Introduction
Approximately 2–5% of all children in North America and Europe have fetal alcohol spectrum disorder (FASD) (Chudley et al., Reference Chudley, Conry, Cook, Loock, Rosales and LeBlanc2005; May et al., Reference May, Gossage, Kalberg, Robinson, Buckley, Manning and Hoyme2009) and the incidence is much higher in South African black and Canadian First Nations populations (May et al., Reference May, Gossage, Marais, Hendricks, Snell, Tabachnick and Viljoen2008; Robinson, Conry, & Conry, Reference Robinson, Conry and Conry1987). The severe deficits and behavioral problems associated with this condition (Kodituwakku, Reference Kodituwakku2007) have made it a challenge for parents/caregivers, health professionals (Chudley et al., Reference Chudley, Kilgour, Cranston and Edwards2007), and the justice system alike (Burd et al., 2003; Fast & Conry, Reference Fast and Conry2009). The range and severity of deficits in FASD are usually, although not always, reflective of the duration and amount of alcohol consumed, the gestational period of consumption (Autti-Rämö & Granstrom, Reference Autti-Rämö and Granstrom1991; Berman & Hannigan, Reference Berman and Hannigan2000), and such mitigating factors as poverty, deprivation, pre- and postnatal undernutrition, multiple foster placements, and abuse and neglect (May & Gossage, Reference May and Gossage2011).
Within the fetal alcohol spectrum, the best-known condition is fetal alcohol syndrome (FAS), which is characterized by a constellation of features that include growth deficiency, distinct facial dysmorphia (e.g., palpebral fissures, long flat philtrum, thin vermillion), and significant central nervous system (CNS) impairments (Stratton, Howe, & Battaglia, Reference Stratton, Howe and Battaglia1996). However the most prevalent form of FASD, estimated at 10 times the frequency of FAS (Stoler & Holmes, Reference Stoler and Holmes1999), is alcohol-related neurodevelopmental disorder (ARND), which involves only CNS abnormalities (Clarren & Smith, Reference Clarren and Smith1978; Riley & McGee, Reference Riley and McGee2005; Stratton et al., Reference Stratton, Howe and Battaglia1996). Because children with ARND lack defining physical features for clinical recognition, a substantial proportion may be misdiagnosed or wrongly diagnosed (Chudley et al., Reference Chudley, Kilgour, Cranston and Edwards2007), signifying an even higher prevalence than formerly estimated.
Research on persons with FASD has revealed IQ reductions (Rasmussen et al., 2008), cognitive impairments (Astley, Carmichael Olson, et al., Reference Astley, Carmichael Olson, Kerns, Brooks, Aylward, Coggins and Richards2009; Nash et al., Reference Nash, Stevens, Rovet, Fantus, Nulman, Sorbara and Koren2013), and social and behavioral difficulties (Stevens et al., Reference Stevens, Nash, Fantus, Nulman, Rovet and Koren2013; see Donaldson et al., Reference Donaldson, Coles, Hagan, Evans, Klain, Kosofsy and Yolton2011) as well as predisposition to mental health issues (Pei, Rinaldi, Rasmussen, Massey, & Massey, Reference Pei, Rinaldi, Rasmussen, Massey and Massey2008). These include attention deficit disorder (Nash et al., Reference Nash, Rovet, Greenbaum, Fantus, Nulman and Koren2006; O'Malley & Nanson, Reference O'Malley and Nanson2002; Rasmussen et al., Reference Rasmussen, Benz, Pei, Andrew, Schuller, Abele-Wester and Lord2010), conduct disorder (Nash, Koren, & Rovet, Reference Nash, Koren and Rovet2011), and autism (Bishop, Gahagan, & Lord, Reference Bishop, Gahagan and Lord2007; Stevens, Nash, Koren, & Rovet, Reference Stevens, Nash, Koren and Rovet2012). As adults, individuals with FASD typically do not complete their educations and are unemployed while many experience depression, suicide risk, and trouble with the law (O'Malley & Huggins, Reference O'Malley and Huggins2005; Streissguth et al., Reference Streissguth, Bookstein, Barr, Sampson, O'Malley and Kogan Young2004). Consequently, FASD confers an enormous life-long burden and a high cost to society (Lupton, Burd, & Harwood, Reference Lupton, Burd and Harwood2004; Stade, Ungar, Stevens, Beyene & Koren, Reference Stade, Ungar, Stevens, Beyene and Koren2006; Stade et al., Reference Stade, Ali, Bennett, Campbell, Johnston, Lens and Koren2009).
