Articles Long-term cognitive and cardiac outcomes after prenatal
Transcript
Articles Long-term cognitive and cardiac outcomes after prenatal
Articles Long-term cognitive and cardiac outcomes after prenatal exposure to chemotherapy in children aged 18 months or older: an observational study Frédéric Amant, Kristel Van Calsteren, Michael J Halaska, Mina Mhallem Gziri, Wei Hui, Lieven Lagae, Michèl A Willemsen, Livia Kapusta, Ben Van Calster, Heidi Wouters, Liesbeth Heyns, Sileny N Han, Viktor Tomek, Luc Mertens, Petronella B Ottevanger Summary Background Chemotherapy for the treatment of maternal cancers during pregnancy has become more acceptable in the past decade; however, the effect of prenatal exposure to chemotherapy on cardiac and neurodevelopmental outcomes of the offspring is still uncertain. We aimed to record the general health, cardiac function, and neurodevelopmental outcomes of children who were prenatally exposed to chemotherapy. Methods We did an interim analysis of a multicentre observational cohort study assessing children who were prenatally exposed to maternal cancer staging and treatment, including chemotherapy. We assessed children at birth, at age 18 months, and at age 5–6, 8–9, 11–12, 14–15, or 18 years. We did clinical neurological examinations, tests of the general level of cognitive functioning (Bayley or intelligence quotient [IQ] test), electrocardiography and echocardiography, and administered a questionnaire on general health and development. From age 5 years, we also did audiometry, the Auditory Verbal Learning Test, and subtasks of the Children’s Memory Scale, and the Test of Everyday Attention for Children, and we also completed the Child Behavior Checklist. This study is registered with ClinicalTrials.gov, number NCT00330447. Findings 236 cycles of chemotherapy were administered in 68 pregnancies. We assessed 70 children, born at a median gestational age of 35·7 weeks (range 28·3–41·0; IQR 3·3; 47 women at <37 weeks), with a median follow-up period of 22·3 months (range 16·8–211·6; IQR 54·9). Although neurocognitive outcomes were within normal ranges, cognitive development scores were lower for children who were born preterm than for those born at full term. When controlling for age, sex, and country, the score for IQ increased by an average 11·6 points (95% CI 6·0–17·1) for each additional month of gestation (p<0·0001). Our measurements of the children’s behaviour, general health, hearing, and growth corresponded with those of the general population. Cardiac dimensions and functions were within normal ranges. We identified a severe neurodevelopmental delay in both members of one twin pregnancy. Interpretation Fetal exposure to chemotherapy was not associated with increased CNS, cardiac or auditory morbidity, or with impairments to general health and growth compared with the general population. However, subtle changes in cardiac and neurocognitive measurements emphasise the need for longer follow-up. Prematurity was common and was associated with impaired cognitive development. Therefore, iatrogenic preterm delivery should be avoided when possible. Funding Research Foundation-Flanders; Research Fund-K U Leuven; Agency for Innovation by Science and Technology; Stichting tegen Kanker; Clinical Research Fund-University Hospitals Leuven; and Belgian Cancer Plan, Ministery of Health. Introduction The unintended use of urethane to treat chronic myeloid leukaemia in a pregnant woman in 1948 was one of the first reports of chemotherapy use during pregnancy.1 Since then, more experience has been gained and chemotherapy is now regularly given after the first trimester if cancer treatment is needed during pregnancy.2,3 Investigators estimate that cancer complicates one in 1000–2000 pregnancies and the incidence increases by 2·5% per calendar year.4–6 However, the effect of the malignancy and its treatment on fetal health remains a serious concern. Chemotherapy during the first trimester increases the risk of congenital malformations, whereas the effects of chemotherapy on the fetus beyond the first trimester could potentially affect the development of the brain and heart. Of concern is the potential effect on the front of the brain (attention, memory, and executive functions), because these are most affected by cytotoxic treatment in adults and children.7 These cognitive functions are also the most vulnerable in infants at risk, such as preterm neonates, children with periventricular leucomalacia, and children in utero exposed to toxic products such as cocaine, tobacco, or alcohol, or to maternal emotional disturbance and distress.8–10 A second concern is the potential cardiotoxic effect of anthracyclines, commonly used in the treatment of breast cancer and haematological malignancies—the most common cancer types during pregnancy.11 www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 Published Online February 10, 2012 DOI:10.1016/S14702045(11)70363-1 See Online/Comment DOI:10.1016/S14702045(11)70408-9 See Series Lancet 2012; 379: 558, 570, and 580 Leuven Cancer Institute, Gynaecologic Oncology (F Amant PhD, B Van Calster PhD, L Heyns MSc, S N Han MD), Department of Obstetrics and Gynaecology (K Van Calsteren PhD, M Mhallem Gziri MD, B Van Calster), and Department of Paediatrics (L Lagae PhD, H Wouters MSc), University Hospitals Leuven, and Division of Imaging and Cardiovascular Dynamics (L Mertens PhD), Katholieke Universiteit Leuven, Belgium; Department of Obstetrics and Gynaecology, Second Medical Faculty, Charles University, Prague, Czech Republic (M J Halaska PhD); Department of Paediatric Neurology, Donders Institute for Brain, Cognition and Behaviour (M A Willemsen PhD), Department of Paediatric Cardiology (L Kapusta PhD), and Department of Medical Oncology (P B Ottevanger PhD), Radboud University Nijmegen Medical Centre, Nijmegen, Netherlands; Edit Wolfson Medical Centre, Holon, Israel (L Kapusta); Paediatric Cardiocentre, Faculty Hospital Motol, Prague, Czech Republic (V Tomek MD); and Division of Cardiology, The Hospital for Sick Children, Toronto, ON, Canada (W Hui MD, L Mertens) Correspondence to: Dr Frédéric Amant, Leuven Cancer Institute, Gynaecologic Oncology, University Hospitals Leuven, Katholieke Universiteit Leuven, Herestraat 49, Leuven 3000, Belgium [email protected] 1 Articles So far, solely retrospective and restricted data exist on the