our study, the proportions of infants exposed to ANS alone and
indomethacin alone was very similar between the groups. After accounting for missing
data, 8.7% of infants with SIP and 9.7% of infants with NEC were exposed to indomethacin
only. Accordingly, 30.4% of infants with SIP and 37.1% of infants with NEC were exposed
to ANS alone. Significant differences were observed when we compared proportions of
infants in each group with no exposure to any of these medications (8.7% in infants with
SIP vs. 38.7% of controls) and with combined exposure to both medications (52.2% vs.
14.5%, respectively). While we did not find significant differences in exposure to PNS in
our subjects, given existing evidence we combined PNS exposure data with that of ANS
24
and indomethacin. In multivariate analysis, the combined exposure to ANS and/or PNS
and indomethacin was found to be an independent risk factor for SIP.
Pneumoperitoneum as an initial radiographic finding was another strong risk factor
that was independently associated with SIP in our study on multivariate modeling. It was
present in 96% of infants with SIP, significantly more common than in NEC patients or
previously described. Overall incidence of abdominal free air as the first indicator of
serious intraabdominal disorder in infants with SIP has been reported to be up to 50-80%
(Adderson et al., 1998; Attridge, Herman, et al., 2006). In patients with NEC, free
abdominal air due to bowel perforation typically appears following other clinical and
radiographic signs of NEC at median interval of 1 day (Najaf, Vachharajani, Warner, &
Vachharajani, 2010). Gasless abdomen was significantly less frequent finding in out study,
and it was similar between the study groups unlike previously reported (Adderson et al.,
1998; Pumberger et al., 2002; Aschner, Deluga, Metlay, Emmens, & Hendricks-Munoz,
1988). None of our infants with SIP had pneumatosis intestinalis, and only one had finding
of portal air reported on abdominal radiograph, consistent with current literature.
Abdominal wall discoloration has been described as a typical finding in patients
with SIP (Ragouilliaux et al., 2007; Pumberger et al., 2002; Adderson et al., 1998). Aschner
et al. reported blue-black abdominal wall discoloration as the first presenting sign of SIP
in 5 of 6 patients as well as the lack of signs of systemic illness (Aschner et al., 1988). Only
one of their patients presented with pneumoperitoneum, while the rest of them had gasless
abdomen on abdominal radiograph. None of the patients were fed prior to development of
perforation. Shorter et al. also reported abdominal wall discoloration in addition to earlier
exposure to indomethacin in infants with SIP (Shorter et al., 1999). Indeed, nearly two-
thirds of our infants with SIP presented with this physical finding that was significantly
more common than in the NEC group; however, almost 40% of NEC patients in our study
also presented with this sign. When included in multivariate analysis, abdominal wall
discoloration was no longer significant. Abdominal distention and bilious residuals were
similarly common among infants in both groups while abdominal wall erythema
demonstrated trend toward significance in NEC patients. Another typical NEC sign, bloody
stools, was almost exclusively present in more than half of infants with NEC.
25
Among other factors that were associated with SIP following adjustment for GA
was gravidity. First pregnancy was a significant risk factor for development of SIP, with
infants with SIP being almost 3 times as likely to be born to a primigravid mother compared
to NEC infants. This risk appeared to decrease with each subsequent pregnancy by almost
30%. We were unable to find other studies reporting this risk factor.
Intrapartum factors also appear to play a role and affect the risk of developing SIP
vs. NEC as indicated by the Apgar scores (Pumberger et al., 2002). A 5-minute Apgar score
may reflect overall acuity as well as resuscitation events immediately after birth, and it was
significantly lower in our infants with SIP, even after adjustment for degree of prematurity.
Similarly, Cass et al. reported lower Apgar scores in infants with SIP in addition to younger
GA, age at disease onset and lower BW when compared to infants with NEC (Cass et al.,
2000), while other authors did not find such a difference (Attridge, Clark, Walker, et al.,
2006a).
Initiation of enteral feeds appeared to be another significant risk factor for SIP in
our study, consistent with other reports (Cass et al., 2000; Holland et al., 2003; Kelleher et
al., 2014; Meyer et al., 1991; Ragouilliaux et al., 2007). More than 90% of our infants with
SIP were either never fed (NPO status, ~42%) or received trophic feeds only (~52%) prior
to development of perforation, whereas almost 75% of NEC patients were on full feeds.
More so, the association between NPO status and SIP remained significant even after
adjustment for either GA or PMA. Furthermore, while a growing amount of evidence
supports protective effects of breast milk in lowering the risk of NEC, based on our limited
data about the type of enteral feeds given prior to illness, maternal breast milk did not
demonstrate protective properties relative to NEC in infants with SIP.
Decreased overall severity of systemic illness in infants with SIP compared to NEC
has been supported by significant differences in laboratory findings between the two
groups in our study. Infants with SIP had lower levels of liver enzymes, a marker of
hepatocellular injury, and had indicators of lactic acidosis less frequently. In contrast,
patients with NEC presented with laboratory signs of prominent inflammatory response as
demonstrated by significantly higher CRP levels, lower white blood cell counts with
leukopenia and lower absolute neutrophil counts, significant left shift with increased
26
