Surveillance Protocol for the Detection of Intracranial Abnormalities in Premature Neonates FREE

CRANIAL ultrasound scans (USs) are widely used to identify preterm neonates at risk for brain injury and subsequent neurodevelopmental defects, most commonly as a consequence of severe intraventricular hemorrhage (IVH) and cystic periventricular leukomalacia (PVL).^{1 - 3} Intraventricular hemorrhage occurs in 40% of premature neonates who weigh less than 1500 g; 90% of hemorrhages occur within the first 3 postnatal days and the remainder by 10 days.^{3 - 6} Intraventricular hemorrhage is usually clinically occult, and detection requires a screening US.³ Periventricular leukomalacia is seen in 3% to 10% of premature neonates.² The sonographic criteria of PVL are increased white matter echogenicity that evolves to cyst formation.³ However, clinically significant white matter injury may occur without frank cyst formation and in the absence of white matter changes on US, as evidenced by development of nonobstructive ventriculomegaly not associated with IVH.⁷ Because of the sensitivity of USs to IVH and periventricular abnormalities and the ease with which USs can be performed at the bedside, surveillance protocols using USs have evolved for screening premature neonates. In recent years, the incidence of all grades of hemorrhage has decreased because of increased antenatal steroid use and improved neonatal care.^{8 - 9} As cost containment has become a more pressing issue, the need for and frequency of screening USs has been raised.^{10 - 12} The objective of this analysis was to determine the optimal timing of cranial USs scanning for identifying preterm neonates weighing less than 1500 g who develop intracranial complications of prematurity.

The study population was derived from 379 neonates with birth weight (BW) of less than 1500 g delivered at Parkland Memorial Hospital, Dallas, Tex, and admitted to the Neonatal Intensive Care Unit between January 1995 and December 1996. Fifty neonates (13%) died within 24 hours from extreme prematurity or other medical conditions and did not undergo a cranial US. In 12 other neonates (3.1%), cranial USs was not performed. Thus, cranial USs were available for review on the remaining 317 neonates.

The long-standing screening protocol in our institution for preterm neonates of less than 1500-g BW includes USs on days 3 to 5, 10 to 14, 28, and/or before discharge from the hospital (hereafter referred to as predischarge). Scans are also obtained more frequently as clinically indicated; these are not considered screening examinations. Standard coronal and parasagittal images are obtained with an Acuson 128 XP ultrasonographic scanner (Acuson, Mountain View, Calif) using a 7.5-mHZ transducer through the anterior fontanel. The USs were prospectively interpreted by 5 pediatric radiologists on a rotating basis as part of standard practice. The results were entered into a database and coded based on the definitions outlined in Table 1.

Ultrasound studies from 150 of these neonates were submitted for review to a pediatric neuroradiologist (N.R.); the images were reviewed in a blinded random manner, not in sequential order, and without knowledge of the results of the previous or subsequent USs. The studies included screening as well as clinically indicated scans and included the cases of severe IVH and its complications, PVL, ventriculomegaly, solitary cysts, some of the cases of lesser grades of IVH and transient periventricular echogenicity (PVE), and 50 normal cases. The BW and gestational age (GA) (as assessed using the modified Ballard measure¹³ ) were also recorded in the database.

Data were analyzed by t test where appropriate. All data are presented as mean ± SD unless otherwise stated.

Two hundred forty-eight neonates (78%) underwent at least 3 USs, 32 (10%) had 2 USs, and 37 (12%) underwent only 1 US. Neonates who were scanned less frequently were larger (ie, 1329 ± 141 g vs 1096 ± 268 g [P = .0001]) than neonates scanned frequently. The initial US obtained on day 3 (range, 1-5 days) was abnormal in 161 neonates (51%) and normal in 156 (49%). The abnormalities included the following: IVH (n = 74) (grade 1, 38 neonates; grade 2, 26 neonates; grade 3 and/or intraparenchymal echogenicity, 10 neonates); increased PVE (68 neonates); ventriculomegaly (7 neonates); solitary cysts (9 neonates) (choroid plexus, 5 neonates, and germinal matrix, 4 neonates); congenital hydrocephalus (1 neonate); lissencephaly (1 neonate); and increased cerebellar echogenicity (1 neonate). No neonate had cystic PVL on the initial scan. The specifics of neonates with severe IVH, cystic PVL, and ventriculomegaly are outlined below.