long-term outcome of children exposed to chemotherapy in utero.12–14 We aimed to assess the general health, cardiac function, and neurodevelopmental outcomes, including intelligence, memory, attention, and executive functions, in children who were prenatally exposed to chemotherapy. Methods Participants See Online for appendix This multicentre cohort study was started on March 18, 2005 in three European countries and on the basis of a collaboration between national referral centres in Belgium (University Hospitals Leuven), the Netherlands (Radboud University Nijmegen Medical Centre), and the Czech Republic (Faculty Hospital Motol, Charles University, Prague). The study is still recruiting. Some children were enrolled and followed up prospectively; however, others were identified retrospectively (ie, were exposed to chemotherapy before 2005), but were followed up prospectively. We included children who were prenatally exposed to cytotoxic drugs for cancer treatment, but excluded children who were exposed to low-dose chemotherapy for skin or inflammatory disorders. Further details of how children were identified are given in the appendix. We recorded long-term toxic effects, which we defined as those occurring after a minimum follow-up of 18 months, secondary to in-utero exposure to chemotherapy. A standardised study protocol was used in the three participating centres. Written parental permission to participate in this study was obtained for each child. The research protocol was approved by the institutional review board of all participating centres. 114 children born after prenatal exposure to chemotherapy 31 retrospective (born between 1991 and 2004) 83 prospective (born aften 2004) 29 excluded 23 younger than 18 months 6 with incomplete medical records 85 eligible 15 excluded 4 lost to follow-up 1 case of sudden infant death syndrome at 2 months 8 parents refused to participate* 2 did not speak the national language 70 included in long-term follow-up study 27 retrospective 43 prospective Figure 1: Study profile *Three parents did not want to be confronted with the hospital environment because of emotional distress, one did not want to burden their child with tests, one did not want to come to the referral hospital, and three did not want to participate in our study. 2 Procedures We recorded maternal disease, staging examinations, and all treatments given in pregnancy. We organised a standardised assessment of the general health and development, and cardiological, cognitive, behavioural, and neurological development of the children, including assessments at birth, at age 18 months (the age range, at first assessment, we allowed in this cohort was 17–29 months), and age 5–6, 8–9, 11–12, 14–15, or 18 years.15–20 We also included children from historical datasets from the three centres and, when they entered the study at an age between our predefined ages, their developmental milestones were assessed by a paediatric neurologist (LL or MAW). When these children reached our predefined ages, we assessed them in accordance with our protocol. All children born after prenatal exposure to chemotherapy (both retrospectively and prospectively) were born in hospital and assessed by a neonatologist. For the prospective group we asked the neonatologists to complete a datasheet (general physical examination and neurological assessment: tonus, reflexes, active and passive movements, eye movements). Since these are standard examinations of neonates, this information was available in the medical files of all patients we included retrospectively. From age 18 months onwards we did assessments at the national study centre. At each visit, biometric data and a questionnaire on general health status, school performance, and recreation and social situation was systematically collected from the parents (appendix). For the cognitive assessment, we developed an age-adapted test battery for the assessment of intelligence, verbal and non-verbal memory, attention, working memory, and executive functions (appendix). The Child Behavior Checklist,20 a questionnaire that screens for behavioural and emotional problems, was completed by the parents while their child was tested. From age 5 years onwards we did an audiometry test once. Tests and questionnaires were completed in the children’s native languages. Cardiac assessment consisted of a 12-lead electrocardiograph (ECG) and a full echocardiographical assessment for structural and functional characteristics (appendix). Statistical analysis We compared our findings with available norms: national data for height, weight, head circumference, by sex,21 and national and international reference data for neurodevelopmental tests.15–20 Some children were assessed twice; we ensured there were no important differences between the two sets of results and used the most recent data. We converted all raw scores from the cognitive tests to standardised scores with published normative data for the specific age group as provided by the respective tests. For the Bayley and Wechsler intelligence tests, 100 plus or minus 15 is thought the normal range of index scores. For the data analysis and www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 Articles representation of the Test of Everyday Attention for Children, the Children’s Memory Scale, and the Auditory Verbal Learning Test, we converted all scores to Z scores, with a normal range defined as a Z score of 0 (SD 1). We assessed the relation between intelligence quotient (IQ) scores and gestational age at birth by linear (ordinary least squares) regression. We estimated the effect size with the ω² measure of explained variance. We added age and sex as covariates, together with a random effect for country. We analysed and compared electrocardiographic measurements with normal values in childhood and adolescence.22 All echocardiographic measurements were obtained in three cardiac cycles and averaged. We compared all cardiac measurements with the measurements in a matched control group (1:1; from Toronto and Leuven) with age (plus or minus 12 months) and sex as matching factors. We compared our cardiac measurements with reference values from a historical dataset and represented these as Z scores. To circumvent violations of the linear regression assumptions, mainly the assumption of normality of the residuals due to skewed distributions for some cardiac variables, we estimated the independent effect of chemotherapy with median regression. The matching was accounted for by stratifying the regression models by age (six strata) and adding sex as a covariate.23 When both age and sex were used as covariates, as a sensitivity analysis, we obtained similar results. We corrected for multiple testing with Holm’s method for all cardiac measurements. Analyses were done with SAS (version 9.2) and Statistical Package Number receiving electrocardiogram and echocardiogram (cardiological assessment; n=65)* for Social Sciences for Windows (version 16). This study is registered with ClinicalTrials.gov, number NCT00330447. Role of the funding source The sponsors of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. Results This interim analysis was done with a data cutoff at March 1, 2011. Figure 1 shows the study profile. 