Late-onset systemic hypotension and oliguria without obvious cause following stable circulation and respiration are not rare in preterm infants, and such episodes are referred to as late-onset circulatory dysfunction (LCD) of premature infants [1]. LCD due to adrenal insufficiency (AI) is one of the major causes of late-onset hypotension [2]. Recently, several studies performed in Japan have reported that preterm infants develop LCD that responds to glucocorticoid therapy, but not to volume expansion or vasopressors, after the first week of life [3-6]. The main problem associated with LCD is the subsequent onset of periventricular leukomalacia (PVL) and poorer neurologic development [1,3]. The purpose of this study was to characterize the brain ultrasonographic findings of LCD due to AI in preterm infants.

This retrospective study was performed with the approval and under the guidelines of the Institutional Review Board. Informed consent was waived.

The ultrasonographic findings of 14 preterm infants with LCD due to AI are shown in Table 1. LCD occurred at a median of postnatal age 18.5 days (range, 9 to 32 days) and a mean of 33.4 weeks (range, 31.9 to 35.0 weeks) of postmenstrual age. The serum cortisol concentration before hydrocortisone administration was median 5.3±2.6 μg/dL (range, 1.7 to 9.2 μg/dL; normal reference, 5 to 25 μg/dL). All 14 patients were stable before the onset of LCD and there was no history of birth asphyxia or hypoxic ischemic encephalopathy. The interval of ultrasonographic examination before and after LCD was a mean of 10.1 days (range, 4 to 18 days), the interval between LCD and ultrasonography was a mean of 2.6 days (range, 0 to 11 days), and the interval of next follow-up ultrasonography after LCD was a mean of 10.3 days (range, 3 to 25 days).

The last brain ultrasonographic findings before the occurrence of LCD showed grade 1 PVE in all 14 patients and germinal matrix hemorrhage (GMH) with focal cystic change in one patient. Brain ultrasonographic findings after LCD demonstrated no significant change in grade 1 PVE and no new lesions in eight (57%), grade 1 PVE with newly appearing GMH in three (21%) (Fig. 1C, D), and increased PVE in three (21%) infants (Figs. 2C, D, 3E, F). Five infants (36%) showed new development (n=4) or increased size (n=1) of GMH (Figs. 1C, D, 3D). The date of occurrence of GMH in the five patients was a median of 34 days (range, 16 to 42 days) after birth (Figs. 1C, D, 3D). Two of three infants with increased PVE developed cystic PVL and showed rapid progression to macrocystic encephalomalacia (Figs. 2, 3), especially in both parieto-occipital lobes. Periventricular cyst formations were noted at the 11th (Fig. 3, case 8) and the 18th (Fig. 2, case 3) days after LCD occurrence, that is, in both cases, the 42nd day of life. The remaining one infant (case 10) with increased PVE developed non-cavitary PVL on magnetic resonance imaging (MRI) at a corrected age of 20 months.

In this study, the incidence of LCD due to AI was 5.4%. Kusuda et al. [8] reported an incidence of LCD due to AI of 4% in 2,145 infants with very low birth weight (≤1,500 g) in Japan during a stay in the NICU in 2003. Choi et al. [7] reported the incidence was 7.6% in 128 preterm infants in Korea from January 2007 to August 2009.

Characteristic ultrasonographic findings in our study are the late development of GMH after 2 weeks of age, increased PVE after the occurrence of LCD, and late development of PVL. GMH-IVH in premature infants is typically diagnosed during the first days of life, 50% on the first day and 90% within the first 4 days [9]. New development or increased size of GMH in five patients occurred at 16-42 days of life in our study. Delayed hemorrhage (GMH-IVH) after 2 weeks of life may be related to postnatal events or insults and should not be overlooked.

The three patients who showed increased PVE after the occurrence of LCD all developed PVL (cystic PVL with macrocystic encephalomalacia [n=2] and non-cavitary PVL [n=1]) associated with adverse neurodevelopmental outcomes. Severe cystic PVL and rapid progression to macrocystic encephalomalacia were found unexpectedly on routine follow-up ultrasonography in two patients, although they had a good clinical response to steroid and had no other causes of hypotension except the occurrence of AI. The cystic changes were more rapid and extensive than typical PVL and predominantly involved both parieto-occipital lobes. The periventricular cyst formations were noted at the 11th and the 18th day after LCD occurred, that is, the 42nd day of life in both of these two infants. Kobayashi et al. [1,3] reported that in infants with PVL, those with LCD had more severe MRI findings and developed more severe cerebral palsy than those without LCD. The pathogenesis of PVL development in patients with LCD is thought to be remarkable hypotension that might lead to a striking decrease in cerebral blood flow (CBF) to the white matter. They presumed the hypotension during LCD is associated with the development of PVL. Young et al. [10] demonstrated that the regional CBF to periventricular and occipital (visual radiations) white matter significantly decreases during severe hypotension in newborn dogs, whereas regional CBF to all gray matter structures is preserved. Although our two cases with severe cystic PVL and macrocystic encephalomalacia are well correlated with previous reports [1,3,10,11], the number of patients is too small to draw conclusions.

In stable preterm infants after 2-4 weeks of age, the interval for follow-up ultrasonography was usually lengthened to 2-4 weeks. However, when LCD occurs due to AI, the interval of ultrasonography should be shortened. Kobayashi et al. [3] reported that periventricular cysts were initially detected far later in cases of PVL with LCD than in those without LCD. This might be attributed to the decreased frequency of ultrasonography performed once an infant has reached 1 month of age. In our study, the interval of ultrasonographic examination before and after LCD was as long as 18 days (mean, 10.1 days; range, 4 to 18 days) and the interval of next follow-up ultrasonography after LCD was up to 25 days (mean, 10.3 days; range, 3 to 25 days). In seven cases, even if ultrasonography was performed shortly after LCD (the interval between LCD and ultrasonography was 0-1 day), the ultrasonographic changes were subsequently detected later because the interval of next follow-up ultrasonography after the occurrence of LCD was up to 25 days (mean, 10.3 days; range, 3 to 25 days). Therefore, if hypotension is suspected to be associated with LCD due to AI, we think the interval of ultrasonographic examination should be shortened to 1 week, at least for 1 month after LCD occurs.

This study had several limitations, such as its retrospective design and small number of patients. Further research is needed to specify the ultrasonographic findings of LCD due to AI in a large population with a control group.

In conclusion, LCD due to AI may be associated with the late development of GMH (36%) after 2 weeks of age, increased PVE (21%) after LCD, and the development of cystic PVL with rapid progression to macrocystic encephalomalacia (14%).