immature to total neutrophils (I:T) ratio, and the trend toward thrombocytopenia. Other
indicators of severe systemic illness were the findings that infants with NEC had higher
odds of requiring high-frequency ventilation that, in fact, became significant after
adjustment for GA, and 17% of NEC patients presented with anuria. Interestingly, infants
with SIP had significantly higher levels of BUN and creatinine, likely as a result of overall
immaturity, fluid restriction and exposure to indomethacin, consistent with previous
reports (Novack et al., 1994).
Hyperglycemia as well as culture-positive sepsis were not significantly different
between the two groups. Kleigman et al. reported that 20-30% of infants with NEC present
with associated bacteremia (Kliegman & Fanaroff, 1984), consistent with our findings.
Novack et al. also found that 4 of 11 infants with SIP had positive cultures for either
Staphylococcus epidermidis or Candida (Novack et al., 1994).
As for the clinical outcomes, we observed significant differences in mortality and
length of stay patterns between the two groups, similar to other reports. NEC patients in
our study had higher overall mortality, with the majority of patients (17/28, or 61%)
expiring within the first two days following onset of the disease. However, we did not
identify an association of increased mortality with decreasing GA or BW in either group,
unlike previously reported (Fisher et al., 2014; Fitzgibbons et al., 2009). More so, infants
with SIP in our study had the lowest mortality presented in the literature: only 6%. Other
authors reported mortality from 10% in VLBW infants with SIP (Uceda et al., 1995) to as
high as 39% in ELBW infants (R. Wadhawan et al., 2014). Shah et al. reported mortality
of 25% in infants diagnosed with SIP compared to 50% mortality when NEC with
perforation was present in infants < 32 weeks GA (Shah et al., 2015). This is consistent
with our results of subgroup analysis that indicated that 46% of NEC patients with GA <32
weeks did not survive till discharge, slightly higher than overall mortality of 42% in all
infants with NEC. This, in turn, was similar to Bhatt et al. who reported mortality of 36-
38% in all infants with surgical NEC (Bhatt et al., 2017).
Another important indicator of the different impact of SIP compared to NEC on
short-term outcomes was the observation that length of stay (LOS) was not significantly
affected by intestinal perforation in infants with SIP, and it was primarily dependent on
27
GA with longer hospital stays in infants of smaller GA at birth. This association was most
evident in infants with SIP that were discharged home: each week of increase in GA was
associated with, on average, 2 week decrease in LOS. In contrast, LOS of NEC patients
was independent of their GA, and was on average approximately 50 days in infants with
GA ranging from 24 to 35 weeks that were transferred back, and nearly twice as long in
infants discharged home.
Our study has many limitations related to the retrospective design including
accuracy of documentation in medical records, inconsistency in definitions and timing of
certain diagnostic and therapeutic interventions, and missing data. For example, lack of
consistent data on timing and indications for indomethacin therapy in medical records
prevents further analysis of early vs. late exposure to indomethacin and risk of SIP vs. NEC
and comparing them to available literature. Similar limitations are applicable to exposure
to postnatal steroids; however, we found combined exposure to both antenatal and/or
postnatal steroids and indomethacin was a strong risk factor for SIP, consistent with other
reports. More importantly, there were no alkaline phosphatase levels prior to development
of SIP or NEC available as it is not routinely checked in newborn infants during the first
weeks of life, nor did we have data about isoforms of alkaline phosphatase. Nevertheless,
significant group differences in alkaline phosphatase levels immediately following disease
onset between the infants of the same PMA as well as rapid decrease in alkaline
phosphatase levels in infants with SIP support our hypothesis.
Lack of GA- or BW-matched control infants that did not have SIP or NEC is
another limitation of the study, particularly with regards to serum alkaline phosphatase
changes. However, exposure rates to many other risk factors in our infants were very
similar to other studies that reported such exposure rates when infants with SIP or NEC
were compared to their controls. Finally, the relatively small sample size limited statistical
power of our study and, thus, the ability to detect statistically significant associations when
multiple independent variables were included in statistical analysis. To overcome all these
limitations, large multicenter prospective studies that would include serial measurements
of serum alkaline phosphatase levels and identification of its isoforms when elevated is
necessary.
28
In conclusion, this is the first study reporting differences in alkaline phosphatase
levels between patients with SIP compared to NEC patients in conjunction with other risk
factors. Patients with SIP had significantly higher levels of alkaline phosphatase
immediately following intestinal perforation, even after adjustment for other established
factors. These abnormally elevated alkaline phosphatase levels decreased rapidly during
the following 3 days, however, they remained significantly higher compared to the ones in
NEC patients even by day 7 of illness. When combined with infant’s gestational age at
birth and age at disease onset as well as previous exposure to both ante- and/or postnatal
steroids and indomethacin, alkaline phosphatase level at the onset of the disease may be
used as a differentiating factor to assist in establishing correct diagnosis in an infant
