| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Psychiatry (T.H.), Molecular Medicine and Surgery (T.H., A.N., A.W., S.L.), Clinical Sciences (A.N.), Public Health Sciences/National Institute for Psychosocial Medicine (F.L.), and Woman and Child Health (E.M.R.), Karolinska Institutet, 171 76 Stockholm, Sweden; and the Department of Psychology (T.L.), Stockholm University, 106 91 Stockholm, Sweden
Address all correspondence and requests for reprints to: Dr. Svetlana Lajic, Department of Molecular Medicine and Surgery, Center for Molecular Medicine L8:02, Karolinska Institutet/Karolinska University Hospital, 171 76 Stockholm, Sweden. E-mail: Svetlana.Lajic{at}ki.se.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Design and Patients: Prenatally treated children, 7 to 17 yr old, were assessed with standardized neuropsychological tests (A Developmental Neuropsychological Assessment and Wechsler Intelligence Scales for Children) and child-completed questionnaires measuring self-perceived scholastic competence (Self-Perception Profile for Children). A parent-completed questionnaire (Child Behavior Checklist/4–18 School Scale) was used to evaluate whether the treatment had any impact on the children’s school performance. In addition, a child-completed questionnaire measuring social anxiety (The Social Anxiety Scale for Children–Revised) was completed by the prenatally treated children aged 8 to 17 yr (n = 21) and age- and sex-matched controls (n = 26).
Results: Of 40 DEX-treated children, 26 (median age, 11 yr) participated in the study. Thirty-five sex- and age-matched healthy children were controls. There were no between-group differences concerning psychometric intelligence, measures of cerebral lateralization, memory encoding, and long-term memory. Short-term treated, CAH-unaffected children performed poorer than the control group on a test assessing verbal working memory (P = 0.003), and they rated lower on a questionnaire assessing self-perception of scholastic competence (P = 0.003). This group also showed increased self-rated social anxiety assessed by The Social Anxiety Scale for Children–Revised (P = 0.026). Prenatally treated, CAH-affected children performed poorer than controls on tests measuring verbal processing speed, although this difference disappeared when controlling for the child’s full-scale IQ.
Conclusions: This study indicates that prenatal DEX treatment is associated with previously not described long-term effects on verbal working memory and on certain aspects of self-perception that could be related to poorer verbal working memory. These findings may thus question future DEX treatment of congenital adrenal hyperplasia. Therefore, we encourage additional retrospective studies of larger cohorts to either confirm or challenge the present findings.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
However, experimental data from animal studies and observations on adverse medical events in human newborns have raised concerns about the safety of the treatment and the impact it may have on fetal programming (5, 6, 7). Furthermore, to be fully effective, the treatment needs to be initiated at the sixth to seventh week of gestation, i.e. several weeks before prenatal diagnosis is possible by a chorionic villous biopsy performed in gestational wk 10. Thus, a majority of the treated fetuses (seven of eight, all unaffected children and CAH-affected boys) will receive DEX unnecessarily (8). Potential harmful effects affecting adult metabolic and neuropsychological status must therefore be evaluated in this context.
In follow-up studies of prenatally treated children, pre- and postnatal growth has been normal (2, 9, 10, 11). However, several adverse events such as mental retardation, corpus callosum agenesia, hydrocephalus, and failure to thrive do not show any obvious causal relationship, but they are reported in prenatally treated children (9, 11, 12, 13). Few follow-up studies on cognitive development have been performed, and none has been based on direct observations of treated children. A pilot study based on reports from maternal questionnaires and comprising 26 long- and short-term treated children suggested that DEX-exposed children showed more shyness and inhibition (14). However, an extended questionnaire study performed by the same group on 174 treated children could not document any significant negative effects on developmental outcome (15).
Experimental data from animals exposed to perinatal corticosteroids have shown adverse effects on both somatic development and cognitive function. In rats, reduced birth weight in rat offspring as well as development of hypertension in adult life (16, 17) and impaired short-term memory (5) have been reported. In rhesus monkeys, a single high dose of DEX in late gestation resulted in altered hippocampal architecture with depletion of hippocampal pyramidal and dentate granular neurons. Moreover, multiple low-dose injections of DEX caused even more severe damage than one single high dose (18). In summary, the doses used in animal experiments have been high in most cases, and there is little information concerning the effects of low-dose prenatal DEX treatment as used in CAH.