In recent years, extensive literature has emerged on the neuroanatomic characteristics of FASD. Findings indicate global brain volume reductions (Archibald et al., Reference Archibald, Fennema-Notestine, Gamst, Riley, Mattson and Jernigan2001) with reduced size of specific brain regions including parietal, temporal, and frontal lobes (Lebel, Rousette, & Sowell, Reference Lebel, Roussotte and Sowell2011; Sowell et al., Reference Sowell, Thompson, Mattson, Tessner, Jernigan, Riley and Toga2002; Spadoni, McGee, Fryer, & Riley, Reference Spadoni, McGee, Fryer and Riley2007), caudate (Cortese et al., Reference Cortese, Moore, Bailey, Jacobson, Delaney-Black and Hannigan2006), cerebellum (Sowell et al., Reference Sowell, Jernigan, Mattson, Riley, Sobel and Jones1996), corpus callosum (Autti-Rämö et al., Reference Autti-Rämö, Autti, Korkman, Kettunen, Salonen and Valanne2002; Riley et al., Reference Riley, Mattson, Sowell, Jernigan, Sobel and Jones1995), and hippocampus (Coles et al., Reference Coles, Goldstein, Lynch, Chen, Kable, Johnson and Hu2011; Willoughby, Sheard, Nash, & Rovet, Reference Willoughby, Sheard, Nash and Rovet2008). Also observed are cortical and subcortical gray matter reductions (Astley, Aylward, et al., Reference Astley, Aylward, Carmichael Olson, Kerns, Brooks, Coggins and Richards2009; Nardelli, Lebel, Rasmussen, Andrew, & Beaulieu, Reference Nardelli, Lebel, Rasmussen, Andrew and Beaulieu2011), abnormalities in cortical morphology (Sowell et al., Reference Sowell, Mattson, Kan, Thompson, Riley and Toga2008; Yang et al., Reference Yang, Roussotte, Kan, Sulik, Mattson, Riley and Sowell2012; Zhou et al., Reference Zhou, Lebel, Lepage, Rasmussen, Evans, Wyper and Beaulieu2011), white matter irregularities (Lebel et al., Reference Lebel, Rasmussen, Wyper, Walker, Andrew, Yager and Beaulieu2008; Wozniak et al., Reference Wozniak, Muetzel, Mueller, McGee, Feerks, Ward and Lim2009), and functional disturbances (Fryer, McGee, Matt, Riley, & Mattson, Reference Fryer, McGee, Matt, Riley and Mattson2007; Malisza et al., Reference Malisza, Allman, Chiloff, Jakobson, Longstaffe and Chudley2005; Sowell et al., Reference Sowell, Lu, O'Hare, McCourt, Mattson, O'Connor and Bookheimer2007). Studies examining children with ARND almost exclusively (Rajaprakash, Chakravarty, Lerch, & Rovet, 2014; Willoughby, Sheard, Nash, & Rovet, Reference Willoughby, Sheard, Nash and Rovet2008) or versus other FASD forms (Astley, Aylward et al., Reference Astley, Aylward, Carmichael Olson, Kerns, Brooks, Coggins and Richards2009; Lebel et al., Reference Lebel, Rasmussen, Wyper, Walker, Andrew, Yager and Beaulieu2008) have found similar neuroanatomic effects between subgroups as in the entire FASD group.
Among individuals with FASD, one consistent finding is a weakness in memory skills (Mattson, Riley, Elis, Stern, & Lyons, Reference Mattson, Riley, Delis, Stern and Lyons1996; Rasmussen, Horne, & Witol, Reference Rasmussen, Horne and Witol2008; Richardson, Ryan, Willford, Day, & Goldschmidt, Reference Richardson, Ryan, Willford, Day and Goldschmidt2002; Uecker & Nadel, Reference Uecker and Nadel1996), particularly on hippocampally mediated tasks (Willford, Richardson, Leech, & Day, Reference Willford, Richardson, Leech and Day2004). In a study using a virtual Morris Water Maze task, the prototype for studying the hippocampus in rodents, individuals with heavy prenatal alcohol exposure (PAE) exhibited comparable learning and memory deficits as the hippocampally damaged rodents (Hamilton, Kodituwakku, Sutherland, & Savage, Reference Hamilton, Kodituwakku, Sutherland and Savage2003). Although studies specifically of the ARND subgroup have reported either equivalent (Astley, Carmichael Olson, et al., Reference Astley, Carmichael Olson, Kerns, Brooks, Aylward, Coggins and Richards2009) or milder deficits as in FAS (Coles et al., Reference Coles, Goldstein, Lynch, Chen, Kable, Johnson and Hu2011), inconsistencies exist among studies. For example, Willoughby et al. (Reference Willoughby, Sheard, Nash and Rovet2008) reported youth with ARND had a wide range of verbal and nonverbal memory weaknesses relative to controls, whereas Rasmussen, McCauley, & Andrew (2008) comparing those with an ARND diagnosis and those with PAE not meeting diagnostic criteria for FASD found the only differences were in face recognition and number repetition and groups otherwise performed similarly. Thus the specific memory difficulties of children with ARND are not yet known.