presenting with bowel perforation. This may have important management and prognostic
implications in clinical settings.
REFERENCES
Abdallah, E. A. A., Said, R. N., Mosallam, D. S., Moawad, E. M. I., Kamal, N. M., & Fathallah, M. G.
E.-D. (2016). Serial serum alkaline phosphatase as an early biomarker for osteopenia of
prematurity. Medicine, 95(37), e4837. https://doi.org/10.1097/MD.0000000000004837
Abrams, S. A., & Committee on Nutrition. (2013). Calcium and vitamin d requirements of
enterally fed preterm infants. Pediatrics, 131(5), e1676-1683.
https://doi.org/10.1542/peds.2013-0420
Adderson, E. E., Pappin, A., & Pavia, A. T. (1998). Spontaneous intestinal perforation in
premature infants: a distinct clinical entity associated with systemic candidiasis. Journal
of Pediatric Surgery, 33(10), 1463–1467.
Adesanya, O. A., O’Shea, T. M., Turner, C. S., Amoroso, R. M., Morgan, T. M., & Aschner, J. L.
(2005). Intestinal perforation in very low birth weight infants: growth and
neurodevelopment at 1 year of age. Journal of Perinatology: Official Journal of the
California Perinatal Association, 25(9), 583–589. https://doi.org/10.1038/sj.jp.7211360
29
Alpan, G., Eyal, F., Vinograd, I., Udassin, R., Amir, G., Mogle, P., & Glick, B. (1985). Localized
intestinal perforations after enteral administration of indomethacin in premature
infants. The Journal of Pediatrics, 106(2), 277–281.
Alvaro, D., Benedetti, A., Marucci, L., Delle Monache, M., Monterubbianesi, R., Di Cosimo, E., …
Alpini, G. (2000). The function of alkaline phosphatase in the liver: regulation of
intrahepatic biliary epithelium secretory activities in the rat. Hepatology (Baltimore,
Md.), 32(2), 174–184. https://doi.org/10.1053/jhep.2000.9078
Aschner, J. L., Deluga, K. S., Metlay, L. A., Emmens, R. W., & Hendricks-Munoz, K. D. (1988).
Spontaneous focal gastrointestinal perforation in very low birth weight infants. The
Journal of Pediatrics, 113(2), 364–367.
Attridge, J. T., Clark, R., & Gordon, P. V. (2006). New insights into spontaneous intestinal
perforation using a national data set (3): antenatal steroids have no adverse association
with spontaneous intestinal perforation. Journal of Perinatology: Official Journal of the
California Perinatal Association, 26(11), 667–670. https://doi.org/10.1038/sj.jp.7211589
Attridge, J. T., Clark, R., Walker, M. W., & Gordon, P. V. (2006a). New insights into spontaneous
intestinal perforation using a national data set: (1) SIP is associated with early
indomethacin exposure. Journal of Perinatology: Official Journal of the California
Perinatal Association, 26(2), 93–99. https://doi.org/10.1038/sj.jp.7211429
Attridge, J. T., Clark, R., Walker, M. W., & Gordon, P. V. (2006b). New insights into spontaneous
intestinal perforation using a national data set: (2) two populations of patients with
perforations. Journal of Perinatology: Official Journal of the California Perinatal
Association, 26(3), 185–188. https://doi.org/10.1038/sj.jp.7211439
Attridge, J. T., Herman, A. C., Gurka, M. J., Griffin, M. P., McGahren, E. D., & Gordon, P. V. (2006).
Discharge outcomes of extremely low birth weight infants with spontaneous intestinal
30
perforations. Journal of Perinatology: Official Journal of the California Perinatal
Association, 26(1), 49–54. https://doi.org/10.1038/sj.jp.7211407
Ballance, W. A., Dahms, B. B., Shenker, N., & Kliegman, R. M. (1990). Pathology of neonatal
necrotizing enterocolitis: a ten-year experience. The Journal of Pediatrics, 117(1 Pt 2),
S6-13.
Barnett, S. M., Davidson, E. D., & Bradley, E. L. (1976). Intestinal alkaline phosphatase and base
deficit in mesenteric occlusion. The Journal of Surgical Research, 20(3), 243–246.
Beumer, C., Wulferink, M., Raaben, W., Fiechter, D., Brands, R., & Seinen, W. (2003). Calf
intestinal alkaline phosphatase, a novel therapeutic drug for lipopolysaccharide (LPS)-
mediated diseases, attenuates LPS toxicity in mice and piglets. The Journal of
Pharmacology and Experimental Therapeutics, 307(2), 737–744.
https://doi.org/10.1124/jpet.103.056606
Bhatt, D., Travers, C., Patel, R. M., Shinnick, J., Arps, K., Keene, S., & Raval, M. V. (2017).
Predicting Mortality or Intestinal Failure in Infants with Surgical Necrotizing
Enterocolitis. The Journal of Pediatrics, 191, 22-27.e3.
https://doi.org/10.1016/j.jpeds.2017.08.046
Blakely, M. L., Tyson, J. E., Lally, K. P., McDonald, S., Stoll, B. J., Stevenson, D. K., … NICHD
Neonatal Research Network. (2006). Laparotomy versus peritoneal drainage for
necrotizing enterocolitis or isolated intestinal perforation in extremely low birth weight
infants: outcomes through 18 months adjusted age. Pediatrics, 117(4), e680-687.
https://doi.org/10.1542/peds.2005-1273
Cass, D. L., Brandt, M. L., Patel, D. L., Nuchtern, J. G., Minifee, P. K., & Wesson, D. E. (2000).
Peritoneal drainage as definitive treatment for neonates with isolated intestinal
31
perforation. Journal of Pediatric Surgery, 35(11), 1531–1536.
https://doi.org/10.1053/jpsu.2000.18299
Drewett, M. S., & Burge, D. M. (2007). Recurrent neonatal gastro-intestinal problems after
spontaneous intestinal perforation. Pediatric Surgery International, 23(11), 1081–1084.
https://doi.org/10.1007/s00383-007-1999-2
Duro, D., Mitchell, P. D., Kalish, L. A., Martin, C., McCarthy, M., Jaksic, T., … Duggan, C. (2011).
Risk factors for parenteral nutrition–associated liver disease following surgical therapy
for necrotizing enterocolitis: A Glaser Pediatric Research Network Study [corrected].
Journal of Pediatric Gastroenterology and Nutrition, 52(5), 595–600.
https://doi.org/10.1097/MPG.0b013e31820e8396
Fawley, J., & Gourlay, D. M. (2016). Intestinal alkaline phosphatase: a summary of its role in
clinical disease. The Journal of Surgical Research, 202(1), 225–234.
https://doi.org/10.1016/j.jss.2015.12.008
Fisher, J. G., Jones, B. A., Gutierrez, I. M., Hull, M. A., Kang, K. H., Kenny, M., … Jaksic, T. (2014).