Because the present study is the first to assess long-term cognitive development based on direct examination of children treated prenatally with DEX, the study was explorative with respect to the number and variation of neuropsychological tests that were used. Based on the described animal studies, memory functions were our primary focus regarding cognition. Because test performance may be influenced by social anxiety and as a result of the findings in Trautman et al. (14), we also investigated the children’s subjective experience of social anxiety.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In Sweden, between the years 1985 and 1995, 40 fetuses at risk of being affected with severe CAH were treated from postmenstrual wk 6 to 7 with DEX to prevent prenatal virilization of affected females (11). The DEX dose offered to mothers with a previous child with severe CAH was 20 µg/kg per day in three divided doses based on prepregnancy weight and maximum 1.5 mg daily.
The final cohort comprised 26 children (12 boys, 14 girls; refusal rate 35%) between 7 and 17 yr (mean ± SD, 10.95 ± 2.33 yr). Based on the results of mutation analysis in chorionic villous biopsies at wk 10 to 12, treatment was either continued to term (four CAH-affected females) or stopped at the end of the first trimester (seven unaffected boys and 10 girls; five CAH-affected boys). One CAH-affected girl was not able to accomplish the neuropsychological tests as a result of low intellectual performance, but parental questionnaires were obtained for this child. One CAH-affected boy had diabetes mellitus since 2 yr of age; one unaffected girl had rheumatoid arthritis; and another unaffected, short-term treated girl was under investigation for proteinuria. The composition of the cohort that participated in the study did not differ from those children/families that denied participation with respect to behavioral problems of the child, psychosocial problems in the family, maternal side effects induced by DEX, and maternal attitude toward future treatment as assessed by questionnaires to the managing pediatrician and/or the mother. In this retrospective study, families with a DEX-treated child were contacted by means of an invitation letter. None of the families were initially treated within the frames of a clinical study, and this fact is most probably the major reason for the high dropout rate.
Two control groups were recruited. First, siblings of children treated with GH (with presumed similar psychosocial prerequisites as the study group) were included: 10 (four boys and six girls) of 28 children (refusal rate 64%) between 7 and 12.5 yr (9.94 ± 2.02). Second, to create a large enough control group, the Swedish National Registry was used for randomly recruiting 25 children matched for age and sex (11 boys and 14 girls, refusal rate 69%) between 7 and 17 yr (10.55 ± 2.46).
Among the DEX-treated children, 54% lived in cities in comparison with 97% of the controls (
2 test, P < 0.001). There were no significant differences in parental socioeconomic status as estimated by parental educational level (2 test, P < 0.162) among the two groups. Moreover, the treated vs. nontreated children were comparable regarding birth length (P = 0.234), birth weight (P = 0.704), and gestational length (P = 0.942).
Study design
The same clinical psychologist performed the standardized neuropsychological assessments for all children. The questionnaires were filled out by the children and their parents when visiting the clinical psychologist. Thus, both parents and children had the opportunity to ask for clarification of items (19). Parents and children filled out the questionnaires independently of each other. All families gave written informed consent, and the study was approved by the Regional Ethics Committee of Stockholm.
Instruments
Neuropsychological tests assessing cognitive functions. Well-standardized and commonly used neuropsychological tests were chosen to ensure good psychometric properties of the tests. The short version of Wechsler Intelligence Scales for Children (20) was used for estimation of psychometric intelligence or full-scale IQ (FSIQ) and the four-factor index scores—verbal comprehension, perceptual organization, freedom from distractibility, and processing speed. Handedness was measured using the Manual Preference test (21). Subtests of A Developmental Neuropsychological Assessment (NEPSY) (22) were used to assess memory for a narrative as well as immediate and delayed memory for names, faces, and a list of words. In addition, a test of immediate and delayed spatial memory (23) was administered. The Arithmetic and Digit Span subtest comprising the Freedom from Distractibility Index (20) was used to measure verbal working memory and a Span-board test (24) for assessment of visuospatial working memory. The Stroop Interference test (25) was given to children who were fluent readers to evaluate the ability to inhibit an overlearned response. Two additional subtests of Stroop were used to investigate verbal processing speed (reading speed and speeded naming of colors). Moreover, Processing Speed Index from the short version of Wechsler Intelligence Scales for Children (20) was used for assessing nonverbal processing speed.
Child-completed questionnaires. The children estimated their scholastic ability by completing the Scholastic Competence subscale from the Self-Perception Profile for Children (26). The questionnaire consists of school-related items tapping the child’s perception of his or her competence within the realm of scholastic performance. The psychometric properties of the scale are considered to be acceptable according to the manual (26).