When the hippocampus was directly examined in individuals with FASD, structural (e.g., Autti-Rämö et al., Reference Autti-Rämö, Autti, Korkman, Kettunen, Salonen and Valanne2002) and functional (Sowell et al., Reference Sowell, Lu, O'Hare, McCourt, Mattson, O'Connor and Bookheimer2007) abnormalities, as well as differences, in laterality were observed (Riikonen, Salonen, Partanen, & Verho, Reference Riikonen, Salonen, Partanen and Verho1999). However in the ARND subgroup specifically, studies showed varying findings. For example, Astley, Aylward, et al. (Reference Astley, Aylward, Carmichael Olson, Kerns, Brooks, Coggins and Richards2009) observed both ARND and FAS youth aged 8 to 16 years had significantly smaller right and left hippocampi than controls. In contrast, Willoughby et al. (Reference Willoughby, Sheard, Nash and Rovet2008) studying ARND youth primarily observed only left hippocampal volume reductions, whereas Coles et al. (Reference Coles, Goldstein, Lynch, Chen, Kable, Johnson and Hu2011) studying adults with heavy PAE found memory deficits were mediated by right hippocampal volume reductions but only if they showed facial dysmorphia (i.e., FAS but not ARND).
In hippocampal research, a recent emphasis has been the ascription of different memory functions to specific hippocampal subregions (Moser & Moser, Reference Moser and Moser1998). According to Poppenk and Moscovitch (Reference Poppenk and Moscovitch2011), the anterior hippocampus subserves the encoding of new information (Fanselow & Dong, Reference Fanselow and Dong2010), whereas the posterior hippocampus contributes to recollective aspects of memory and event reconstruction (see also, Poppenk, Evensmoen, Moscovitch, & Nadel, Reference Poppenk, Evensmoen, Moscovtch and Nadel2013). For example, London taxi drivers, who have a vast knowledge of spatial locations, show enlarged posterior hippocampi at the expense of their anterior hippocampus (Maguire et al., Reference Maguire, Gadian, Johnsrude, Good, Ashburner, Frackowiak and Frith2000). Poppenk and Moscovitch (Reference Poppenk and Moscovitch2011) studying young adults observed those with large posterior but small anterior hippocampal regions had better recall of previously learned proverbs than age-matched counterparts. However, it is not known (a) whether posterior and anterior regions are differentially affected in children with FASD, and the ARND subgroup specifically, and (b) what are the implications of regional differences for specific memory difficulties.
An extensive literature on rodents with prenatal ethanol exposure has shown severe memory impairments and associated hippocampal abnormalities (Berman & Hannigan, Reference Berman and Hannigan2000; Klintsova et al., Reference Klintsova, Helfer, Calizo, Dong, Goodlett and Greenough2007; Livy, Miller, Maier, & West, Reference Livy, Miller, Maier and West2003). While mice exposed to ethanol at gestational day 7 exhibited the FAS facial deformity, those exposed later had different or no unusual facial features, as in ARND (Lipinski et al., Reference Lipinski, Hammond, O'Leary-Moore, Ament, Pecevich, Jang and Sulik2012), however, they still showed hippocampal volume reductions, especially on the right side (Parnell et al., Reference Parnell, O'Leary-Moore, Godin, Dehart, Johnson, Johnson and Sulik2009). Additionally, ethanol-exposed rodents show sex differences in hippocampally mediated memory functions. For example, exposed females had larger encoding deficits than males (Minetti, Arolfo, Virgolini, Brioni, & Fulginiti, Reference Minetti, Arolfo, Virgolini, Brioni and Fulginiti1996), who instead showed larger recall deficits than females (Kelly, Leggett, & Cronise, Reference Kelly, Leggett and Cronise2009). Furthermore, when rodents were exposed to ethanol in the period corresponding to the human third trimester, males and females both had spatial learning impairments but males exhibited deficits with just 2 days of exposure while the females needed 4 days, suggesting greater male vulnerability (Goodlett & Peterson, Reference Goodlett and Peterson1995).