Mortality associated with laparotomy-confirmed neonatal spontaneous intestinal
perforation: a prospective 5-year multicenter analysis. Journal of Pediatric Surgery,
49(8), 1215–1219. https://doi.org/10.1016/j.jpedsurg.2013.11.051
Fitzgibbons, S. C., Ching, Y., Yu, D., Carpenter, J., Kenny, M., Weldon, C., … Jaksic, T. (2009).
Mortality of necrotizing enterocolitis expressed by birth weight categories. Journal of
Pediatric Surgery, 44(6), 1072–1075; discussion 1075-1076.
https://doi.org/10.1016/j.jpedsurg.2009.02.013
Gordon, P., Rutledge, J., Sawin, R., Thomas, S., & Woodrum, D. (1999). Early postnatal
dexamethasone increases the risk of focal small bowel perforation in extremely low
32
birth weight infants. Journal of Perinatology: Official Journal of the California Perinatal
Association, 19(8 Pt 1), 573–577.
Gordon, P. V., Young, M. L., & Marshall, D. D. (2001). Focal small bowel perforation: an adverse
effect of early postnatal dexamethasone therapy in extremely low birth weight infants.
Journal of Perinatology: Official Journal of the California Perinatal Association, 21(3),
156–160. https://doi.org/10.1038/sj.jp.7200520
Harms, K., Lüdtke, F. E., Lepsien, G., & Speer, C. P. (1995). Idiopathic intestinal perforations in
premature infants without evidence of necrotizing enterocolitis. European Journal of
Pediatric Surgery: Official Journal of Austrian Association of Pediatric Surgery ... [et Al] =
Zeitschrift Für Kinderchirurgie, 5(1), 30–33. https://doi.org/10.1055/s-2008-1066158
Holland, A. J. A., Shun, A., Martin, H. C. O., Cooke-Yarborough, C., & Holland, J. (2003). Small
bowel perforation in the premature neonate: congenital or acquired? Pediatric Surgery
International, 19(6), 489–494. https://doi.org/10.1007/s00383-003-0967-8
Kaplan, M. M. (1972). Alkaline phosphatase. The New England Journal of Medicine, 286(4), 200–
202. https://doi.org/10.1056/NEJM197201272860407
Kawase, Y., Ishii, T., Arai, H., & Uga, N. (2006). Gastrointestinal perforation in very low-
birthweight infants. Pediatrics International: Official Journal of the Japan Pediatric
Society, 48(6), 599–603. https://doi.org/10.1111/j.1442-200X.2006.02282.x
Kelleher, J., Salas, A. A., Bhat, R., Ambalavanan, N., Saha, S., Stoll, B. J., … GDB Subcommittee,
Eunice Kennedy Shriver National Institute of Child Health and Human Development
Neonatal Research Network. (2014). Prophylactic indomethacin and intestinal
perforation in extremely low birth weight infants. Pediatrics, 134(5), e1369-1377.
https://doi.org/10.1542/peds.2014-0183
33
Kliegman, R. M., & Fanaroff, A. A. (1984). Necrotizing enterocolitis. The New England Journal of
Medicine, 310(17), 1093–1103. https://doi.org/10.1056/NEJM198404263101707
Koseesirikul, P., Chotinaruemol, S., & Ukarapol, N. (2012). Incidence and risk factors of
parenteral nutrition-associated liver disease in newborn infants. Pediatrics International:
Official Journal of the Japan Pediatric Society, 54(3), 434–436.
https://doi.org/10.1111/j.1442-200X.2012.03627.x
Koyama, I., Matsunaga, T., Harada, T., Hokari, S., & Komoda, T. (2002). Alkaline phosphatases
reduce toxicity of lipopolysaccharides in vivo and in vitro through dephosphorylation.
Clinical Biochemistry, 35(6), 455–461.
Meyer, C. L., Payne, N. R., & Roback, S. A. (1991). Spontaneous, isolated intestinal perforations
in neonates with birth weight less than 1,000 g not associated with necrotizing
enterocolitis. Journal of Pediatric Surgery, 26(6), 714–717.
Najaf, T. A., Vachharajani, N. A., Warner, B. W., & Vachharajani, A. J. (2010). Interval between
clinical presentation of necrotizing enterocolitis and bowel perforation in neonates.
Pediatric Surgery International, 26(6), 607–609. https://doi.org/10.1007/s00383-010-
2597-2
Neu, J. (1996). Necrotizing enterocolitis: the search for a unifying pathogenic theory leading to
prevention. Pediatric Clinics of North America, 43(2), 409–432.
Novack, C. M., Waffarn, F., Sills, J. H., Pousti, T. J., Warden, M. J., & Cunningham, M. D. (1994).
Focal intestinal perforation in the extremely-low-birth-weight infant. Journal of
Perinatology: Official Journal of the California Perinatal Association, 14(6), 450–453.
Paquette, L., Friedlich, P., Ramanathan, R., & Seri, I. (2006). Concurrent use of indomethacin and
dexamethasone increases the risk of spontaneous intestinal perforation in very low
34
birth weight neonates. Journal of Perinatology: Official Journal of the California Perinatal
Association, 26(8), 486–492. https://doi.org/10.1038/sj.jp.7211548
Pumberger, W., Mayr, M., Kohlhauser, C., & Weninger, M. (2002). Spontaneous localized
intestinal perforation in very-low-birth-weight infants: a distinct clinical entity different
from necrotizing enterocolitis. Journal of the American College of Surgeons, 195(6), 796–
803.
Ragouilliaux, C. J., Keeney, S. E., Hawkins, H. K., & Rowen, J. L. (2007). Maternal factors in
extremely low birth weight infants who develop spontaneous intestinal perforation.
Pediatrics, 120(6), e1458-1464. https://doi.org/10.1542/peds.2006-2804
Rentea, R. M., Liedel, J. L., Welak, S. R., Cassidy, L. D., Mayer, A. N., Pritchard, K. A., … Gourlay, D.
M. (2012). Intestinal alkaline phosphatase administration in newborns is protective of
gut barrier function in a neonatal necrotizing enterocolitis rat model. Journal of Pediatric
Surgery, 47(6), 1135–1142. https://doi.org/10.1016/j.jpedsurg.2012.03.018
Riggle, K. M., Rentea, R. M., Welak, S. R., Pritchard, K. A., Oldham, K. T., & Gourlay, D. M. (2013).
Intestinal alkaline phosphatase prevents the systemic inflammatory response associated
with necrotizing enterocolitis. The Journal of Surgical Research, 180(1), 21–26.
https://doi.org/10.1016/j.jss.2012.10.042
Schmidt, B., Davis, P., Moddemann, D., Ohlsson, A., Roberts, R. S., Saigal, S., … Trial of
Indomethacin Prophylaxis in Preterms Investigators. (2001). Long-term effects of
