This is a multi-centered cross-sectional study conducted in Namazi, Hafez and Hazrat Zeinab hospitals, affiliated to Shiraz University of Medical Sciences, Shiraz, Iran. During January 2016 to July 2016, neonates who were admitted with a diagnosis of asphyxia based on the Sarnat and Sarnat clinical criteria (12) in the mentioned health care centers, were considered for entry in the study. Neonates born between 37 and 42 weeks who had signs of fetal distress such as abnormal fetal monitoring and presence of meconium in the amnion, with an umbilical cord blood PH of less than 7.1, five minute Apgar score of less than five, were included in the study.
Neonates born earlier than 37 weeks or later than 42 weeks, those who had any congenital disorders (for example patent ductus arteriosus), and those with myocardial dysfunction, were excluded from the study.
We excluded patients with myocardial dysfunction and congenital abnormalities as some studies have shown these patients to have variable findings regarding DS indexes (13).
Protocol and patient evaluation
Patients were initially evaluated by a neonatologist and were scored based on the Sarnat and Sarnat scoring system (previously mentioned) as mild, moderate and severe asphyxia. After diagnosis of asphyxia through clinical evaluation by the neonatologist, and after patients were clinically stable, during 6 to 24 hours after birth, each patient underwent DS (Shenzhen Mindray Bio-Medical Electronics Co., China) of the cerebral arteries using a curved probe of 5 MHz and a linear probe of 7.5 MHz. As DS studies are dynamic, DS was done after the first 6 hours (up to 24 hours) of birth in order to obtain stability and more importantly minimize changes in DS findings.
For DS of the ACA, imaging was conducted on bilateral sides on parasagittal planes through the anterior fontanelles. Parameters related to the ACA were measured using a branch of the anterior artery, which were anterior to the corpus callosum. For the MCA, imaging was conducted on bilateral sides by both temporal bones between the eye socket and the ear above the zygoma on axial planes. For the BA, imaging was obtained in the sagittal planes located just before the pons. Examination of the neonates was performed during sleep.
Magnetic resonance imaging (MRI) was done for each neonate during the first month, after discharge or during hospital admission, when the patients obtained a clinically stable condition. Evaluation of MRI findings was done as followed: limited hyper intense T2 signal in brain cortex and subcortical white matter represented mild degree of hypoxic ischemic encephalopathy. Involvement of basal ganglia and thalamus was considered severe hypoxic ischemic encephalopathy. Moreover, mild to moderate encephalopathy were considered disseminated signal changes of the cortex and subcortical weight matter (14). Accordingly, patients were categorized into three groups of normal, mild to moderate (stage 2) and severe (stage 3) regarding asphyxia.
MRI and sonography findings were interpreted by two different radiologists who were both unaware of the clinical staging and the staging done by the other radiologist.
Data including sex, one minute and five minute APGAR scores, type of delivery, birth weight, gestational age, clinical severity of asphyxia, severity of asphyxia based on sonography (based on ACA, MCA and BA), and severity of asphyxia based on MRI findings, were all registered in a data gathering sheet.
We used the Sarnat and Sarnat criteria for clinical diagnosis and classification of severity of asphyxia. This criteria considers six variables for the diagnosis of asphyxia including: alertness, muscle tone, seizure, pupils, respiration pattern, and duration of symptoms (12).
For sonography based staging of asphyxia, the Pourcelots's resistive index (RI) was used to estimate cerebral blood flow status, due to its easiness of use, reproducibility and its independence of the angle of insonation (15).
Peak systolic velocity (PSV), end diastolic velocity (EDV) and resistive index (RI) of the anterior cerebral arteries (ACA), middle cerebral artery (MCA), and the basilar arteries (BA) were measured and the severity of disease was defined, accordingly. Severity of disease was measured according to the RI, which is calculated as the peak systolic velocity minus the end diastolic velocity divided by the systolic velocity. The RI was measured three consecutive times for each artery and the average was considered the final RI.
RI provides a tool to evaluate the dynamics of cerebral blood flow and cerebral pressure (16). Appropriate cut-off points for classifying patients based on RI as sever, moderate and mild were obtained using previous literature (17). According to the mentioned study a cut-off of RI ≤ 0.57 was defined as severe asphyxia, moreover according to our final results for the diagnosis of asphyxia (using our own obtained cut-off point for the diagnosis of asphyxia based on RI) the rest of the cut-offs were categorized accordingly. According to our results and that of previous literature (as mentioned before), severe asphyxia was considered as RI ≤ 0.57, moderate as RI = 0.58–0.62, mild was considered as RI = 0.63–0.67, and normal DS was considered as RI = 0.68–0.72.
In the end, staging based on sonography was compared with the staging based on MRI findings.
All patients were hospitalized in the neonatal intensive care units and received related intensive care according to standard protocols, therefor factors such as thermoregulation and vasogenic edema which may have affected RI were controlled and were similar for all the patients.
Data was analyzed using the Statistical Package for Social Sciences software (SPSS Inc., Chicago, IL, USA) for windows, version 16.
In order to evaluate the linear correlation between DS findings of the ACA, MCA and BA with MRI, the Spearman's correlation test was used.
To determine the ideal cut-off point regarding RI of the three arteries, in discriminating between normal neonates and those with asphyxia (considering MRI as the gold standard), the receiver operating characteristics (ROC) analysis was used, reporting its area under the curve (AUC), sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR) and negative likelihood ratio (NLR), where appropriate. In addition, a pairwise comparison of ROC curves was performed between the ACA, MCA and BA to determine the difference between the examined arteries.
We calculated the optimum RI cut-off point using the Youden index (18). Using this model, the ideal cut-off point on the ROC curve is considered optimum which has the maximum sensitivity + specificity. Upper and lower limits of cut-off points were also determined considering the point on the ROC curve with the highest sensitivity and highest specificity.
Data are presented as means ± standard deviations (SD) or frequency and percentage, where appropriate.