Severe IVH occurred in 17 neonates (5.3%) with a BW, of 799 ± 184 g, and a GA of 25.3 ± 1.53 weeks. Thirteen (11.4%) of 114 neonates of less than 1000-g BW and 4 (5%) of 79 neonates from 1000- to 1250-g BW. In 11 (65%) of 17 neonates, the IVH was clinically unsuspected. For neonates of 1000-g BW, IVH was diagnosed by days 3 to 5 in 10 (77%), days 10 to 14 in 11 (84%), and day 28 in all 13 cases (Table 1). For neonates of 1001- to 1250-g BW, IVH was diagnosed by days 3 to 5 in 1 (25%), days 10 to 14 in 2 (50%), and day 28 in 3 (75%) of 4 cases. One infant with a grade 1 hemorrhage on day 28 was diagnosed as having a grade 3 IVH on a routine US obtained predischarge. Two neonates (11%) developed posthemorrhagic hydrocephalus during the first postnatal month, ultimately requiring ventricular decompression.

Cystic PVL developed in 9 (2.8%) of the 317 neonates (Table 2). These neonates were larger (BW, 970 ± 228 g) and more mature (GA, 27.2 ± 1.83 weeks) than neonates with severe IVH (P<.01). Six of the 9 neonates were of less than 1000-g BW. In 2 of the 6 cases, increased PVE were observed on the initial US and cyst formation was already obvious by the second US. One infant with 2 normal prior USs had cystic PVL on the third US, and 3 of the neonates developed cystic PVL that was only identified on screening US predischarge. Of the 3 neonates who weighed between 1251 and 1500 g, 2 had increased PVE observed on the initial US and cyst formation was obvious by the second US. The third infant had 3 normal screening USs and subsequently developed meningitis complicated by cystic PVL.

Fourteen neonates with a BW of 918 ± 192 g and a GA of 26.7 ± 1.9 weeks had ventriculomegaly without IVH or cystic PVL (Table 3). The ventriculomegaly varied from mild to moderate and was limited to the frontal horns and atrial trigones of the lateral ventricles (Figure 1). Seven neonates had ventriculomegaly on the initial screening US. In 2 of 7, the ventriculomegaly resolved by the third US or day 28. By day 28, 3 additional neonates had developed ventriculomegaly. Thus, 8 neonates had nonobstructive ventriculomegaly by day 28. Four additional neonates had ventriculomegaly on a predischarge sonogram. Thus, 12 neonates had ventriculomegaly at the time of discharge from the hospital (Table 4). In 10 of 12 cases, the ventriculomegaly was bilateral, and in 2 cases, unilateral.

Twenty-one neonates died during hospitalization, 37 neonates had a single US performed during hospitalization, and 3 neonates were transferred to a secondary hospital. Thus, 256 neonates had a predischarge US. The predischarge US was normal in 181 neonates (72%) and abnormal in 75 (28%). The abnormalities included resolving IVH (n = 31) and late-onset clinically occult IVH (n = 5) (ie, grade 1 IVH, 4 neonates; grade 3 IVH, [1 neonate; hydrocephalus/porencephaly (n = 5); ventriculomegaly (n = 12); PVL (n = 9); solitary cysts (n = 5); and new-onset PVE (n = 2). Thalamic echoes, either as isolated phenomena or in association with one of the above abnormalities, developed in 10 neonates.

Nine significant abnormalities (ie, severe IVH [1 neonate], cystic PVL [4 neonates], and ventriculomegaly [4 neonates]) were detected by the predischarge US. The ventriculomegaly was bilateral in 3 and unilateral in 1 case (Table 5).

The overall concordance between prospective interpretation and the review of the 150 US studies was 98%. Discrepancies in interpretation related to the conspicuousness and significance of white matter echogenicity.