32 girls and 38 boys were exposed to chemotherapy in utero (236 cycles) during 68 pregnancies (two twin pregnancies). Distribution of children by country was 42 in Belgium, 20 in the Netherlands, and eight in the Czech Republic. Maternal disease, staging examinations, and drugs administered as supportive care are shown in the appendix. During pregnancy, 34 women received chemotherapy, one received chemotherapy and radiotherapy, 27 received chemotherapy and surgery, and six received all three treatment modalities. 19 different regimens of chemotherapy were administered (appendix) and anthracyclines were the most common agents (53 patients). The median cumulative dose for doxorubicin (n=33) was 180 mg/m² (range 50–400; IQR 120), for epirubicin (n=14) was 450 mg/m² (200–600; 240), for idarubicin (n=3) was 72 mg/m² (36–108; 36), and for daunorubicin (n=4) was 240 mg/m² (120–360; 60). Number receiving paediatric examination (n=70) plus questionnaire for parents (n=57; general health assessment)† Neurological examination (n=70) Neonatal 21 70‡ Clinical neurological examination 18 months 40§ 40 Clinical neurological examination and Mental Development Index of the Bayley Scales of Infant Development15 5–6 years 15 16 Clinical neurological examination, audiometry, and age-adapted test battery¶ 8–9 years 11|| 13 Clinical neurological examination and age-adapted test battery** 11–12 years 2 2 Clinical neurological examination and age-adapted test battery** 14–15 years 1 1 Clinical neurological examination and age-adapted test battery** 18 years 2 2 Clinical neurological examination and age-adapted test battery†† Inclusion between standardised ages‡‡ 6§§ 6 Clinical neurological examination Only children in the long-term follow up are represented. *One child tested at age 18 months and 6 years, five at age 6 and 9 years, one at age 9 and 12 years, and one at age 15 and 18 years. †One child tested at age 18 months and 6 years, five at age 6 and 9 years, two at age 4 (between standardised ages) and 9 years, one at age 9 and 12 years, and one at age 15 and 18 years. ‡Parents were not asked to complete questionnaire since it is not appropriate for neonates. §In six children not all echocardiographic data were obtained because of the child’s movements; in three of these children echocardiography was repeated after 1 or 2 years to complete the data set. Two of the 40 children were tested at age 1 year, the examination was not repeated 6 months later since it was within normal ranges. ¶For this age group the age-adapted test battery comprised the Wechsler Preschool and Primary Scale of Intelligence (one child was tested with the Snijders-Oomen non-verbal intelligence test because their native language was not the national language), subtasks from the Children’s Memory Scale (Numbers for verbal memory; Dot Locations, Faces, and Pictures for non-verbal memory),17 and the Child Behavior Checklist (only available in Dutch and English).20 ||In a twin with an important neurodevelopmental delay, cardiac assessment was done at 4 years and was within normal ranges. At age 9 years cardiac assessment was not done at the request of the parents. Neurological assessments were done at ages 4 and 9 years. **For these age groups the age-adapted test battery comprised the Wechsler Intelligence Scale for Children (one child was tested with the Snijders-Oomen non-verbal intelligence test because their native language was not the national language), the Test of Everyday Attention for Children,16 subtasks from the Children’s Memory Scale (Numbers for verbal memory; Dot Locations, Faces, and Pictures for non-verbal memory), the Auditory Verbal Learning Test,18,19 and the Child Behavior Checklist (only available in Dutch and English). ††For this age group the age-adapted test battery comprised the Wechsler Adult Intelligence Scale, the Auditory Verbal Learning Test, and the Child Behavior Checklist. ‡‡Ages of children in this group were 1·9, 2·1, 2·3, 3·3, 3·7, and 3·7 years (median 2·8 years; range 1·9–3·7). §§Echocardiography was not complete for one child because of the child’s movements. Table 1: Medical assessments by age group www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 3 Articles A B 100 Weight (kg) 80 60 40 20 0 C D 200 Height (cm) 150 100 50 0 E F Head circumference (cm) 80 60 40 20 0 0 50 100 Age (months) 150 200 0 50 100 Age (months) 150 200 Figure 2: Biometry of children after prenatal exposure to chemotherapy Girls’ weight (n=32; A), boys’ weight (n=38; B), girls’ height (n=31; C), boys’ height (n=30; D), girls’ head circumference (n=29; E), and boys’ head circumference (n=30; F). For each child we used the last available measurement. The lines are the 5th, 50th, and 95th percentiles; the dots are the measurements from our study population. For the customised weight centile calculator see http:// www.gestation.net/ 4 Median maternal age was 32·9 years (range 23·4–41·7; IQR 6·0) and gestational age at diagnosis was 18·1 weeks (1·7–33·1; 9·0). Median gestational age at birth was 35·7 weeks (28·3–41·0; 3·3). Seven children were born at 28·0–31·9 weeks, nine at 32·0–33·9 weeks, 31 at 34·0–36·9 weeks, and 23 at term (≥37 weeks). Median birthweight was 2612 g (720–3970; 955). Birthweight was below the tenth percentile for gestational age and sex in 14 of 70 children (20·6%, 95% CI 12·3–30·8; onetailed binomial test p=0·009). The incidence and type of congenital malformations were similar to the general population (appendix). Furthermore, neonatal