In humans not exposed gestationally to alcohol, sexual dimorphisms in hippocampal volume are consistently reported. Females, for example, typically show larger hippocampi (per brain size) and more rapid rates of hippocampal growth than males (Filipek, Richelme, Kennedy, & Caviness, Reference Filipek, Richelme, Kennedy and Caviness1994; Giedd et al., Reference Giedd, Vaituzis, Hamburger, Lange, Rajapakse, Kaysen and Rapoport1996; Giedd, Castellanos, Rajapakse, Vaituzis, & Rapoport, Reference Giedd, Castellanos, Rajapakse, Vaituzis and Rapoport1997), a finding attributed to hormonal effects (Lord, Buss, Lupien, & Preussner, Reference Lord, Buss, Lupien and Preussner2008). Nevertheless inconsistencies exist as to what hippocampus is most affected in males versus females and where within the hippocampus disturbances occur. Neufang et al. (Reference Neufang, Specht, Hausmann, Gunturkun, Herpertz-Dahlmann, Fink and Konrad2009) found that in 8- to 15-year-olds, females had larger hippocampi bilaterally than same-age males. In contrast, Gogtay et al. (Reference Gogtay, Nugent, Herman, Ordonez, Greenstein, Hayashi and Thompson2006) found in typically developing youth aged 4 to 18 years, sex differences were confined to specific hippocampal subregions. These reflected more prominent growth in the left posterior hippocampus and greater volume loss at the posterior hippocampal pole in females than males and the greater age-related volume loss at the head of the hippocampus in males. However, sex differences in hippocampal volumes have not been studied in youth with FASD, particularly those with the ARND subtype.
The present study was conducted in the context of a larger investigation of memory and the hippocampus in children with the ARND variant of FASD. Currently, we sought to address some of the knowledge gaps and inconsistencies identified above. Specifically, we asked whether youth with ARND differ from non-exposed typically developing controls in memory abilities, size of their hippocampi and specific hippocampal segments, sex differences in memory functions and hippocampal structure, and correlations between specific memory deficits and hippocampal reductions.
Methods
Participants
Participants were 35 children and adolescents aged 11.1 to 14.8 years, 18 with a diagnosis of ARND along the FASD spectrum and 17 typically developing controls. All had no MRI counter-indications (e.g., braces, implants).
The ARND group consisted of 11 males and 7 females previously diagnosed at the Motherisk Follow-up Clinic at the Hospital for Sick Children (SickKids), a regional FASD diagnostic facility, which included a pediatrician, several psychologists and psychometrists, and a speech therapist. Most children attending this clinic were brought by a foster or adoptive parent who sought to ascertain whether the child's current cognitive or behavioral problems were related to PAE and then obtain the necessary services. To be assessed in this clinic, a history of PAE first had to be confirmed from direct report of the mother or a relative or valid documentation that the mother was an alcoholic or received treatment for alcoholism during pregnancy or had the child removed by the Children's Aid Society (CAS) at birth due to her alcohol abuse. If suitable, the child received a thorough physical and neurological assessment by the pediatrician and a detailed medical history was obtained from the caregiver or an also accompanying social worker. Finally, children underwent a comprehensive neuropsychological evaluation. Children diagnosed with FAS showed the requisite facial dysmorphology, growth retardation, and either an IQ score below 70 or a neuropsychological profile indicating deficits (scores 2 standard deviations below test mean) on subtests in any three of the domains specified by the Canadian FASD Diagnostic Guideline system (Chudley et al., Reference Chudley, Conry, Cook, Loock, Rosales and LeBlanc2005). Children not showing the physical features but having the requisite neuropsychological profile were considered to have ARND. Because the partial FAS classification was not being used in the clinic at the time current participants received their diagnoses, all children were classified as having ARND. Only children with IQs >70 were included presently.
Controls were 17 (10 males and 7 females) typically developing children recruited from among non-ARND foster or step-siblings, participant lists from previous studies, and local advertising within SickKids. They were matched for age (within 6 months) and sex with children in the ARND group. All of their mothers reported not drinking alcohol or taking medications for a major illness during pregnancy. Any child with a reported neurological disease, head injury, chronic illness, learning disability, psychiatric disorder, or obtaining a low IQ score was excluded.
Procedures
The study consisted of two visits over a 6-month period. The first session involved a 4-hr assessment that included an intelligence test, multiple tests of memory and other abilities (e.g., executive functioning, reported elsewhere). The assessment was conducted by a team of psychometrists and advanced graduate students trained on all tasks and masked as best as possible to group status. Snack and lunch breaks were provided as needed. The second session involved two 1-hr same-day MRI scans in a 1.5 Tesla Signa Excite (General Electric, Milwaukee, WI) scanner in the SickKids Diagnostic Imaging Unit. Both structural and functional sequences were performed (functional results described elsewhere; e.g., Rovet, Sheard, Wheeler, & Skocic, Reference Rovet, Sheard, Wheeler and Skocic2010). To minimize movement during structural scanning, children viewed movies via MRI-compatible goggles.
Initially, parents/caregivers gave written consent and participants gave oral assent. After each session, participants received two movie passes and a certificate of participation for high school credit hours; parents/caregivers were compensated for travel expenses. At the end of scanning, participants also received a CD containing their own brain images. Parents/caregivers received a detailed report of the child's performance within 2 months of the assessment. Additionally, all scans were reviewed for gross abnormalities by a staff neuroradiologist masked to group status and her report was sent to each child's physician. The Research Ethics Board at SickKids approved all procedures.
Tests and Measures
The following tests were currently selected from our larger test battery: Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, Reference Wechsler1999) based on Vocabulary and Matrix Reasoning subtests; selective Children's Memory Scale subtests (CMS; Cohen, Reference Cohen1997); complete Test of Memory and Learning (TOMAL; Reynolds & Bigler, Reference Reynolds and Bigler2007); Rey-Osterrieth Complex Figure task (ROCF; Bernstein & Waber, Reference Bernstein and Waber1999; Osterrieth & Rey, Reference Osterrieth and Rey1944; Taylor, Reference Taylor1991); and three Cambridge Neuropsychological Test Automated Battery (CANTAB, 1998) subtests (see Table 1).
Descriptions of cognitive measures
Image Acquisition and Processing
The structural sequence consisted of a 7-min high-resolution axial T1 3D FSPGR sequence (fast spoiled gradient recalled echo) with inversion recovery to provide increased T1 weighting and allow for enhanced contrast of gray and white matter tissue. Approximately 125 1.5-mm-thick slices were obtained per scan to provide whole brain coverage. T1 image acquisition parameters were: repetition time = 10.3 ms, echo time = 4.2 ms, inversion time = 400 ms, flip angle = 20°, and a 256 × 192 acquisition matrix. Three additional clinical sequences (10-min) read by the neuroradiologist for gross brain abnormalities were: a Sagittal T1 Flair, a Coronal T2 Fast Relaxation Fast Spin Echo, and an Axial T2 Flair.
All images were subsequently transferred to a Linux workstation running Fedora 7 (Red Hat, Raleigh, NC). In Analyze 9.0 (Mayo Clinic, Rochester, MN), Dicom files were reconstructed to have isotropic voxels of less than 1 mm and transformed into standard space using AC-PC (anterior commissure-posterior commissure) alignment. For all tracing, an optical wheel mouse was used to manually define regions of interest (ROIs). Right and left hippocampi were identified in each plane and verified using predetermined landmarks (Duvernoy, Reference Duvernoy2005; Pruessner et al., Reference Pruessner, Li, Serles, Pruessner, Collins, Kabani and Evans2000; Willoughby et al., Reference Willoughby, Sheard, Nash and Rovet2008). Each hippocampus included volumes of the dentate gyrus, subicular complex, and the cornu ammonis regions, but not the fimbria or alveus, which served as boundaries; the entorhinal cortex was not measured. This conservative tracing approach was adopted to minimize partial voluming effects.
Hippocampi were traced in the coronal plane in an anterior-to-posterior direction beginning at the rostral end when the head first appeared below the amygdala and terminating at the caudal end when the crura of the fornices separate from the hippocampal tail (Figure 1a). ROIs were further examined in sagittal (Figure 1b) and axial planes to verify superior and inferior boundaries. The anterior region was defined by the emergence of the uncal recess of the hippocampal head in the superomedial region of the hippocampus; the posterior division was defined as the first appearance of ovoid mass of gray matter inferomedial to the trigone of the lateral ventricle (Figure 1c; see also Poppenk et al., Reference Poppenk, Evensmoen, Moscovtch and Nadel2013). The first author (J.D.) traced all hippocampi and delineated their anterior and posterior regions while the second author (J.S.) traced ∼25% of scans. Their inter-rater reliabilities using Cronbach's alpha were 0.91 and 0.84 for left and right hippocampal volumes, respectively. Intracranial volume (ICV), consisting of total gray matter, white matter, and cerebrospinal fluid (CSF), was determined using Christian Gaser's VBM Toolbox v.1.18 for SPM5.
a: Coronal view of hippocampus. b: Right sagittal view of hippocampus; c: Right sagittal view of anterior and posterior hippocampal regions.
Statistical Analyses
Analyses were conducted in SPSS version 21.0 for Macintosh. Regarding demographic indices, categorical variables were analyzed for group differences using χ2 and continuous variables (viz., age, intelligence, ICV, and total hippocampal volume) using t tests. Effects of handedness on hippocampal volumes were determined using the Mann-Whitney U test. To assess effects on memory, we conducted separate group by sex multivariate analyses of variance (MANOVA) for each global test and examined effects on individual subtests via the univariate analyses provided within each MANOVA. Hippocampal measurements were examined for group and sex differences by two methods: (i) repeated measures analysis of variance (ANOVA) with group and sex as between-group factor variables and side as the repeated measure and (ii) group by sex MANOVA on the four individual segments. For all analyses, effect sizes were determined using partial eta-squared. To ascertain the relations between memory and hippocampal size, we performed partial correlations with age removed. To restrict the number of correlations and possibility of a type-1 statistical error, we used only memory indices showing highly significant (p < .005) between-group differences. For all analyses, significance was set at p < .05.