indomethacin prophylaxis in extremely-low-birth-weight infants. The New England
Journal of Medicine, 344(26), 1966–1972.
https://doi.org/10.1056/NEJM200106283442602
Shah, J., Singhal, N., da Silva, O., Rouvinez-Bouali, N., Seshia, M., Lee, S. K., … Canadian Neonatal
Network. (2015). Intestinal perforation in very preterm neonates: risk factors and
35
outcomes. Journal of Perinatology: Official Journal of the California Perinatal
Association, 35(8), 595–600. https://doi.org/10.1038/jp.2015.41
Shields, H. M., Bair, F. A., Bates, M. L., Yedlin, S. T., & Alpers, D. H. (1982). Localization of
immunoreactive alkaline phosphatase in the rat small intestine at the light microscopic
level by immunocytochemistry. Gastroenterology, 82(1), 39–45.
Shorter, N. A., Liu, J. Y., Mooney, D. P., & Harmon, B. J. (1999). Indomethacin-associated bowel
perforations: a study of possible risk factors. Journal of Pediatric Surgery, 34(3), 442–
444.
Snyder, C. L., Gittes, G. K., Murphy, J. P., Sharp, R. J., Ashcraft, K. W., & Amoury, R. A. (1997).
Survival after necrotizing enterocolitis in infants weighing less than 1,000 g: 25 years’
experience at a single institution. Journal of Pediatric Surgery, 32(3), 434–437.
Stark, A. R., Carlo, W. A., Tyson, J. E., Papile, L. A., Wright, L. L., Shankaran, S., … National
Institute of Child Health and Human Development Neonatal Research Network. (2001).
Adverse effects of early dexamethasone treatment in extremely-low-birth-weight
infants. National Institute of Child Health and Human Development Neonatal Research
Network. The New England Journal of Medicine, 344(2), 95–101.
https://doi.org/10.1056/NEJM200101113440203
Stavel, M., Wong, J., Cieslak, Z., Sherlock, R., Claveau, M., & Shah, P. S. (2017). Effect of
prophylactic indomethacin administration and early feeding on spontaneous intestinal
perforation in extremely low-birth-weight infants. Journal of Perinatology: Official
Journal of the California Perinatal Association, 37(2), 188–193.
https://doi.org/10.1038/jp.2016.196
Uauy, R. D., Fanaroff, A. A., Korones, S. B., Phillips, E. A., Phillips, J. B., & Wright, L. L. (1991).
Necrotizing enterocolitis in very low birth weight infants: biodemographic and clinical
36
correlates. National Institute of Child Health and Human Development Neonatal
Research Network. The Journal of Pediatrics, 119(4), 630–638.
Uceda, J. E., Laos, C. A., Kolni, H. W., & Klein, A. M. (1995). Intestinal perforations in infants with
a very low birth weight: a disease of increasing survival? Journal of Pediatric Surgery,
30(9), 1314–1316.
Van Dongen, J. M., Kooyman, J., Visser, W. J., Holt, S. J., & Galjaard, H. (1977). The effect of
increased crypt cell proliferation on the activity and subcellular localization of esterases
and alkaline phosphatase in the rat small intestine. The Histochemical Journal, 9(1), 61–
75.
van Veen, S. Q., van Vliet, A. K., Wulferink, M., Brands, R., Boermeester, M. A., & van Gulik, T. M.
(2005). Bovine intestinal alkaline phosphatase attenuates the inflammatory response in
secondary peritonitis in mice. Infection and Immunity, 73(7), 4309–4314.
https://doi.org/10.1128/IAI.73.7.4309-4314.2005
Vongbhavit, K., & Underwood, M. A. (2017). Intestinal perforation in the premature infant.
Journal of Neonatal-Perinatal Medicine, 10(3), 281–289. https://doi.org/10.3233/NPM-
16148
Wadhawan, R., Oh, W., Hintz, S. R., Blakely, M. L., Das, A., Bell, E. F., … NICHD Neonatal Research
Network. (2014). Neurodevelopmental outcomes of extremely low birth weight infants
with spontaneous intestinal perforation or surgical necrotizing enterocolitis. Journal of
Perinatology: Official Journal of the California Perinatal Association, 34(1), 64–70.
https://doi.org/10.1038/jp.2013.128
Wadhawan, Rajan, Oh, W., Vohr, B. R., Saha, S., Das, A., Bell, E. F., … Higgins, R. (2013).
Spontaneous intestinal perforation in extremely low birth weight infants: association
with indometacin therapy and effects on neurodevelopmental outcomes at 18-22
37
months corrected age. Archives of Disease in Childhood. Fetal and Neonatal Edition,
98(2), F127-132. https://doi.org/10.1136/archdischild-2011-300659
Watterberg, K. L., Gerdes, J. S., Cole, C. H., Aucott, S. W., Thilo, E. H., Mammel, M. C., … Shaffer,
M. L. (2004). Prophylaxis of early adrenal insufficiency to prevent bronchopulmonary
dysplasia: a multicenter trial. Pediatrics, 114(6), 1649–1657.
https://doi.org/10.1542/peds.2004-1159
Whitehouse, J. S., Riggle, K. M., Purpi, D. P., Mayer, A. N., Pritchard, K. A., Oldham, K. T., &
Gourlay, D. M. (2010). The protective role of intestinal alkaline phosphatase in
necrotizing enterocolitis. The Journal of Surgical Research, 163(1), 79–85.
https://doi.org/10.1016/j.jss.2010.04.048
Yee, W. H., Soraisham, A. S., Shah, V. S., Aziz, K., Yoon, W., Lee, S. K., & Canadian Neonatal
Network. (2012). Incidence and timing of presentation of necrotizing enterocolitis in
preterm infants. Pediatrics, 129(2), e298-304. https://doi.org/10.1542/peds.2011-2022
Yılmaz, Y., Kutman, H. G. K., Ulu, H. Ö., Canpolat, F. E., Uraş, N., Oğuz, S. S., & Dilmen, U. (2014).
Preeclampsia is an independent risk factor for spontaneous intestinal perforation in very
preterm infants. The Journal of Maternal-Fetal & Neonatal Medicine: The Official Journal
of the European Association of Perinatal Medicine, the Federation of Asia and Oceania
Perinatal Societies, the International Society of Perinatal Obstetricians, 27(12), 1248–
1251. https://doi.org/10.3109/14767058.2013.851189
Zamir, O., Goldberg, M., Udassin, R., Peleg, O., Nissan, S., & Eyal, F. (1988). Idiopathic
gastrointestinal perforation in the neonate. Journal of Pediatric Surgery, 23(4), 335–337.
38
Table 1. Patient Demographics and Prenatal Factors.
* – Missing values present
δ
– Continuous variables reported as Median [interquartile range (IQR)], unless specified
otherwise
§
– PPROM = preterm premature rupture of membranes