physical examination and echocardiographic assessments (n=21) were within normal limits (data not shown). Neonatal neurological examination was normal in 64 children (91%). One child, born at 28 weeks, presented with a contracture of the right elbow, which resolved later. In five children a transient neonatal hypotonia was noted; one of these children also presented with benign sleep myoclonus. www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 Articles Average group results on the different subtasks for verbal and non-verbal memory were within normal ranges (figure 4). Attention function for the 12 children with available data was within the normal range. Figure 4 lists detailed results grouped per attention function: focused attention, sustained attention, attention flexibility, divided attention, and response inhibition. The Child Behavior Checklist was available for 21 children at a median age of 8·7 years (range 5·0–15·9; IQR 3·0) and the average score was within the normal ranges (figure 4). For internalising, A 60 Intelligence test between age 5 and 18 years Bayley test at age 18 months Clinical neurological examination 49 Number 40 20 10 0 5 5 x – (>2 SD) x – (1–2 SD) 1 x ± (1 SD) x + (1–2 SD) x + (>2 SD) B 140 130 120 IQ score 110 100 90 80 70 60 50 28 29 30 31 32 33 34 35 36 37 Pregnancy duration (weeks) C 150 140 Verbal IQ Performance IQ Total IQ 130 120 IQ score The median follow-up period at last examination of the children was 22·3 months (range 16·8–211·6; IQR 54·9). Medical assessments by age are shown in table 1. Biometric data showed that children had normal values for weight, height, and head circumference (figure 2). The questionnaire on general health status and development was completed by 57 parents (81%) at the study visits. Medical problems reported by the parents (appendix) are also common in the general population. We did not identify any malignancies. General development was reported as seriously impaired in both children of a twin pregnancy. Apart from these children, all school-aged children attended a normal school (n=25). Social and leisure activities were similar to the general population (data not shown). Children with missing questionnaires mainly lived in the Netherlands and the French speaking part of Belgium. Apart from their different cultural background, we had less personal contact with these parents since we were not the treating physicians. However, we did not identify differences in maternal history and children’s outcomes when compared with parents who completed the questionnaires. We recorded normal cognitive development in most children. Figure 3 shows the distribution of the results for the last cognitive evaluation (Bayley test [n=40], Wechsler or Snijders-Oomen non-verbal intelligence test [n=26], and clinical neurological examination between standardised ages [n=4]). Children who scored below normal ranges were mainly from the preterm group. A univariable linear regression model showed that the average IQ score increased by 11·1 points (95% CI 5·4–16·8; p=0·0003) for each month increase in pregnancy duration (explained variation in IQ scores was 16%; figure 3). When controlling for age, sex, and country, the effect remained: the score increased on average by 11·6 (95% CI 6·0–17·1, p<0·0001) for each month increase in pregnancy duration. Two children (one of the sets of twins) had a significant neurodevelopmental delay, which made it impossible to do the complete proposed in-depth cognitive test battery (appendix). We excluded these two children from further cognitive and behavioural assessment. In one of these children, a cortical malformation with multiple dysmorphic characteristics was recorded (appendix). Although a syndromal entity is probable in this child, we were unable to diagnose a clinical syndrome. The parents refused a brain MRI for the other twin, but this child has no physical abnormalities. Total IQ scores for the children were within normal ranges (n=25; 103 [SD 14·5]). Comparing verbal and performance IQ scores obtained in the Wechsler tests (n=23), paired t test showed a significant difference (verbal score 104·8 [SD 14·5] vs performance score 97·9 [14·1]; paired t test p=0·033). This observation was driven by nine children (39%) where there was a significant difference between verbal and performance IQ scores (figure 3); the remaining 14 (61%) did not show a significant difference. 110 100 90 80 70 60 50 www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 Individual child 38 39 40 41 Figure 3: Cognitive development (A) General cognitive development after prenatal exposure to chemotherapy (n=70). The distribution of the result of the last performed neurological evaluation for each child (Bayley test at age 18 months [n=39], intelligence test between age 5 and 18 years [n=24 Wechsler IQ test, n=2 Snijders-Oomen non-verbal intelligence test, n=1 Bayley test], clinical neurological examination in between standardised ages [n=4]). Normal range is defined as x plus or minus 1 SD. (B) Cognitive outcome in relation to gestational age at birth. IQ score was obtained from Wechsler test (n=24), Snijders-Oomen non-verbal intelligence test (n=2), Bayley test (n=40), clinical neurological examination whereby findings within normal ranges were translated into a mean score of 100 (n=4). A linear regression model suggested that the average IQ score increases by 2·5 (95% CI 1·2–3·9) for each week increase in pregnancy duration (p=0·0003, ω²=0·16). Circles are individual data, solid line is predicted IQ score, and dashed lines are the 95% CIs for the prediction. (C) IQ scores from Wechsler (n=23) and Snijders-Oomen non-verbal intelligence tests (n=2; last two columns). Vertical lines represent a significant difference between verbal and performance IQ scores. Shaded band represents the normal range. IQ=intelligence quotient. 