The data in this report indicate that up to 50% of neonates weighing less than 1500 g exhibited some abnormality on the initial US, although abnormalities considered to be of clinical significance occurred in less than 5% and were seen only in the smallest neonates. Severe IVH was observed in approximately 11% of the neonates weighing less than 1000 g and in 5% of those between 1000- and 1250-g BW. Cystic PVL was noted in 5% of the neonates weighing less than 1000-g and in approximately 1% of those between 1250 and 1500 g. Ventriculomegaly in the absence of IVH, considered to be a marker of occult white matter injury, was noted in 7% of neonates of less than 1000-g BW.^{7 ,14} The overall incidence of these abnormalities is consistent with previous reports.^{3 ,6} However, the timing or evolution of these lesions seems to be changing. Previous studies have indicated that most cases of severe IVH (>90%) occur in the first 3 days of life,^{3 ,6} whereas, in this report, approximately 65% of the cases were detected within the first week. The remaining cases were detected in the second or third postnatal week; 1 infant developed severe IVH after day 28. Cystic PVL also tended to occur later and often in the absence of prior sonographic abnormalities. This is consistent with some previous observations^{15 - 16} but is in conflict with other studies, which indicate that most cases of cystic PVL are heralded by prior hyperechogenicity in the early postnatal period.^{3 ,17 - 18} The onset of ventriculomegaly was more varied. In 50% of neonates, ventriculomegaly was present on the initial US, whereas in the remaining cases, the ventricles enlarged over time; in 25% of the latter cases, ventriculomegaly only became apparent after day 28. Recent data indicate the importance of ventriculomegaly as an independent predictor of adverse neurodevelopmental outcome.^{19 - 21}

Although 50% of neonates had an abnormal initial US, 72% of these were interpreted as normal by the time of discharge from the hospital. Of the approximately 30% of neonates with an abnormal predischarge US, the abnormalities were severe in 9 cases (12%), and only detected on the predischarge US.

The observations in this report may provide a basis for making recommendations regarding the frequency and timing of screening USs in high-risk premature neonates. However, any screening protocol needs to take into account that most intracranial pathologic processes in the sick premature neonate is clinically occult,^{3 ,11} that severe lesions occur most often in the tiniest of premature neonates, and that there seems to be a shift toward a delayed presentation of the clinically more significant lesions. Based on the above observations, we would recommend that screening USs be performed as follows: for neonates of less than 1000-g BW, an initial screening US should be obtained on days 3 to 5; this should identify at least 75% of cases of IVH and may identify early periventricular white matter abnormalities. A second US on days 10 to 14 should detect most cases of IVH and identify early hydrocephalus.³ Early cyst formation within white matter may also be apparent at this time. We would recommend a third scan on day 28. This should detect almost all cases of severe IVH and assess periventricular white matter and ventricular size. Given the unexpectedly high yield of significant sonographic abnormalities (ie, 20% noted on the predischarge US), we consider a predischarge US is indicated in all neonates of less than 1000-g BW. For neonates of 1001- to 1250-g BW, we recommend a screening US on days 3 to 5, which should detect at least 40% of significant lesions; a second scan at day 28, which should detect at least 70% of significant lesions; and a final predischarge US. For neonates between 1250- and 1500-g BW, we recommend a US on days 3 to 5 and a second scan before discharge from the hospital if the clinical course is complicated in any way. Scans should obviously be obtained more frequently as clinically indicated for all BW categories. This reduction in frequency of screening USs in the relatively "larger high-risk neonate" (ie, >1000 g) (for our population, this approximates 300 per year) should not reduce the ability to detect significant neuropathologic processes even in the absence of clinical symptoms, and should, at the same time, address the concerns of cost containment (approximately $30,000 per year in our institution). Clearly, these issues can only be definitively addressed in a prospective randomized study.

Accepted for publication March 30, 2000.

Corresponding author: Jeffrey M. Perlman, MB, Department of Pediatrics, University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75235-9063 (e-mail: jperlm@mednet.swmed.edu).