Parameter
SIP (cases)
n = 48
NEC (controls)
n = 66
p-value
Demographic characteristics:
Birth weight, g (n=48 / 65)* 785 [640-930] 1,111 [790-1,758] 0.0001
Gestational age, weeks 25.6 [24.9-27.0] 28.2 [25.6-31.1] 0.0002
Males, n (%) 30 (62.5) 38 (57.6) 0.60
Race, n (%)
White
Black
Hispanic
Asian
Other

5 (10.4)
6 (12.5)
32 (66.7)
2 (4.2)
3 (6.2)

12 (18.2)
7 (10.6)
40 (60.6)
7 (10.6)
0

0.16
Maternal age, years (mean ± SD)
(n=47 / 63)*
28.3 ± 6.5 29.7 ± 7.0 0.27
Gravidity (n=47 / 66)*
Primigravida, n (%)*
Parity (n=47/ 66)*
2 [1-3]
19 (39.6)
2 [1-3]
3 [2-5]
10(15.2)
2 [2-4]
0.002
0.003
0.009
Prenatal factors:
Prenatal care, n (%) (n=48 / 60)* 48 (100) 55 (91.7) 0.06
Substance abuse, n (%) (n=47 / 66)* 2 (4.3) 5 (7.6) 0.70
Intrauterine growth restriction, n (%) 3 (6.3) 9 (13.6) 0.24
Oligohydramnios, n (%) (n=48 / 63)* 4 (8.3) 2 (3.2) 0.40
Preeclampsia, n (%) (n=46 / 55)* 5 (10.9) 4 (7.3) 0.73
Eclampsia, n (%) (n=46 / 55)* 2 (4.4) 2 (3.6) 1.00
Pregnancy-induced hypertension, n (%)
(n=46 / 55)*
5 (10.9) 8 (14.6) 0.77
PPROM
§
, n (%) (n=48 / 64)* 16 (33.3) 16 (25.0) 0.33
Chorioamnionitis, n (%) (n=48 / 63)* 8 (16.7) 5 (7.9) 0.23
Placental abruption, n (%) (n=48 / 65)* 6 (12.5) 8 (12.3) 0.98
Antenatal steroids, n (%) (n=46 / 63)* 38 (82.6) 33 (52.4) 0.001
Magnesium, n (%) (n=48 / 63)* 20 (41.7) 20 (31.8) 0.32
39
Table 2. Intrapartum and Postnatal factors.
Parameter
SIP (cases)
n = 48
NEC (controls)
n = 66
p-value
Intrapartum factors:
C-section, n (%) 35 (72.9) 39 (59.1) 0.13
Apgar score (n=46 / 61)*:
1 min
5 min
≤7 at 5 min, n (%)

5 [2-6]
7 [5-8]
29 (63.0)

6 [5-8]
8 [8-9]
13 (21.3)

0.0005
<0.0001
<0.0001
Therapeutic factors:
Mechanical ventilation, n (%):
None
CMV
§

HFOV/HFJV
§

3 (6.3)
33 (68.8)
12 (25.0)

6 (9.1)
38 (57.6)
22 (33.3)

0.50
Oxygen therapy, n (%) 46 (95.8) 65 (98.5) 0.57
Surfactant, n (%) (n=47/ 66)* 42 (89.4) 40 (60.6) 0.0006
Steroids, n (%)
Dexamethasone, n (%)
Hydrocortisone, n (%)
13 (27)
3 (6.3)
12 (25.0)
8 (12)
1 (1.5)
8 (12.1)
0.052
0.31
0.086
NSAIDs
§
, n (%) (n=48 / 65)*
Indomethacin, n (%) (n=48 / 65)*
Multiple courses, n (%)*
Number of doses (median [IQR])*
Ibuprofen, n (%)
35 (73)
30 (62.5)
11 (23)
3 [0-3]
5 (10.4)
25 (38)
17 (26.2)
2(3)
0 [0-1]
9 (13.6)
0.0003
0.0001
0.002
<0.0001
0.77
Steroids + indomethacin, n (%) (n=47 /
62)*
26 (55.3) 10 (16.1) <0.0001
Vasopressor-inotropes, n (%) 33 (68.8) 40 (60.6) 0.43
Feeds:
Feeding amount:
No feeds (NPO) since birth, n (%)
Trophic feeds, n (%)
Partial feeds, n (%)
Full feeds, n (%)
20 (41.7)
25 (52.1)
3 (6.2)
0
2 (3.0)
7 (10.6)
8 (12.1)
49 (74.3)

<0.0001

Feeding type (n = 21 / 44)*:
Maternal breastmilk, n (%)
Donor breastmilk, n (%)
Formula, n (%)
Breastmilk + formula, n (%)
Pedialyte, n (%)

15 (71.4)
2 (9.5)
3 (14.3)
1 (4.8)
0

11 (25)
2 (4.5)
22 (50)
8 (18.2)
1 (2.3)

0.002

40

* – Missing values present
δ
– Continuous variables reported as Median [IQR], unless specified otherwise
§
– CMV = conventional mechanical ventilation,

HFOV = high-frequency oscillatory ventilation,
HFJV = high-frequency jet ventilation, NSAIDs = nonsteroidal anti-inflammatory drugs