5 Articles Task or subtask Behaviour CBCL Verbal memory AVLT Verbal memory span CMS Numbers Non-verbal memory CMS Dots CMS Faces Tested variable N Median age (range; IQR) in years Internalising Externalising Total problems 21 21 21 8·7 (5·0–15·9; 3·0) 8·7 (5·0–15·9; 3·0) 8·7 (5·0–15·9; 3·0) Learning Immediate recall Delayed recall 12 12 12 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) Forward Backward Total numbers 25 25 25 8·5 (5·0–17·6; 3·0) 8·5 (5·0–17·6; 3·0) 8·5 (5·0–17·6; 3·0) Learning Immediate recall Delayed recall Immediate recall Delayed recall 25 25 24 25 24 8·5 (5·0–17·6; 3·0) 8·5 (5·0–17·6; 3·0) 8·2 (5·0–17·6; 3·0) 8·5 (5·0–17·6; 3·0) 8·2 (5·0–17·6; 3·0) 25 8·5 (5·0–17·6; 3·0) 12 12 12 12 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 12 12 12 12 12 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) 12 11 12 12 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 1·0) 9·0 (8·5–15·9; 0·6) 9·0 (8·5–15·9; 0·6) Non-verbal memory span CMS Pictures Selective attention Tea-Ch Sky Search Correct score Time score Attention score Tea-Ch Map Mission Sustained attentions Tea-Ch Score Tea-Ch Code Transmission Tea-Ch Sky Search DT* Tea-Ch Score DT* Tea-Ch Walk/Don’t Walk† Attentional flexibility Tea-Ch Creature Counting Correct score Time score Tea-Ch Opposite World Same world Opposite world –1 0 Z score 1 Figure 4: Cognitive development and behaviour (n=25, 5–18-year-old children) *Also tests divided attention. †Also tests response inhibition. CBCL=Child Behavior Checklist. AVLT=Auditory Verbal Learning Test. CMS=Children’s Memory Scale. Tea-Ch=Test of Everyday Attention for Children. externalising, and total problems, six (29%) of 21 children with available data had an increased score (Z score >1·0). No significant relation was evident with prematurity. All these children attended a normal school. Clinical neurological examination (n=68) did not show any focal or other neurological abnormalities. The mean Bayley mental developmental index score was 96·8 (SD 14·9; n=40, at a median age of 18·2 months [range 16·8–28·6; IQR 0·7]). Index scores for children born at term (n=13) were 103·1 (SD 13·6) and preterm (n=27) were 94·6 (SD 14·6). We tested auditory function in 21 children at the median age of 6·5 years (range 5·0–17·4; IQR 2·9). We did not detect abnormalities in 18 children (86%; four received cisplatin); in three children with hearing loss, middle ear infection (n=1) and neurodevelopmental problems (n=2) were confounding factors (appendix). 69 children aged 1–18 years underwent a non-sedated ECG and echocardiographic assessment (two children were tested off-protocol at 12 months). One child refused echocardiography and we excluded four children because it was impossible to complete the assessment because of a lack of cooperation; so data of 65 children were available for analysis. When compared with controls, we did not identify any statistically significant differences by weight, height, body surface area, and systolic blood pressure (all p values ≥0·45). 49 pregnancies (78%) of 64 had been exposed to anthracyclines. We noted a higher heart rate in our study participants (median 109, range 61–152) by comparison with the control group (104, 54–139; p=0·25); however, analysis of ECG measurements revealed no arrhythmia or conduction abnormalities (appendix). During the echocardiographic examination we did not identify any structural cardiac defects. Table 2 summarises the echocardiographic measurements of cardiac dimensions and systolic function in children exposed to anthracyclines. All cardiac dimensions were within the normal range for children N* Study children (n=50) Controls (n=50) Shortening fraction (%) 98 35 (30 to 43) 39 (28 to 51) –4·0 (–5·9 to –2·1) <0·0001† Ejection fraction (%) 98 66 (60 to 75) 71 (57 to 83) –6·0 (–8·3 to –3·7) <0·0001† Left ventricle mass index (g/m²) 94 48·3 (37·4 to 65·5) 55·4 (40·0 to 82·6) –2·53 (–7·89 to 2·82) 0·35 Left ventricle mass index (left ventricle mass per height²·⁷) 93 35·9 (19·8 to 54·5) 35·3 (20·6 to 28·8) –0·94 (–3·79 to 1·91) 0·51 Left ventricle Tei index 94 0·33 (0·27 to 0·45) 0·32 (0·15 to 0·53) Left ventricle end diastolic diameter (Z score) 96 0·02 (–2·28 to 3·19) 0·25 (–2·22 to 3·00) Left ventricle posterior wall thickness (Z score) 95 0·16 (–2·32 to 1·81) Interventricular septum thickness (Z score) 95 –1·09 (–2·50 to 0·45) Right ventricle end diastolic diameter (Z score) 93 0·76 (–1·48 to 1·83) Effect (95% CI) 0·014 (–0·025 to 0·052) p value 0·47 –0·13 (–0·64 to 0·38) 0·61 –0·20 (–1·90 to 1·98) 0·35 (–0·25 to 0·94) 0·25 –0·64 (–2·10 to 1·65) –0·47 (–0·89 to –0·05) 0·029 0·30 (–1·91 to 2·21) 0·29 (–0·19 to 0·76) 0·24 Data are median (range) unless otherwise stated. We estimated the independent effect of chemotherapy with stratified median regression with six strata for age and with sex as covariate. *Missing data are from both the study and control groups: shortening fraction (2 control), ejection fraction (2 control), left ventricle mass index, g/m² (2 study, 4 control), left ventricle mass index, left ventricle mass per height²·⁷ (3 study, 4 control), left ventricle Tei index (2 study, 4 control), left ventricle end diastolic diameter (2 study, 2 control), left ventricle posterior wall thickness (2 study, 3 control), interventricular septum thickness (2 study, 3 control), and right ventricle end diastolic diameter (2 study, 5 control). †Statistically significant after correction for multiple testing with Holm’s method including the variables from the appendix. Table 2: Echocardiographic data of children exposed to anthracyclines in utero and their matched controls 6 www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 Articles exposed to anthracyclines and for control children. Compared with the control group, we noticed a clinically small but statistically significant decrease for the patient group concerning ejection fraction, fractional shortening, and interventricular septum thickness. However, all patients were within the normal range and no patient had an abnormal value. Diastolic variables were also within the normal range with some clinically small but statistically significant differences between both groups. Mitral valve E velocity was lower, mitral valve A duration was shorter, and isovolumetric relaxation time was shorter in the patient group compared with the controls (appendix). These variables are heart-rate dependent. Discussion Despite prenatal exposure to chemotherapy, radiotherapy, staging examinations, and co-medications, the outcome for children in our study is not different from the general population. In particular, we noted ageadequate cognitive development and normal cardiac outcome in a cohort of children aged at least 18 months who were prenatally exposed to chemotherapy and tested at predefined ages. In our cohort, we confirmed the negative prognostic effect of prematurity on cognitive development (Bayley or IQ score).8,9 However, we cannot exclude an additional effect of chemotherapy and other drugs or radiation exposure on these children born preterm. The decision to induce the delivery for nonobstetric reasons—mostly to start cancer treatment or in a few cases because of deterioration of the maternal health—was taken in 38 pregnancies (56%) of 68. 