The Social Anxiety Scale for Children–Revised (SASC-R) (27, 28) was used to assess the children’s subjective experience of social anxiety and the behavioral consequences of the anxiety, i.e. avoidance or inhibition. Three separate factors have been identified, and these factors were used to study: 1) Fear of Negative Evaluation (SASC-FNE)—fear of negative evaluation from peers; 2) Social Avoidance and Distress–New (SASC-New)—avoidance and distress with new situations or unfamiliar peers; and 3) Social Avoidance and Distress-General (SASC-General)—generalized social distress and inhibition (28).
Parent-completed questionnaire evaluating children’s scholastic performance. Parents evaluated academic performance, need for special class or any type of remedial special class, whether the child had repeated a grade as well as any other type of major school problems of the children by completing the School Scale of Child Behavior Checklist for Ages 4–18 (19), a frequently used screening questionnaire. In Sweden, grades are not given at school until 14 yr of age. Thus, school grades could not be used to estimate school performance.
Statistical analysis
Data on all measures were initially analyzed by a series of one-way ANOVA for continuous scales and 2 tests for binary scales using three groups (DEX-treated children and the two different control groups) as independent variables. Similar results were obtained for the two control groups on all measures, and therefore data for the two groups were combined. Subsequent comparisons between DEX-exposed and -unexposed children were performed with t tests, whereas categorical data were analyzed by 2 tests. Although only negative effects, if any, were expected for the DEX-treated group, two-tailed statistical tests were chosen to reduce the risk of type I errors. The statistically significant results obtained from t tests were also analyzed by one-way analysis of covariance with the short version of Wechsler Intelligence Scales for Children FSIQ entered as the covariate to exclude the impact of the general cognitive ability on specific cognitive functions (29). Because all other variables were normally distributed, the Freedom from Distractibility Index was the only variable that was log-transformed to correct for non-Gaussian distribution. The results in this parameter are reported for both the corrected and noncorrected variable.
Because CAH-affected children have a different prenatal hormonal milieu and continue to receive GC after birth, statistically significant findings were reevaluated using one-way ANOVA with pairwise post hoc comparisons of the three groups (CAH-affected, CAH-unaffected, and control children). Furthermore, to assess whether the effects of treatment were sex-specific (30), the significant findings were also analyzed separately for males and females. Moreover, because many comparisons were performed on a relatively small sample size, the significant findings were Bonferroni-corrected to avoid type I errors. Only children above 7 yr of age filled in the self-rating scale SASC-R, and data were therefore not analyzed separately for boys and girls in this part of the study as a result of small subgroups. The impact of social anxiety on test performance and scholastic ability was excluded by one-way analyses of covariance on statistically significant results with SASC-Total entered as the covariate.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Psychometric intelligence.
DEX-exposed children and controls performed equally in Verbal Comprehension and Perceptual Organization Indexes (P > 0.10) (Table 1
). In addition, a comparison of the mean values of FSIQ did not show any statistically significant difference between the treated vs. nontreated groups (P = 0.08) (Table 1) or among the three subgroups (CAH-unaffected, CAH-affected, and controls) using pairwise post hoc comparisons (data not shown).
|
2 test, P = 0.258).
Learning and memory.
Measures for immediate (memory encoding) and long-term memory could not identify any differences between DEX-exposed and DEX-unexposed children (Table 1
). There were ceiling effects in both groups in two of five tests measuring learning and memory, i.e., in visuospatial tests assessing visuospatial learning and long-term memory as well as in the test measuring long-term memory for names.
Working memory.
The Freedom from Distractibility Index was log-transformed to correct for non-Gaussian distribution. There was a difference between the treated vs. nontreated groups both before (P = 0.009) and after (P = 0.007) transformation. However, the difference was specifically found in the subtest Digit Span (P = 0.001), a test measuring verbal working memory (Table 1). When analysis of covariance was performed with FSIQ entered as a covariate, the effect on verbal working memory remained significant (P = 0.002). Pairwise post hoc comparisons showed that both CAH-affected (P = 0.025) and CAH-unaffected (P = 0.001) treated children had a lower performance compared with controls. After Bonferroni corrections for multiple comparisons, the difference remained significant only for the CAH-unaffected, short-term treated group (P = 0.003). The results obtained for the parameter Digit Span were subsequently analyzed separately for females and males. DEX-exposed girls (n = 13) performed poorer than control girls (n = 20) (P = 0.003), even when CAH-affected girls (n = 3) were excluded (P = 0.005). Likewise, DEX-exposed boys (n = 12) performed poorer than control boys (n = 15) (P = 0.019), although the difference dropped to below significance when CAH-affected boys (n = 5) were excluded (P = 0.073).
DEX-exposed children performed equally as well as controls in tests assessing visuospatial working memory (Span-board Test), although the results were approaching significance (P = 0.083).
Impulse inhibition.
There were no differences between DEX-exposed and -unexposed children in the test of impulse inhibition (Stroop Interference) (Table 1).
Speed of processing.
There were no between-group differences in the Processing Speed Index (P = 0.110). However, DEX-exposed children performed poorer than controls in the Stroop subtests assessing verbal processing speed (speeded reading of words, P = 0.043; speeded naming of colors, P = 0.016) (Table 1). This effect disappeared when FSIQ was entered as a covariate in an analysis of covariance, although there was still a tendency toward significance in the test assessing speeded naming of colors (P = 0.069) (Table 1).
Child-completed questionnaires
Most of the children above 7 yr of age completed the Scholastic Competence questionnaire (from Self-Perception Profile for Children) and SASC-R (CAH-unaffected, n = 13; CAH-affected, n = 8; control, n = 26).
Scholastic ability.
The comparison of the children’s ratings on the Scholastic Competence subscale revealed a significant difference between the groups (P = 0.007) (Table 2). The difference was observed only for the item in which the DEX-treated children reported that they "have trouble figuring out the answers at school" (P = 0.001). This result was independent of the psychometric intelligence (analysis of covariance, P = 0.003). Subsequent pairwise post hoc comparisons showed that the CAH-unaffected, short-term treated children scored lower than the controls in both the total subscale (P = 0.001) and the previously mentioned item (P = 0.001). This difference remained significant after Bonferroni correction (P = 0.003). The CAH-affected children did not differ significantly from the controls on these ratings (P = 0.344).
|
SASC-New) when CAH-affected, CAH-unaffected, and control children were compared separately (ANOVA with post hoc comparison of the three groups). However, the analysis also showed that the CAH-unaffected, short-term treated children rated themselves as more socially anxious than the controls for all SASC-R subscales (data summarized in Table 3), although when correcting for multiple comparisons, this difference reached statistical significance only for the SASC-General subscale (Table 3). Because social anxiety may affect both test performance and scholastic ability, we wanted to exclude that our findings on working memory, assessed by the test Digit Span and the Scholastic Competence questionnaire, were not the result of increased social anxiety in the DEX-treated group. One-way analyses of covariance with SASC-Total entered as the covariate showed that the differences between the groups remained significant in the Digit Span test (P = 0.002) and the Scholastic Competence self-rating questionnaire (P = 0.023).
|
Among the DEX-treated group, three children (11.5%) were in a special class/special school or had repeated a class, which was comparable to two children (6%) among the controls (2 test, P = 0.412). Furthermore, four DEX-exposed children (15%) were described to have had "any academic or other problems in school" by their parents. An analogous result was revealed in the control group in which five children (14%) have had those types of problems (2 test, P = 0.905).
A series of t tests were computed on parental ratings of the four academic subjects that are scored by parents in Child Behavior Checklist (Swedish version), i.e. reading (P = 0.097), writing (P = 0.418), spelling (P = 0.822), and mathematics (P = 0.023). The significant difference between the groups on mathematics (DEX-treated children having poorer results) dropped to below significance when controlled for the FSIQ in analysis of covariance (P = 0.224).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Working memory is the ability to retain information for a mental operation during a short period of time (31). According to the model of Baddeley and Hitch (32), working memory has a domain-specific organization with verbal and nonverbal substores that are regulated by the central executive, a third entity that operates over the domain specific substores. Our findings on verbal working memory and self-perception of scholastic competence had a large effect size (d = 0.95 and d = 0.82, respectively), but major cognitive measures such as IQ, learning, and long-term memory were not affected in the DEX-exposed group. Learning and long-term memory are cognitive functions mediated by neural networks that include hippocampal structures. Such GC effects on hippocampal functions have been observed in rats (33), but it is also known that the profile of brain development is species-specific and that sensitivity and expression of corticosteroid receptors varies between different time points in different species (30, 34). Also in humans, we have to consider timing effects, i.e., early prenatal treatment may give different effects compared with treatment during the entire gestational period. In addition, differences in doses of GC administered to CAH-affected children during the neonatal period may also be of importance. GC administration to neonates that are at risk of developing lung disease of prematurity leads to much more substantial effects on neuromotor and cognitive functions (35) than effects observed by us and others after prenatal treatment.
The observed effects on verbal working memory/short-term memory would suggest that prenatal DEX treatment affects the frontostriatal loop, but the present study does not elucidate whether the effects are functional and/or structural. The CAH-affected group is also relatively small and heterogeneous consisting of full-term DEX-treated girls and short-term DEX-treated boys. The fact that none of the families were initially treated within the frames of a clinical study is most probably the major reason to the high dropout rate. Prospective studies such as PREDEX (4) that include magnetic resonance imaging of treated children may give us further information on this issue. Future studies should also include a control group consisting of non-DEX-exposed CAH children.
In our study group, there was a nonsignificant tendency toward higher FSIQ in the control group probably as a result of a higher refusal rate. Although this difference was not statistically significant, we wanted to exclude the possibility that the poorer working memory performance observed in the DEX-treated group was an effect of a lower IQ (29, 36). An analysis of covariance with FSIQ entered as a covariate showed that the observed effect on verbal working memory for the DEX-treated children still remained significant. Another difference that was observed was that CAH-affected, prenatally treated children performed less well in tests measuring verbal processing speed, but this difference did not reach significance when controlling for FSIQ (data summarized in Table 1). Both these observations suggest that prenatal DEX treatment in early pregnancy may affect specific verbal cognitive functions.
The question of family selection bias has been approached. The only family background variable that differed between the groups was that more families with a treated child were living in rural areas. However, this type of living does not imply any effects on cognitive functions (37, 38). The prevalence of child behavioral problems and mental disorders is similar in rural and urban areas (36) or lower in rural areas (39, 40). Rural–urban variation in childhood psychopathology is associated with economic and cultural differences and not with urbanization per se (41). Thus, living in rural areas should not be a risk factor for enhanced social anxiety in our sample.
To investigate if the increased social anxiety in CAH-unaffected, short-term treated children had any negative effect on the test assessing verbal working memory (Digit Span) or on children’s perception of their scholastic ability (Scholastic Competence questionnaire), we performed analyses of covariance with SASC-Total as the covariate. Increased social anxiety did not have any impact on the impaired verbal working memory or on the scholastic ability in the DEX-treated group, because both results remained significant.
In accordance with previous studies (14, 15), no differences in school performance were observed using parental questionnaires as an assessment tool. The clinical significance of these results cannot be fully evaluated within the frames of the present study. Thus, neither direct testing of treated children nor parental questionnaires revealed any differences on measures assessing learning and long-term memory, i.e., cognitive functions mediated by neural networks that include hippocampal structures. Hippocampal structures as well as amygdala are also important for the regulation and expression of fear and anxiety. In rats, prenatal stress and GC exposure increases the levels of corticotrophin releasing hormone in amygdala as well as the levels of GC receptors and/or mineralocorticoid receptors (33, 42, 43, 44, 45). Furthermore, brain-specific knockout of GC receptors in mice markedly reduces the level of anxiety (46). The tendency of the DEX-exposed children toward more shyness in parental ratings (14) was not confirmed with the extended study by the same group (15). In the present study, we observed increased self-reported social anxiety in the CAH-unaffected, short-term treated group. This finding should be interpreted cautiously as a result of small sample size.
In conclusion, this study indicates that prenatal DEX treatment has previously not described long-term effects on verbal working memory and on aspects of self-perception. These facts open up an ethical dilemma regarding the future use of this treatment, especially because it also affects CAH-unaffected siblings who do not benefit from the treatment. We find it imperative that all future prenatal treatment of CAH will be performed within large, multicenter clinical trials to obtain more data on metabolic and neuropsychological effects. The ongoing trial PREDEX (4) is one example in which these questions are being addressed in depth in a prospective manner. Nevertheless, we find it unsatisfactory to only await the results of such long-term studies. Therefore, we urge the scientific society to perform additional retrospective studies of larger cohorts of treated children and young adults including direct neuropsychological testing to either confirm or contradict the present findings. Until then, it is important that parents are thoroughly informed about the potential risks and uncertainties as well as the benefits of this treatment (47).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure Statement: The authors have nothing to disclose.
First Published Online December 5, 2006
Abbreviations: CAH, Congenital adrenal hyperplasia; DEX, dexamethasone; FSIQ, full-scale IQ; GC, glucocorticoid(s); SASC-R, The Social Anxiety Scale for Children–Revised.
Received June 21, 2006.
Accepted November 27, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Read all eLetters
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