Results
Demographic Measures
Table 2 which presents the demographic data, shows that groups did not differ in age, sex, or handedness. However they did differ in their caregiving environments (p < .001), secondary prenatal drug exposures (p < .001), receipt of medications (p < .001), and comorbid ADHD diagnoses (p < .001). Specifically, the ARND group was more likely than controls to be (a) living in foster or adoptive homes rather than with biological mothers; (b) prenatally exposed to cigarettes, cocaine, or marijuana; (c) currently receiving stimulant medications; and (d) diagnosed with ADHD. Groups also differed in Full Scale IQ [F(1,31) = 43.4; p < .001] with the ARND group scoring below controls. However, we chose not to use IQ as a covariate in light of the Dennis et al. (Reference Dennis, Francis, Cirino, Schachar, Barnes and Fletcher2009) claim that this is unnecessary when IQ is a defining feature of a neurodevelopmental disability. The ARND group also scored below controls on WASI Vocabulary [F(1,31) = 31.7; p < .001] and Matrix Reasoning subtests [F(1,31) = 73.2; p < .001].
Demographic, IQ, and brain volume information for ARND and Control groups
aHandedness information was missing on 1 ARND and 3 controls; bHome care information was missing on 2 ARND; cAll for ADHD; dPresented as Mean (SD)
Memory Test Results
Table 3 presents both groups’ mean scores on the subtests from the various memory tests. MANOVAs performed separately for each instrument revealed significant omnibus group differences on all instruments: CMS [F(10,22) = 3.095; p = .013; η 2 = .585], TOMAL [F(12,20) = 3.184; p = .011; η 2 = .656], CANTAB [F(3,29) = 6.725; p = .001; η 2 = .410], and Rey-O [F(2,32) = 3.61; p = .028; η 2 = .184]. There were no sex differences or group by sex interactions. Univariate analyses contained within each MANOVA showed the ARND group scored significantly below controls on: (i) every CMS subtest, except Dots Learning (effect sizes ranged from .168 for Dots Long Delay to .330 for Dots Short Delay); (ii) every TOMAL subtest except Word and Visual Delayed Selective Reminding (effect sizes ranged from .126 for Visual Selective Reminding to .410 for Visual Sequential Recall); (iii) all CANTAB subtests (effect sizes ranged from .168 for Spatial Span to .241 for Paired Associates Learning); and (iv) both Rey-O subtests (η 2 = .157 and .125 for Copy and Delayed Recall, respectively). These results suggest a broad spectrum of memory deficiencies in children with ARND.
Mean (SD) Memory Test Results for ARND and Control groups
*Effect size reflects partial eta-squared.
ICV and Hippocampal Measurements
As preliminary analyses revealed no effects of handedness or age (range of r-values: −0.058 to −0.128) on hippocampal volumes, we combined left and right-handed participants and did not use age as a covariate in subsequent analyses. However, as we found a marginally significant effect of overall brain size with the Mann-Whitney U test applied (z = −1.72, p = 0.086, See Table 2), we used proportion scores adjusting hippocampal values for ICV in subsequent analyses. Table 4 contains the groups’ mean raw hippocampal measurements while Figure 2 shows the proportion scores.
Mean (SD) hippocampal volume (mm3) measurements for ARND and Control groups
Intracranial volumes (ICV) -adjusted hippocampal volumes (with standard error bars) showing right and left anterior and posterior regions for alcohol-related neurodevelopmental disorder (ARND) males (dark bars) and females (dark striped bars) and control males (light bars) and females (light striped bars). Univariate analyses within multivariate analyses of variance indicated the ARND group differed significantly from controls in right (p = .001) and left (p = .004) posterior segments (smaller in ARND), whereas males had significantly smaller right (p = .001) and left (p = .003) anterior hippocampal segments than females.
To examine for group and sex differences on hippocampal proportions, we first conducted a repeated measures ANOVA with group and sex as between-subjects factors and side as the repeated measure. Results revealed significant main effects of group [F(1, 31) = 12.03; p = 0.002, η 2 = .28] and sex [F(1, 31) = 10.31; p = 0.003, η 2 = .25] but no effect of side or any interactions. Both hippocampi were smaller in the ARND group than controls.
The next analysis involved a group by sex MANOVA on the four segments. Results revealed significant omnibus effects for group [F(4,28) = 3.36; p = .023; η 2 = .325] and sex [F(4,28) = 3.53; p = .019; η 2 = .335] and no group by sex interaction. Univariate analyses (provided within MANOVA) indicated the ARND group had smaller right and left posterior segments [p = .001 and .004, respectively] but not anterior hippocampal segments than controls, as shown in Figure 2. In contrast, the significant sex difference reflected the smaller anterior [p-values: right = .001; left = .003] but not posterior volumes in males than females.
Structure-Function Correlations
Although groups did not differ in age, we chose to partial out the effects of age from the subsequent correlational analyses given previous findings of different age trajectories for structure-function correlations with the hippocampus (Giedd et al., Reference Giedd, Vaituzis, Hamburger, Lange, Rajapakse, Kaysen and Rapoport1996). To limit the possibility of a Type-1 statistical error, we used only memory indices showing a highly significant group difference (p < .005; see Table 3). As there were no lateralized group differences, we combined results across right and left hippocampi or segments in these analyses.
For the ARND group, results revealed significant positive correlations between: (i) CMS Dots Short Delay and total hippocampal size (r = 0.440; p = .007), (ii) CMS Stories Immediate Recall and overall and anterior hippocampal volumes (r = 0.664, p = .004 and r = 0.537, p = .026), and (iii) CMS Stories Delayed Recall and global and anterior hippocampal volumes (r = 0.707; p = .001 and r = 0.616; p = .009). However, a significant negative correlation was also observed between TOMAL Visual Sequential Recall and total hippocampal volume (r = −0.560; p = .019). No significant correlations were observed for controls.
Discussion
The present study sought to examine whether youth with ARND show memory weaknesses and reduced hippocampal size. Our findings revealed that relative to controls, youth with ARND exhibited a broad spectrum of memory difficulties, a marginally reduced brain size, and smaller right and left hippocampi, particularly in posterior subregions. Unlike rodents with prenatal ethanol exposure, however, we did not observe any sexual dimorphisms between ARND and control groups. Instead, females in both groups had larger hippocampi (adjusted for brain size) than did males, especially in their anterior hippocampal segments. ARND and control groups also differed in their patterns of structure/function correlations with better story recall being associated with larger global and anterior hippocampal volumes and better visual sequential recall being associated with smaller global hippocampal volumes, in the ARND group only.
Although current findings showed the ARND group was significantly outperformed by controls on most memory indices, it might be argued that since our ARND group was cognitively impaired, this group by definition would have had memory problems. However, it should be noted that to be assigned an ARND diagnosis using the Canadian Guidelines (Chudley et al., Reference Chudley, Conry, Cook, Loock, Rosales and LeBlanc2005), one needs to show impairments in any three of the multiple domains listed and memory deficits alone are not sufficient for diagnosis. Relevantly, Nash et al. (Reference Nash, Stevens, Rovet, Fantus, Nulman, Sorbara and Koren2013) reported considerable heterogeneity in the neuropsychological profile of children receiving an ARND diagnosis from our clinic while few areas of memory were affected in the clinic sample as a whole.
Group differences in memory varied between tasks as well as across subtests for each instrument. Largest differences were observed when participants had to remember a visual sequence and recall story details after a delay or a dot pattern immediately. Groups did not differ on word and object selective reminding tasks of the TOMAL or learning a dot pattern on the CMS, which are thought to engage the hippocampus (Zimmerman et al., Reference Zimmerman, Pan, Hetherington, Katz, Verghese, Buschke and Lipton2008). Selective reminding results are inconsistent with findings from adults with PAE showing correlations between performance and right hippocampal size (Coles et al., Reference Coles, Goldstein, Lynch, Chen, Kable, Johnson and Hu2011). However, it should be noted that in the latter study, persons with the most facial dysmorphia (i.e., FAS) showed effects and this was not present in our sample. According to Brickman, Stern, and Small (Reference Brickman, Stern and Small2011), selective reminding engages the entorhinal cortex primarily, which we did not include in our conservative tracing approach. Therefore, the spared performance of our ARND group may reflect normal development of this area in youth with ARND. Regarding the Dots task, it is interesting to note that while the ARND group performed adequately in the learning (i.e., encoding) phase, they did show difficulty reconstructing it subsequently.
Current findings showed that youth with ARND had smaller hippocampi bilaterally than did controls, especially in the posterior segment. In contrast, males from both groups showed smaller anterior hippocampal segments than did females. The latter finding is consistent with previous research showing larger bilateral hippocampi in females than males (Neufang et al., Reference Neufang, Specht, Hausmann, Gunturkun, Herpertz-Dahlmann, Fink and Konrad2009) and that age-related volume loss is greater in posterior hippocampal regions in males than females (Gogtay et al., Reference Gogtay, Nugent, Herman, Ordonez, Greenstein, Hayashi and Thompson2006). Although our sample's lack of a sexual dimorphism in hippocampal size is inconsistent with research on ethanol-exposed rodents, other aspects of hippocampal structure and function that may be sexually dimorphic in humans with PAE still need to be examined. Notably, the smaller anterior hippocampal region in males did not contribute to any sex differences in memory performance.