28 (74%) of these 38 patients were delivered, iatrogenically, preterm (median gestational age 35·3 weeks; range 31·3–36·3). We suggest that the start of cancer treatment (including chemotherapy) during pregnancy might prevent iatrogenic prematurity and add to the preservation of long-term neurocognitive outcome. These data should be interpreted with caution since two children, both members of a twin pregnancy, presented with an important neurodevelopmental delay. The fetal cortex changes during development from a smooth cerebral surface at 14 gestational weeks to a complex association of sulci and gyri. Polymicrogyria is the endpoint of different pathological processes.24 We cannot rule out that prenatal exposure to chemotherapy after 15 weeks of gestational age affected the developmental delay, although the syndromal presentation makes the causal relation less probable. Moreover, in the children older than 6 years who did a Wechsler intelligence test (n=23), a disharmonic intelligence profile—discrepancies between verbal and performance IQ values—was evident in 39% (15% in the normal population).25 Although this finding does not directly suggest neuropsychological impairment, verbal and performance IQ discrepancies have been associated with several neurological disorders and learning problems. Furthermore, children with disharmonic intelligence profiles more often have behavioural problems than children with harmonic intelligence profiles.26 In our series the average results of the Child Behavior Checklist were within normal ranges, although in six (29%) of the 21 children the total problem score was raised, which is higher than we expected (15% in the general population).9 Although the general neurodevelopmental assessment was within normal ranges, these findings show that more subtle changes in neurodevelopment are possible. Therefore we need to remain prudent until a larger group of children has been assessed with a longer follow-up. In our study, fetal exposure to chemotherapy in utero was not associated with congenital cardiac abnormalities. For the children exposed to anthracyclines, variables for systolic and diastolic function were all within normal ranges. We identified a significant difference in ejection fraction and fractional shortening between the patients and age-matched and sex-matched controls. However, because values obtained in all patients in the anthracycline group were normal, as in the report of Aviles and colleagues,14 we do not believe this difference is clinically relevant. Some of the diastolic variables (isovolumetric relaxation time and mitral A duration) were different between patients and controls, but also these data were within the normal range and the differences can probably be explained by the higher heart rate in the study group. It is reassuring that cardiac dimensions, wall thickness, and left ventricular mass index were all within normal ranges. We discussed the need for matching for prematurity with our cardiologists although prematurity is not a predictor of heart dysfunction. Furthermore, we did not identify heart dysfunction for premature children. We believe that three important factors contributed to our overall reassuring outcome. First, chemotherapy was only given after the first trimester, which is the most vulnerable time for toxic effects.27 Chemotherapy given during the second and third trimester does not increase congenital malformations.11,28 Second, by contrast with the belief that the fetal blood–brain barrier is immature and leaky, recent data suggest that the fetal brain is well protected.29–31 Compared with the situation in adults, fetal cerebrospinal fluid contains high concentrations of proteins related to specialised transcellular transfer and needed for fetal brain development.30 Erroneously, these high protein concentrations were once interpreted as a consequence of a leaky blood–brain barrier. During embryogenesis, almost all neurons are formed by 6–18 weeks of gestation, but the brain continues to develop later in pregnancy and also postnatally, through neuronal migration, differentiation and synaptic maturation, and myelinisation. The underlying morphological features of the blood–brain barrier are the presence of tight junctions, low rates of transcytosis, and the expression of specialised influx and efflux transporters, www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 7 Articles Panel: Research in context Systematic review To our knowledge, ours is the first comprehensive report on the long-term outcome of children after prenatal exposure to chemotherapy. We searched PubMed for reports published from 1990 to 2011, with the search terms “pregnancy”, “cancer”, “children”, “chemotherapy”, and “outcome”. We did not limit our search by language. We also searched review papers. We identified very limited published data on the examination of children long term after prenatal exposure at the time of our search, indicating a need for information in this area. Interpretation Our findings suggest that general health and growth, and CNS, cardiac, and auditory function of children born to mothers treated with chemotherapy during pregnancy did not differ from the general population. Our findings do not support a strategy of delay in chemotherapy administration or iatrogenic preterm delivery with post-partum chemotherapy administration to avoid harm to the fetus. However, we noted subtle changes and underscore the need for longer follow-up in more children. For more on Cancer in Pregnancy see http://www. cancerinpregnancy.org/ 8 which are present early in embryological development.30 Moreover, pericytes inhibit the expression of molecules that increase vascular permeability and CNS immunecell infiltration.29 Tight junctions are present between cerebral endothelial cells and between choroid plexus epithelial cells, and restrict the entry of proteins into the brain and cerebrospinal fluid. In the immature brain there are additional morphological barriers at the interface between the cerebrospinal fluid and brain tissue: strap junctions at the inner neuroependymal surface and these and other intercellular membrane specialisations at the outer (pia-arachnoid) surface. These barriers disappear later in development and are absent in adults.30 Apart from this morphological protection, the presence of functional efflux transporters, including P-glycoprotein, reduces brain penetration of drugs.32 Virgintino and colleagues31 reported on the expression of P-glycoprotein early in human fetuses during cerebral cortex formation. At the earliest examined stage, 12 weeks of gestation, P-glycoprotein was detectable as diffuse cytoplasmic labelling of the endothelial cells lining the primary cortex microvessels. At 18 weeks of gestation, a punctate P-glycoprotein staining pattern was detected on cortex and subcortical vessels and on their side branches. At 22 weeks of gestation, P-glycoprotein staining was linear and concentrated on endothelial cell membranes. Third, the placenta filters cytotoxic drugs and shields a proportion from the fetus. Data on human beings are anecdotal, but transfer rates in pregnant baboons vary and range from 0 to 57% of the maternal serum concentrations.33,34 The limitations of our study include our small sample size, a short follow-up period, and a lack of a direct comparison with identically assessed children, born at the same gestational age but without prenatal exposure to chemotherapy. To tackle the latter, we have now started to build a control group: children matched for the gestational age at birth, which will be followed up in accordance with the same protocol as the children exposed to chemotherapy. Because of the short follow-up period, our study does not allow us to investigate for secondary malignancies. In children, the use of low-dose etoposide is associated with secondary leukaemia, but genetic susceptibility to neoplasia might be a confounding factor.35 So far, only one case of secondary neoplasm after transplacental exposure to chemotherapy has been described. One member of a twin exposed to DNAdamaging cyclophosphamide and prednisone for acute lymphocytic leukaemia was born with congenital malformations and was diagnosed with papillary thyroid cancer at age 11 years and neuroblastoma at age 14 years.36 Given the unknown effect on secondary malignancies, where possible, preference to microtubule-targeting agents and especially taxanes should be given.2,37 Longterm follow-up is thus mandatory but sensitive to survivor bias. Nevertheless, all children should be followed up and reasons for exclusion, loss of follow-up, or refusal for participation need to be documented properly. We show that children who were prenatally exposed to chemotherapy do as well as other children. Based on our data, the long-term effect of prenatal exposure to chemotherapy seems to outweigh the maternal need for treatment during pregnancy, especially when we consider confounding factors such as prematurity and maternal stress.8–10 Although we cannot exclude a role of chemotherapy in the poor outcome of a twin and a higher incidence of disharmonic intelligence profiles, our findings suggest that chemotherapy could be used during pregnancy, if needed. The decision to administer chemotherapy should follow the same guidelines as in non-pregnant patients.3 In practice, it is possible to administer chemotherapy from 14 weeks gestational age onwards with specific attention to prenatal care.3 To allow the bone marrow to recover and to minimise the risk of maternal and fetal sepsis and haemorrhage, delivery should be planned at least 3 weeks after the last cycle of chemotherapy, and chemotherapy should not be given after 35 weeks since spontaneous labour becomes more probable.3 Furthermore, neonates, especially preterm babies, have restricted capacity to metabolise and eliminate drugs because of liver and renal immaturity. The delay of delivery after chemotherapy will allow fetal drug excretion via the placenta.3 Given the negative prognostic effect of prematurity on cognitive development, preterm birth should be avoided, if possible. Only time will inform us of the full consequences, including fertility and secondary malignancies (especially if DNA damaging drugs are used), of fetal exposure to chemotherapy. Therefore, we continue our international collaborative initiative, Cancer in Pregnancy, and strive to follow up a greater number of children over a longer period (panel). Contributors FA thought of the concept of the study. FA and KVC searched the published work. FA, MJH, and PBO were the national study coordinators. FA, KVC, MMG, LH, SNH, MJH, and PBO identified www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 Articles children and organised examinations in the national study centres; they collected all clinical data and study results. MMG, LK, and VT did the cardiac assessments. FA, MMG, LM, WH, and LK did the cardiac data analysis and interpretation. FA, KVC, MJH, LL, MAW, and HW did the cognitive tests, cognitive data analysis, and interpretation. KVC and BVC did the statistical analyses. FA and KVC wrote the first draft of the report. All authors approved the final version of the report. Conflicts of interest We declare that we have no conflicts of interest. Acknowledgments This report was presented as a late breaking abstract during the European Multidisciplinary Cancer Congress in Stockholm, Sweden (Sept 26, 2011). FA is senior clinical researcher and BVC postdoctoral researcher for the Research Foundation-Flanders; this research is supported by Research Foundation-Flanders Project G 0358.06, “Stichting tegen kanker Project SCIE2006-17”, Research Fund-K.U.Leuven (OT/07/053), Flemish Government, Agency for Innovation by Science and Technology Project TBM070706-IOTA3, Clinical Research Fund-UZ Gasthuisberg, and Belgian Cancer Plan, Ministry of Health NKP 29 038. We thank the parents and children for their participation in our study, and all colleagues who contributed to this study. We thank Javier Ganame and Daisy Thijs (Leuven), Bob Rijk and Imke Tomasouw-Janssen (Nijmegen) for echocardiographic assessment of the children. We thank Ilse De Croock, Griet De Mulder, and Caroline Sterken (Leuven); Anja Vinck (Nijmegen); and Jitka Zackova (Prague) for cognitive evaluation of the children. We thank Anna Jansen from Vrije Universiteit Brussel for mutational analysis in two patients. We thank Liesbet Van Eycken from the Belgian Cancer Registry and Frank De Smet from the National Alliance of Christian Sickness Funds for helping with data extraction. We thank Marieke Taal, Diane Wolput, and Marie-Astrid Van Hoorick for their administrative support. The full list of investigators is in the appendix. References 1 Creskoff A, Fitz-Hugh TJ, Frost J. Urethane therapy in leukemia. Blood 1948; 3: 896–910. 