41
Table 3. Disease Presentation and Characteristics.

Parameter
SIP (cases)
n = 48
NEC (controls)
n = 66
p-value
DOL on admission 8 [6-10] 14 [8-23] 0.0001
Day of illness on admission 1 [1-2] 2 [1-3] 0.009
PMA on admission 27 [26-28.4] 31.2 [29.4-33.6] <0.0001
Admission weight, g (n=48/ 65)* 800 [685-938]
1,310
[1,000-1,800]
<0.0001
Clinical characteristics:
Bilious residuals, n (%) 16 (33.3) 12 (18.2) 0.079
Bloody stools, n (%) 1 (2.1) 36 (54.6) <0.0001
Abdominal distention, n (%)
(n=47 / 66)*
44/47 (93.6) 63 (95.5) 0.69
Abdominal wall discoloration, n (%) 29 (60.4) 26 (39.4) 0.037
Abdominal wall erythema, n (%) 3 (6.3) 13 (19.7) 0.056
Radiographic findings:
Gasless abdomen, n (%) 5 (10.4) 2(3.0) 0.13
Pneumatosis, n (%) 0 53 (80.3) <0.0001
Portal air, n (%) 1 (2.1) 23 (34.9) <0.0001
Free air, n (%) 46 (95.8) 30 (45.5) <0.0001
Laboratory findings:
Alkaline Phosphatase, Units/L (n=47 /
65)*
782 [575-1283] 236 [169-399] <0.0001
Alkaline Phosphatase >400 Units/L, n
(%)
38 (80.9) 20 (30.8) <0.0001
Alkaline Phosphatase >500 Units/L, n
(%)
38 (80.9) 12 (18.5) <0.0001
ALT, Units/L 23 [18-27] 34 [24-53] <0.0001
AST, Units/L 31 [24-49] 87 [46-150] <0.0001
Bilirubin (total), mg/dL 4.3 [3.6-6.3] 4.6 [2.6-7.5] 0.76
Direct bilirubin, mg/dL (n=16/ 15)* 0.65 [0.25-1.2] 0.6 [0-2.9] 0.78
WBC, 10
3
/microL 16.3 [10.2-25.2] 4.9 [2.8-8.8] <0.0001
ANC, 1/microL
9,229
[5,364-18,016]
1,543 [610-3,840] <0.0001
Severe neutropenia (<500/ microL) 1 (2.1) 15 (22.7) 0.002
Hct, % (mean ± SD) 35.1 ± 6.1 36.4 ± 6.3 0.29
CRP, mg/dL (n=41/ 57)* 2.8[0.9-5.2] 9.0 [3.2-18.8] <0.0001
42

Table 3. Disease Presentation and Characteristics, cont’d.
* – Missing values present
δ
– Continuous variables reported as Median [IQR], unless specified otherwise

Parameter
SIP (cases)
n = 48
NEC (controls)
n = 66
p-value
Laboratory findings ( co n t’ d ):
I:T ratio 0.19 [0.09-0.32] 0.56 [0.32-0.78] <0.0001
I:T ratio > 0.2, n (%) 22 (45.8) 57 (86.4) <0.0001
Thrombocytopenia (<10
5
/microL), n (%) 14 (29.2) 32 (48.5) 0.053
Coagulopathy, n (%) (n=44/ 66)* 32 (72.7) 51 (77.3) 0.59
pH, ( mean ± SD) 7.3 ± 0.1 7.25 ± 0.15 0.047
Acidosis (pH<7.25, BE <-4.0), n (%) 10 (20.8) 23 (34.9) 0.103
Lactate, mg/dL (n=27 / 40)* 11 [7.9-19.3] 24.8 [12.2-68.4] 0.004
Lactate >20 mg/dL, n (%)* 6 (22.2) 22 (55) 0.008
BUN, mg/dL 41 [30-55] 24 [14-34] <0.0001
Creatinine, mg/dL 0.8 [0.7-1.0] 0.6 [0.5-0.9] 0.0003
Creatinine > 1mg/dL, n (%) 11 (22.9) 7 (10.6) 0.075
BUN/ creatinine ratio 46 [34-63] 35 [23-56] 0.012
BUN/ creatinine ratio > 20, n (%) 45 (93.8) 53 (80.3) 0.056
Comorbidities:
Hyperglycemia (>150 mg/dL), n (%) 16 (33.3) 16 (24.2) 0.29
Treated with insulin, n (%) 4 (25.0) 8 (50.0) 0.27
Anuria, n (%) (n=47/ 66)* 0 11 (16.9) 0.002
Culture-positive sepsis
Bacterial, n (%) 9 (18.8) 20 (30.3) 0.16
Candidal, n (%) 3 (6.3) 4 (6.1) 1.0
Affected bowel (n=34/ 66)*:
Ileum, n (%) 33 (97.0) 39 (59.1)
Colon, n (%) 0 8 (12.1)
Ileum + colon, n (%) 1 (3.0) 6 (9.1)
Pan-intestine, n (%) 0 13 (19.7)
43
Table 4. Bivariate analysis (adjusted for GA).

Parameter OR crude (95%CI) p-value OR adj (95%CI) p-value
Demographic characteristics:
Birth weight*, 100 g
0.80
(0.71; 0.9)
0.0001
0.83
(0.67; 1.02)
0.079
Gestational age
0.73
(0.63; 0.86)
0.0001 N/A
Gravidity*
0.7
(0.54; 0.89)
0.0037
0.72
(0.56; 0.94)
0.014
Primigravida*
3.80
(1.56; 9.25)
0.0041
2.73
(1.05; 7.06)
0.039
Parity*
0.72
(0.54; 0.95)
0.02
0.76
(0.57; 1.03)
0.074
Prenatal and intrapartum factors:
Antenatal steroids*
4.32
(1.74; 10.71)
0.0016
3.25
(1.23; 8.57)
0.017
Apgar score*
1 min
5 min
≤7 at 5 min
0.72 (0.6; 0.86)
0.54 (0.4; 0.74)
6.3 (2.7; 14.8)
0.0004
<0.0001
<0.0001
0.83 (0.68; 1.01)
0.66 (0.49; 0.88)
3.28 (1.27; 8.47)
0.056
0.005
0.014
Postnatal factors:
Mechanical ventilation
(HFOV/HFJV)
0.67
(0.29; 1.53)
0.41
0.33
(0.13; 0.85)
0.02
Surfactant
5.46
(1.91; 15.6)
0.0006
1.76
(0.48; 6.4)
0.39
Indomethacin*
4.7
(2.1; 10.5)
0.0001
2.65
(1.1; 6.37)
0.029
Number of doses, dose*