According to Poppenk et al. (Reference Poppenk, Evensmoen, Moscovtch and Nadel2013), the posterior portion of the long axis of the hippocampus, which includes the dentate gyrus, is critical for recall and event reconstruction, whereas the anterior portion includes substructures needed for encoding the event. Our findings from the CMS Dots task are consistent with this notion since we observed groups differed in recalling the pattern but not in learning it initially and they also differed in posterior hippocampal size. In the present study, both right and left hippocampi were similarly affected in the ARND group supporting observations of Astley, Aylward, et al. (Reference Astley, Aylward, Carmichael Olson, Kerns, Brooks, Coggins and Richards2009) on a slightly younger ARND group, but not our own previous findings of a left hippocampal effect (Willoughby et al., Reference Willoughby, Sheard, Nash and Rovet2008) or that of Coles et al. (Reference Coles, Goldstein, Lynch, Chen, Kable, Johnson and Hu2011) showing a right hippocampal effect in adults with PAE. Differences from our earlier study may be explained by several factors: the slightly younger age distribution of our current sample, different distributions of FASD subtypes with none currently having FAS, and our current more conservative tracing approach.
Different structure-function relation patterns were observed between groups with hippocampal size being correlated with selective memory indices only in the ARND group. Our findings of a positive association between story recall with global and anterior hippocampal volumes, and a negative association between visual sequential recall and global hippocampal volumes in the ARND group only, may mean controls used other regions to efficiently recall the story while the ARND group relied heavily on their hippocampi. As we did not study other brain regions currently, we cannot address this issue.
Strengths of our study include: (i) examining children with ARND exclusively, (ii) studying children within a relatively narrow age range, and (iii) subdividing hippocampi into anterior and posterior segments, which is novel within the FASD population. However, several study limitations warrant further discussion. Our sample size was quite small, thus precluding our ability to find meaningful sex dimorphisms or structure-function correlations. Additionally, we were not able to control for mediating factors such as medication usage, comorbidity, family adversity, poor nutrition, stress, any of which could have affected the memory or hippocampal results. However supplementary analyses comparing hippocampal volumes by these factors (e.g., secondary prenatal drug exposures and ADHD comorbidity) failed to show any meaningful effects on hippocampal size (data not shown). Also our conservative tracing approach did not allow us to examine for regions such as the fimbria or parts of subiculum and well as cortical regions such as the entorhinal cortex, which may have been important for some of the memory indices on which we did obtain effects. Finally, current results were based solely on volumes and so did not examine other aspects of hippocampal integrity (viz., shape, contour), which may be sensitive to effects of PAE or to sex differences. Thus, future hippocampal studies using higher resolution MRI to allow for these finer analyses are warranted.
Overall, current findings showed reduced posterior hippocampal volumes and selective memory deficits in youth with ARND. These findings have implications for treating the memory difficulties of children with ARND that can impact on their school functioning, everyday memory functions (Agnihotri, Sheard, Keightley, & Rovet, Reference Agnihotri, Sheard, Keightley and Rovet2012), and social relationships. Since techniques to facilitate recall and maintain newly learned events in memory, as well as to improve hippocampal integrity (e.g., exercise, music training, memory games), would also be beneficial for this population, future studies need to address these possibilities.
Acknowledgments
This research was conducted as part of a larger investigation examining cognitive abilities and brain imaging in FASD funded to JR and ES by Canadian Institutes of Health Research (MOP #3203919) and the Canadian Foundation for Fetal Alcohol Research, as well as RESTRACOMP and Ontario Mental Health Foundation postdoctoral fellowships to ES. The authors wish to thank Anishka Leis for her work in recruitment and coordinating the project; Sarah Wheeler, Karen Willoughby, Victoria Martin, Rosie Bell, and Dragana Ostojic for participation in various stages of the project; Susan Blaser for examining the MRI scans; Ruth Weiss, Tammy Rayner, Garry Detzler, and Lydia Sproule from Diagnostic Imaging for the scanning; and the Motherisk Clinic team for the original FASD diagnoses. The authors are especially grateful to the parents/caregivers and their children for their participation in this study. Finally, we are appreciative of the thoughtful and insightful critiques provided by Mary Pat McAndrews and the reviewers of this study. There are no conflicts of interest regarding this work.