2 Amant F, Van Calsteren K, Halaska MJ, et al. Gynecologic cancers in pregnancy: guidelines of an international consensus meeting. Int J Gynecol Cancer 2009; 19 (suppl 1): S1–12. 3 Amant F, Loibl S, Neven P, Van Calsteren K. Breast cancer in pregnancy. Lancet 2012; 379: 570–79. 4 Nieminen U, Remes N. Malignancy during pregnancy. Acta Obstet Gynecol Scand 1970; 49: 315–19. 5 Potter JF, Schoeneman M. Metastasis of maternal cancer to the placenta and fetus. Cancer 1970; 25: 380–88. 6 Stensheim H, Moller B, van Dijk T, Fossa SD. Cause-specific survival for women diagnosed with cancer during pregnancy or lactation: a registry-based cohort study. J Clin Oncol 2009; 27: 45–51. 7 Mennes M, Stiers P, Vandenbussche E, et al. Attention and information processing in survivors of childhood acute lymphoblastic leukemia treated with chemotherapy only. Pediatr Blood Cancer 2005; 44: 478–86. 8 Lohaugen GC, Gramstad A, Evensen KA, et al. Cognitive profile in young adults born preterm at very low birthweight. Dev Med Child Neurol 2010; 52: 1133–38. 9 van Baar AL, Vermaas J, Knots E, de Kleine MJ, Soons P. Functioning at school age of moderately preterm children born at 32 to 36 weeks’ gestational age. Pediatrics 2009; 124: 251–57. 10 Van den Bergh BR, Van Calster B, Smits T, Van Huffel S, Lagae L. Antenatal maternal anxiety is related to HPA-axis dysregulation and self-reported depressive symptoms in adolescence: a prospective study on the fetal origins of depressed mood. Neuropsychopharmacology 2008; 33: 536–45. 11 Van Calsteren K, Heyns L, De Smet F, et al. Cancer during pregnancy: an analysis of 215 patients emphasizing the obstetrical and the neonatal outcomes. J Clin Oncol 2010; 28: 683–89. 12 Aviles A, Neri N. Hematological malignancies and pregnancy: a final report of 84 children who received chemotherapy in utero. Clin Lymphoma 2001; 2: 173–77. 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Hahn KM, Johnson PH, Gordon N, et al. Treatment of pregnant breast cancer patients and outcomes of children exposed to chemotherapy in utero. Cancer 2006; 107: 1219–26. Aviles A, Neri N, Nambo MJ. Long-term evaluation of cardiac function in children who received anthracyclines during pregnancy. Ann Oncol 2006; 17: 286–88. van der Meulen BF, Ruiter SAJ, Spelberg HCL, Smrkovský M. BSID-II-NL: Bayley Scales of Infant Development, 2nd edn— Nederlandse versie. Lisse: Swets Test Publishers, 2002. Manly T, Robertson I, Anderson V, Nimmo-Smit I. TEA-Ch: test of everyday attention for children—Nederlandstalige vertaling. Amsterdam: Harcourt Test Publishers, 2007. Cohen MJ. Children’s memory scale. Paris: Les Editions du Centre de Psychologie Appliquée, 1997. Forrester G, Geffen G. Performance measures of 7- to 15-year-old children on the auditory verbal learning test. Clin Neuropsychol 1991; 5: 345–59. Rey A. L’examen clinique en psychologie. Paris: Presses Universitaires de France, 1964. Achenbach T. Child behavior checklist/4-18, Nederlandstalige vertaling. Rotterdam: ASEBA, Erasmus MC-Sophia Kinderziekenhuis, Afdeling Kinder-en jeugdpsychiatrie, 1991. Roelants M, Hauspie R. Groeicurven vlaanderen 2004. Brussel: Laboratorium voor Antropogenetica, Vrije Universiteit Brussel, 2004. Dickinson DF. The normal ECG in childhood and adolescence. Heart 2005; 91: 1626–30. Koenker R. Quantile regression. New York: Cambridge University Press, 2005. Verrotti A, Spalice A, Ursitti F, et al. New trends in neuronal migration disorders. Eur J Paediatr Neurol 2010; 14: 1–12. Kaufman AS, Lichtenberger EO. Essentials of WISC-III and WPPSI-R assessment. New York, NY: John Wiley and Sons, 2000. Isen J. A meta-analytic assessment of Wechsler’s P>V sign in antisocial populations. Clin Psychol Rev 2010; 30: 423–35. Cardonick E, Iacobucci A. Use of chemotherapy during human pregnancy. Lancet Oncol 2004; 5: 283–91. Cardonick E, Usmani A, Ghaffar S. Perinatal outcomes of a pregnancy complicated by cancer, including neonatal follow-up after in utero exposure to chemotherapy: results of an international registry. Am J Clin Oncol 2010; 33: 221–28. Daneman R, Zhou L, Kebede AA, Barres BA. Pericytes are required for blood–brain barrier integrity during embryogenesis. Nature 2010; 468: 562–66. Saunders NR, Knott GW, Dziegielewska KM. Barriers in the immature brain. Cell Mol Neurobiol 2000; 20: 29–40. Virgintino D, Errede M, Girolamo F, et al. Fetal blood–brain barrier P-glycoprotein contributes to brain protection during human development. J Neuropathol Exp Neurol 2008; 67: 50–61. Schinkel AH, Wagenaar E, Mol CA, van Deemter L. P-glycoprotein in the blood-brain barrier of mice influences the brain penetration and pharmacological activity of many drugs. J Clin Invest 1996; 97: 2517–24. Van Calsteren K, Verbesselt R, Devlieger R, et al. Transplacental transfer of paclitaxel, docetaxel, carboplatin, and trastuzumab in a baboon model. Int J Gynecol Cancer 2010; 20: 1456–64. Van Calsteren K, Verbesselt R, Beijnen J, et al. Transplacental transfer of anthracyclines, vinblastine, and 4-hydroxy-cyclophosphamide in a baboon model. Gynecol Oncol 2010; 119: 594–600. Le Deley MC, Vassal G, Taibi A, Shamsaldin A, Leblanc T, Hartmann O. High cumulative rate of secondary leukemia after continuous etoposide treatment for solid tumors in children and young adults. Pediatr Blood Cancer 2005; 45: 25–31. Reynoso EE, Shepherd FA, Messner HA, Farquharson HA, Garvey MB, Baker MA. Acute leukemia during pregnancy: the Toronto Leukemia Study Group experience with long-term follow-up of children exposed in utero to chemotherapeutic agents. J Clin Oncol 1987; 5: 1098–106. Mir O, Berveiller P, Ropert S, et al. Emerging therapeutic options for breast cancer chemotherapy during pregnancy. Ann Oncol 2008; 19: 607–13. www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1 9 Articles 10 www.thelancet.com/oncology Published online February 10, 2012 DOI:10.1016/S1470-2045(11)70363-1