1.49
(1.2; 1.85)
0.0003
1.3
(1.04; 1.63)
0.021
Steroids + Indomethacin*
6.4
(2.6; 15.6)
<0.0001
3.7
(1.4; 9.7)
0.01
NPO
22.9
(5.0; 104.5)
<0.0001
19.0
(4.03; 89.6)
0.0002
Disease presentation:
DOL on admission, day
0.9
(0.85; 0.95)
0.0003
0.82
(0.76; 0.89)
<0.0001
PMA on admission, week

0.55
(0.44; 0.68)
<0.0001 N/A
Admission weight*, 100g

0.70
(0.6; 0.81)
<0.0001 N/A
44
Table 4. Bivariate analysis (adjusted for GA), cont’d.
Parameter OR crude (95%CI) p-value OR adj (95%CI) p-value
Clinical characteristics:
Bloody stools
0.02
(0.002; 0.14)
<0.0001
0.03
(0.003; 0.22)
<0.001
Abdominal wall discoloration
2.35
(1.1; 5.02)
0.037
1.82
(0.8; 4.16)
0.16
Free air
27.6
(6.2; 123.2)
<0.0001
25.3
(5.5; 117.7)
<0.0001
Laboratory findings:
Alkaline Phosphatase*, 100
Units/L
1.41
(1.22; 1.64)
<0.0001
1.34
(1.16; 1.55)
<0.0001
Alkaline Phosphatase level >
400 Units/L, n (%)*
9.5
(3.87; 23.3)
<0.0001
5.84
(2.25; 15.15)
0.0003
Alkaline Phosphatase level >
500 Units/L, n (%)*
18.6
(7.1; 48.7)
<0.0001
12.7
(4.7; 34.3)
<0.0001
ALT, 10 Units/L
0.54
(0.38; 0.78)
0.0008
0.54
(0.37; 0.79)
0.0014
AST, 10 Units/L
0.77
(0.68; 0.88)
0.0001
0.79
(0.69; 0.91)
0.0007
WBC, 10
3
/microL
1.23
(1.07; 1.18)
<0.0001
1.095
(1.04; 1.154)
0.0006
ANC, 10
3
/ microL
1.18
(1.09; 1.27)
<0.0001
1.14
(1.05; 1.23)
<0.001
Severe neutropenia (<500/
microL)
0.07
(0.01; 0.57)
0.002
0.09
(0.011; 0.79)
0.03
CRP*, mg/dL
0.86
(0.79; 0.93)
0.0002
0.86
(0.8; 0.93)
0.0003
I:T ratio > 0.2
0.13
(0.05; 0.33)
<0.0001
0.12
(0.05; 0.34)
<0.0001
pH, 10
-1
units
1.37
(1.004; 1.86)
0.047
1.45
(1.01; 2.08)
0.044
BUN, 10 mg/dL
1.56
(1.24; 1.96)
0.0001
1.35
(1.05; 1.72)
0.018
Creatinine, 10
-1
mg/dL
1.18
(1.05; 1.34)
0.008
1.09
(0.96; 1.24)
0.19
BUN/ creatinine ratio, 10 units
1.2
(1.02; 1.41)
0.033
1.1
(0.92; 1.31)
0.31
* – Missing values present

45
Table 5. Multivariate logistic regression analysis (SIP vs NEC).
Predictor variable OR adj 95% CI p-value
Gestational age at birth, per week 0.77 0.61, 0.98 0.023
Day of life on admission, per day 0.83 0.74, 0.93 0.001
Steroids + Indomethacin 14.8 2.0, 111.4 0.009
Alkaline Phosphatase >500 units/L 4.3 1.04, 17.4 0.044
Pneumoperitoneum 21.9 2.5, 187.9 0.005

46
Table 6. Outcomes.
Outcome Indicator
SIP (cases)
n = 48
NEC (controls)
n = 66
p-value
Expired, n (%) 3 (6.3) 28 (42.4) <0.0001
LOS before death, days 8 [5-19] 2 [1-8] 0.09
Survived, n (%) 45 (93.7) 38 (57.4) <0.0001
LOS in survivors, days 90 [43-138] 78 [39-110] 0.37
Transferred back, n (%) 23 (51.1) 16 (42.1)
0.41
Discharged home, n (%) 22 (48.9) 22 (57.9)
LOS, transferred, days 43 [26-79] 36 [24-45] 0.33
LOS, discharged home, days 121 [99-161] 90 [78-150] 0.13
δ
– Continuous variables reported as Median [IQR], unless specified otherwise

47
Figure 1.

48
Figure 2.

49
Figure 3.

50
Figure 4. Alkaline Phosphatase adjusted for GA.

51
Figure 5.

52
Figure 6.

53
54
Figure 7.

55

56
Figure 8.

57
Figure 9. Comparative ROC curve analysis of various predictive models.

58
Figure 10.

59
Figure 11.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Early postnatal cardiovascular maladaptation and peri-intraventricular hemorrhage in extremely preterm infants

PDF

Insulin’s effect on lactate levels in extremely low birth weight neonates. a multi-center, observational study

PDF

Predictors of thrombosis in hospitalized children with central venous catheters: a multi-center predictive study from the CHAT Consortium

Asset Metadata

Creator Azhibekov, Timur A. (author)

Core Title Alkaline phosphatase levels in infants with spontaneous intestinal perforation vs. necrotizing enterocolitis with perforation

School Keck School of Medicine

Degree Master of Science

Degree Program Clinical, Biomedical and Translational Investigations

Publication Date 10/16/2019

Defense Date 10/16/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag alkaline phosphatase,necrotizing enterocolitis,Newborn,OAI-PMH Harvest,predictive modeling,risk factors,spontaneous intestinal perforation

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Mack, Wendy (committee chair), Friedlich, Philippe (committee member), Noori, Shahab (committee member), Seri, Istvan (committee member)

Creator Email tazhibekov@chla.usc.edu,timsky75@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-80892

Unique identifier UC11671440

Identifier etd-AzhibekovT-6853.pdf (filename),usctheses-c89-80892 (legacy record id)

Legacy Identifier etd-AzhibekovT-6853.pdf

Dmrecord 80892

Document Type Thesis

Format application/pdf (imt)

Rights Azhibekov, Timur A.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA