OBJECTIVE: Although children with postnatal-onset microcephaly (POM) generally have poor development, we speculated that better somatic growth would predict better development in these children.
PATIENTS AND METHODS: We followed 57 children with POM for an average of 4.2 years (13 encephaloclastic, 14 dysgenetic, 6 with Rett syndrome, 24 idiopathic) and calculated the developmental quotient (DQ) at each visit (DQ > 0.70 was considered normal). SD scores (SDS) for measurements were analyzed using a repeated measures mixed-effects model to assess effect of weight, height, head circumference (HC), and age on DQ. Pearson's correlation was used to examine the independent influence of each variable on final DQ.
RESULTS: Forty-four children (77%) had a low DQ (mean: 0.33), but 13 (23%) had a normal DQ (mean: 0.93), including 10 idiopathic and 3 encephaloclastic. Mean HC fell below −2 SDS in all before 1 year (destructive at 3.3 months, idiopathic low-DQ at 7.5 months, dysgenetic at 8.5 months, Rett syndrome at 11 months, and idiopathic normal-DQ at 11.5 months). Mean weights and heights both fell below −2 SDS for all low-DQ groups but remained normal in both normal-DQ groups. Weight, height, and HC were independent predictors of DQ (P < .0001). Final DQ correlated with weight (r = 0.27), height (r = 0.41), and HC (r = 0.13).
CONCLUSIONS: Most children with POM have poor later development. Whatever the cause of POM, persons in whom postnatal body growth (weight, height, HC) is better sustained have more favorable development, and in one-quarter of such persons (mostly idiopathic POM), final DQ is normal.
WHAT'S KNOWN ON THIS SUBJECT:
Microcephaly is a common disorder of childhood with a generally unfavorable prognosis. Most microcephaly is congenital, although some cases first appear postnatally. The outcome of postnatal-onset microcephaly varies, and no predictors of development have previously been identified.
WHAT THIS STUDY ADDS:
This study found that developmental outcome in postnatal-onset microcephaly was normal in a substantial minority (25%). Three-quarters of those with normal development were idiopathic, with the remainder encephaloclastic. Better somatic growth (weight, height) was associated with a normal developmental outcome.
The natural history of postnatal-onset microcephaly (POM) is poorly understood, especially with respect to cognitive development and somatic growth. Although developmental outcome in children with POM is generally poor, some have normal intelligence.1 Although the etiology of POM is a predictor of developmental outcome, the outcome still varies greatly within each etiologic group. In otherwise healthy children, the link between malnutrition and poor cognitive development is clear, but an association between poor somatic growth and developmental outcome in POM has never been established. Our objective was to elucidate the natural history of POM and to determine whether somatic growth could predict developmental outcome in these children.
PATIENTS AND METHODS
We reviewed records of 87 children suspected of POM seen consecutively from 1982 to 2001 at Boston Medical Center and Tufts Medical Center, and we collected data on their somatic growth and cognitive development. All data, with exception of birth values, were derived from pediatric neurology visits at these institutions. Parental growth data were not collected because none of the literature on POM indicates a predictive relationship between parental head size and head growth in offspring, unlike certain cases of inherited congenital microcephaly. Postnatal-onset microcephaly was defined as a head circumference (HC) that was normal at birth (2nd to 98th percentile), became microcephalic (<2nd percentile) after the neonatal period, and remained so. We required 1 normal HC within the first 28 days of life and at least 2 subsequent HC measurements.
The 1968 Nellhaus HC charts2 and the 2000 National Center for Health Statistics weight and height charts3 were used in conjunction with Fenton's charts for premature weight, height, and HC4 (for premature children up to 38 weeks' conceptual age). Comparisons were made using SD scores generated with the LMS (λ, μ, σ) technique.5 The LMS calculation determines exact percentiles on the basis of a summary of the median (μ), coefficient of variation (σ), and skewness of distribution (λ) of the data. The National Center for Health Statistics charts were originally created using LMS, and the Fenton charts have recently been converted to LMS format.6 The Nellhaus charts were converted to LMS by the present authors.
Developmental age was determined at each visit using a composite developmental screening instrument compiled by the authors (see Supplemental Table 3). A developmental quotient (DQ) was calculated for each participant, using the method of Illingworth,7 dividing developmental age by chronological age (DQ = (gross motor age + fine motor age + language age + cognitive age)/(4 × actual age)). A developmental recategorization was then performed on all participants using a cognitively weighted DQ formula, which lessens the effect of severe motor impairment on calculation of a child's DQ. Developmental delay was defined as a DQ < 0.7.8 The DQ calculated at the most recent visit was used to divide participants into normal and abnormal groups.
Participants were categorized on the basis of etiology of their microcephaly and DQ, and birth and final SD scores were compared among categories using analysis of variance, with age included as a covariate. A repeated measures mixed-effects model was used to determine longitudinal relationships between the outcome variable, DQ and potential predictors: weight; height; HC; and participant age. Pearson's correlation was used to examine the independent influences of final measurements on the last DQ. Locally weighted scatterplot smoothing was used to generate mean size curves for each category. All analyses were performed using SAS 8 (SAS Institute, Inc, Cary, NC).9
Of the 87 cases, 30 were excluded: 19 because of resolution of microcephaly at a later age or because children referred to as microcephalic in their narratives were in fact normocephalic, according to the Nellhaus LMS charts; 8 because no birth HC was available; 2 because of insufficient follow-up growth data; and 1 who was reclassified as microcephalic at birth. The remaining 57 (27 male, 30 female) were followed for a mean of 4.2 years. During 702 visits to 1 of the 2 research sites, 457 body weights, 283 heights, 578 HCs, and 303 DQs were recorded (Table 1). Four etiologies emerged: 13 encephaloclastic, 14 dysgenetic, 6 with Rett syndrome, and 24 idiopathic (see Supplemental Table 4). Thirteen participants (23%) had a normal DQ (10 idiopathic, 3 encephaloclastic). Birth weight and HC were similar among groups (P: 0.92–0.97), but the dysgenetic group was significantly shorter at birth compared with all others (P = .02). Most encephaloclastic participants were female, whereas most idiopathic low-DQ participants were male; as expected, all Rett syndrome participants were female (Table 2).
Comparing combined low-DQ and combined normal-DQ groups, birth measurements were similar (P: 0.34–0.85). In the 44 cases with a low DQ, the mean DQ was 0.33 (±0.18), whereas in the 13 normal-DQ cases, the mean DQ was 0.93 (±0.18). Mean weight in the normal-DQ group remained in the normal range, whereas mean low-DQ weight fell steeply below the 2nd percentile at 19 months and continued to decline up to 63 months (Fig 1A). Mean height in the normal-DQ group also remained in the normal range, whereas mean low-DQ height fell below the 2nd percentile at 26 months and continued to decline up to 60 months (Fig 1B). Mean HC fell below the 2nd percentile in both DQ groups and plateaued in both between 24 and 36 months (Fig 1C).
Body weight, height, and HC were each significant independent predictors of DQ (P < .0001). This disparity persisted for height (P = .007), weight (P = .001), and HC (P = .03) when age was included as a covariate. Moreover, the results were unchanged when the cognitively weighted DQ formula was used in place of the standard formula. Correlation coefficients between final DQ and final body measurements were significant for weight, height, and HC (P < .05). Of these, height had the strongest linear relationship with DQ (r = 0.41), followed by weight (r = 0.27), and then HC (r = 0.13).
For each of the POM etiologies, mean weights fell below the 2nd percentile in the low-DQ group, but plateaued at or above the 2nd percentile in the normal-DQ group (Fig 1D). Likewise, mean heights fell below the 2nd percentile for the low-DQ group, but plateaued above the 2nd percentile in the normal-DQ group (Fig 1E). For each of the etiologies in both the low-DQ and normal-DQ groups, the mean HC fell below the 2nd percentile before 1 year of age (Fig 1F). Notably, final weight percentile was lowest in the Rett syndrome group, the group with the lowest DQ, and final height percentile was lowest in the dysgenetic group, the group with the second-lowest DQ.
The literature on POM is scanty, with 1 recent article,1 1 older one,10 and 1 book chapter11 forming the main corpus of that literature. The recent practice parameter of the American Academy of Neurology dedicates little text to POM, other than stating that it is usually apparent by 2 years and that infants who develop POM by 1 year are likely to be severely retarded.12
Some of the confusion surrounding POM is derived from the lack of a universally accepted definition. Microcephaly is usually defined as a HC > 2 SDs below the mean for age and gender,12,–,14 whereas severe microcephaly is a HC > 3 SD below the mean.11,12,14 Although 2.3% of children should be microcephalic on the basis of a normal distribution, published estimates are far lower, ∼0.5%, totaling 25 000 infants in the United States annually. By contrast, 0.1% of children would be expected to be severely microcephalic, which is in agreement with published estimates of severe microcephaly in 0.1% of newborns.12
“Acquired” microcephaly can refer either to a nongenetic microcephaly acquired prenatally or postnatally,12 or to a HC that was normal at birth (within 2 SD of the mean) and then declined to the microcephalic range after the newborn period, so-called “postnatal-onset” microcephaly,1,10,12 also referred to as progressive microcephaly1 and secondary microcephaly.11 These last 2 terms are potentially misleading. “Progressive” microcephaly suggests that the acquired microcephaly has continued to worsen (not universally true) or that it is caused by a progressive brain disease (rare in POM). “Secondary” microcephaly suggests that the microcephaly is caused by some specific disorder or disease (often the cause remains unknown). Moreover, a specific cause is found at least as often in congenital microcephaly as when microcephaly develops postnatally.
Postnatal-onset microcephaly can be classified in several ways.1,10,12 The classification proposed by Ashwal et al12 is perhaps the most useful; it divides all postnatal-onset microcephalies into genetic and acquired etiologies. Genetic causes are subdivided into metabolic inborn errors (eg, amino acidopathies) and syndromic (eg, Rett syndrome). Acquired etiologies include disruptive, infectious, toxic, and deprivational.12
Of our 57 cases of POM, 13 were encephaloclastic (combining “disruptive” and “infectious” of Ashwal et al12), and 14 were dysgenetic (“syndromic” of Ashwal et al12). Although Ashwal et al12 included Rett syndrome cases with other syndromic etiologies, we considered our Rett syndrome cases separately because Rett syndrome is clinically distinctive and is the prototype of POM.15 Our largest group, 24 idiopathic cases, had no identifiable cause.
In a recent study, Baxter et al1 reviewed 51 children with POM referred to a pediatric neurology clinic during a 10-year period. They identified 5 etiologic groups and 3 head growth patterns, but found no correlation between etiology and pattern of growth. Moreover, they found no associations between DQ/intelligence quotient (IQ), head circumference, etiology or growth pattern, except for an association between more severe microcephaly and poor somatic growth.
Among our 57 cases of POM, 13 (23%) had a normal final DQ. This indicates that the prognosis in acquired microcephaly, as stated by Ashwal et al12 (“likely to be severely delayed”) may be unduly pessimistic. Although too few of Baxter et al1 subjects had DQ/IQ measurements to compare causal groups statistically, their highest median DQ/IQ score (0.83) was in their idiopathic group and their lowest DQ/IQ score (0.45) in their syndromic group (encompassing our dysgenetic and Rett syndrome groups), consistent with our results.
In addition, we found a striking correlation between the pattern of somatic growth in POM and developmental outcome. In the low-DQ group, mean body weights fell to <2% and continued to decline up to 63 months, whereas in the normal-DQ group, mean body weights plateaued at 5% at 15 months. Similarly, mean body heights in the low-DQ group continued to decline below 2% up to 60 months, whereas mean body heights in the normal-DQ group never fell to 2%. Baxter et al1 had only limited somatic growth data, but in 9 cases with a decrease in weight (6) or in weight and height (3), only 1 was idiopathic (median DQ/IQ: 0.83); median DQs in the groups with underlying brain pathology were lower (0.40–0.62), as in our study.
The association of normal development in children with POM who have better somatic growth suggests a nexus between body growth and brain function in POM. Illingworth and Lutz16 studied infants in their first 10 months of life and found that in children with developmental delay, a head circumference disproportionately low for body weight may foreshadow mental subnormality, whereas a head circumference and body weight both low for age may reflect initial maturational delay with later normal development. Subsequent studies, however, have disagreed about whether proportionate microcephaly (similar weight, height, and HC percentiles) predicts a better prognosis.12,17
Recent studies have found both deficient and excessive growth to impact negatively on developmental outcome. In infants with intrauterine growth restriction, growth in the first 4 postnatal months was an independent risk factor for cognitive outcome at 7 years old, with both heavier and lighter extremes associated with lowered cognitive testing.18 In a study of 5771 children born at term, Emond et al19 found that when weight faltered from birth to 9 months, IQ at 8 years was lower, whereas weight gain from birth to 8 weeks had a positive linear association with IQ at age 8. Height is similarly linked to developmental outcome, with height at ages 9 and 13 years found to be significantly positively associated with IQ at 11 years.20
Both experimentally and clinically, malnutrition results in poor cognitive development. Brain weight, brain cholesterol, and brain DNA were all significantly lower in rats nutritionally deprived from birth, and these deficits were not corrected when the nutrition of the deprived animals was subsequently normalized.21 Clinically, Stoch et al22 found that children with severe and protracted undernutrition as infants became and remained microcephalic, with a significantly lower IQ on follow-up compared with a matched group of well nourished infants. Despite improved nutrition in the undernourished group, while weight and height trended toward normal, head circumference did not. This implies an early life “window” of nutritional vulnerability, which, if opened by nutritional deprivation, can never again be closed.
Brain development and body growth are intimately interrelated. Abnormal number, structure, or function of neurons or glia can interfere with normal growth and development. For example, in Rett syndrome (our group with lowest final body weights and lowest DQs), cortical and subcortical cell populations are normal in number, but they function abnormally, resulting in microcephaly (through poor dendritic arborization),23 and profound growth failure (through inability to consume and absorb sufficient calories to maintain weight). Although motor delays may have contributed to poor weight gain in some of our cases, this is not likely the entire explanation. For example, girls with Rett syndrome have difficulty sustaining normal weight because of oromotor apraxia, gastric dysmotility, and gastroesophageal reflux. Yet, this growth failure can only be partially reversed after gastrostomy tube supplementation, which indicates that other mechanisms, probably brain-based, contribute to their growth failure.24
Finally, neurotrophins may play an important role. Nerve growth factor levels are lower in newborns with intrauterine growth restriction than in those born appropriate for gestational age.25,26 Also, β-nerve growth factor in human cord blood is significantly lower in preterm infants than in those born at term and is lower in microcephalic children than in controls.27 Thus, lowered neurotrophins probably play an additive role in contributing to poor developmental outcome in children with POM and poor somatic growth.
In most children with POM, the developmental outcome is very poor. Of 57 children with POM followed for a mean duration of 4 years, most were severely delayed. In one-quarter, however, most of them idiopathic, later development was normal and was accompanied by better somatic growth, independent of age. These data reveal that when somatic growth is well maintained in a child with POM, especially of idiopathic causation, the child's later DQ may be entirely normal.
We thank Robyn Treadwell, MD, who assisted with the literature review.
- Accepted January 5, 2011.
- Address correspondence to N. Paul Rosman, MD, Division of Pediatric Neurology, Boston Medical Center, 1 Boston Medical Center Place, Dowling 3 S, Boston, MA 02118. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- POM =
- postnatal-onset microcephaly •
- HC =
- head circumference •
- LMS =
- λ, μ, σ •
- DQ =
- developmental quotient •
- IQ =
- intelligence quotient
- Baxter PS,
- Rigby AS,
- Rotsaert MH,
- et al
- Nellhaus G
- Kuczmarski RJ,
- Ogden CL,
- Grummer-Strawn LM,
- et al
- Illingworth R
- Custer J,
- Rau R
- 9.↵SAS Institute, Inc. 8 ed. Cary, NC; 2000
- Seaver L,
- Holden K
- Maria B
- Holden K,
- Lyons MJ
- Ashwal S,
- Michelson D,
- Plawner L,
- et al
- Maria BL
- Vermeulen RJ,
- Wilke M,
- Horber V,
- et al
- Illingworth RS,
- Lutz W
- Emond AM,
- Blair PS,
- Emmett PM,
- et al
- Pearce MS,
- Deary IJ,
- Young AH,
- et al
- Armstrong DD
HELMETS AND THE RISK OF INJURY: After a series of serious injuries this past year, the National Football League banned certain types of hits to the head. The injuries and subsequent rules spawned a lively debate about the balance between providing protection and maintaining the essence of the sport. Almost all agreed that football players need better head protection. However, what if the helmets themselves were to blame for the risks that players take? According to an article in The New York Times (February 16, 2011: Sports) that debate is being played out in women's lacrosse. As opposed to the men, who are required to wear helmets in all age groups, women lacrosse players do not wear helmets at all. And that, many experts say, is the safest approach. As reported in the article, players encased in helmets think they are safer than they actually are and engage in more dangerous and possibly injurious play. Advocates of the continued ban on helmets in women's lacrosse point out that checking in the National Hockey League became more violent after the adoption of helmets in the 1980s. As the size of the helmets and face masks used in football increased in size so did the number of head-to-head collisions. Advocates of protective head gear for women lacrosse players note that helmets may decrease the number of concussions. While women's lacrosse has a lower rate of concussion than women's soccer and basketball, the rate of concussion is only approximately 15 percent less than in men's lacrosse. The debate about whether helmets increase the risk of injury can be contentious. However, most would agree that scaling back helmet use in football or hockey would be too dangerous as violent play is now part of the game. Whether requiring women to use helmets in lacrosse would increase violent play is not known but as reported in the article, most college players think it would. For me, it is pretty simple. Having watched a lot of lacrosse, it seems safety equipment brings dangers of its own. When my son plays lacrosse, he leans into the contact. When my daughter plays lacrosse, she strategically moves so as to draw off her attacker and moves away from contact. I prefer the latter.
Noted by WVR, MD
- Copyright © 2011 by the American Academy of Pediatrics