Heterogeneity within the autism diagnosis obscures the genetic basis of the disorder and impedes our ability to develop effective treatments. We found that by using two readily available tests, autism can be divided into two subgroups, "essential autism" and "complex autism," with different outcomes and recurrence risks. Complex autism consists of individuals in whom there is evidence of some abnormality of early morphogenesis, manifested by either significant dysmorphology or microcephaly. The remainder have "essential autism." From 1995 to 2001, 260 individuals who met DSM-IV criteria for autistic disorder were examined. Five percent (13/260) were microcephalic and 16% (41/260) had significant physical anomalies. Individually, each trait predicted a poorer outcome. Together they define the "complex autism" subgroup, comprising 20% (46/233) of the total autism population. Individuals with complex autism have lower IQs (P=0.006), more seizures (P=0.0008), more abnormal EEGs (46% vs. 30%), more brain abnormalities by MRI (28% vs. 13%). Everyone with an identifiable syndrome was in the complex group. Essential autism defines the more heritable group with higher sib recurrence (4% vs. 0%), more relatives with autism (20% vs. 9%), and higher male to female ratio (6.5:1 vs. 3.2:1). Their outcome was better with higher IQs (P=0.02) and fewer seizures (P=0.0008). They were more apt to develop autism with a regressive onset (43% vs. 23%, P=0.02). Analysis of the features predictive of poor outcome (IQ<55, functionally non-verbal) showed that microcephaly was 100% specific but only 14% sensitive; the presence of physical anomalies was 86% specific and 34% sensitive. The two tests combined yielded 87% specificity, 47% sensitivity, and an odds ratio of 4.8:1 for poor outcome. Separating essential from complex autism should be the first diagnostic step for children with autism spectrum disorders as it allows better prognostication and counseling. Definition of more homogeneous populations should increase power of research analyses.

Children with phenylketonuria (PKU) obtain a great deal of their protein and mineral intakes from synthetic elemental formulae devoid of phenylalanine. To assess the effect of such diets and/or the disease on bone mineralization, children with PKU were compared to normal children for many parameters of mineral homeostasis and bone mineralization. A total of 11 children with PKU of mean age 10.9 +/- 4.2 years were compared to a large group of normal control children mean age 11.4 +/- 4.2, and an age and sex matched subset (n = 11). Children with PKU had lower serum calcium (9.1 +/- 0.9 vs 10.4 +/- 1.9 mg/dl P < 0.01) amd magnesium (1.67 +/- 1.4 vs 2.07 +/- 0.16 mg/ dl, P < 0.001) but normal values for phosphorus, zinc, and copper. The percentage tubular reabsorption of phosphorus was increased in PKU (93 +/- 3% vs 88 +/- 6%, P < 0.05) suggesting a lower phosphorus intake and/or absorption. Serum 25-hydroxyvitamin D, parathyroid hormone and 1,25 dihydroxyvitamin D were similar in PKU and control children. Serum albumin and lean body mass by dual energy X-ray absorption were not different suggesting that protein intake was adequate. In the 11 pairs, a decreased bone mineral density was seen for the lumbar spine (0.61 +/- 0.15 vs 0.72 +/- 0.24 P < 0.05), and lower extremities (1.56 +/- 0.30 vs 1.87 +/- 0.56 P < 0.05) by paired t-test. Compared to the total controls and the paired controls, decreases were seen in markers of bone formation; bone alkaline phosphatase, (72 +/- 30 vs 126 +/- 43 P < 0.001), osteocalcin (10.7 +/- 3.4 vs 13.1 +/- 2.0 P < 0.05) and procollagen type I carboxyterminal propeptide. No differences were seen in the bone resorption markers tartrate resistant acid phosphatase and urine Ca/Cr. The changes noted could not be related after age correction to serum phenylalanine levels, protein intake, or mineral intakes. It is unclear whether deficits in bone mineralization relate to the disease process itself or its treatment.

In an effort to delineate more homogeneous autism subgroups for genetic study, we evaluated 133 consecutive individuals referred to the University of Missouri Autism Center. Each index case underwent a diagnostic evaluation, including a clinical morphology examination, laboratory studies, brain MRI, EEG, and collection of historical, medical, and family data. The 71% (94/133) who fulfilled DSM-IV and CARS autism diagnostic criteria were included in this study. Six of 94 were diagnosed with a known genetic disorder. Of the remaining 88 with apparently "idiopathic autism," 58% (51/88) were phenotypically normal, 22% (19/88) were clearly abnormal, and for 20% (18/88) the clinical morphology examination was equivocal. The percentage of phenotypically abnormal individuals is higher than generally thought and disagrees with the perception that children with autism are usually normally formed. The phenotypically abnormal individuals were 10 times more likely to be diagnosed with a known genetic syndrome (21% vs. 2%) and were more than twice as likely (29% vs. 14%) to have structurally abnormal brain MRIs than the phenotypically normal propositi. Moreover, the male to female ratio correlated with the presence of physical anomalies. The total study group had a male to female ratio of 4.2:1; the morphologically normal subgroup, defined on the basis of a normal physical examination, had a sex ratio of 7.5:1 and the normal subgroup, defined on the basis of both a normal physical examination and a structurally normal brain by MRI had a 23:1 sex ratio. For the phenotypically abnormal subgroup, the sex ratio was 1.7:1. Since differences in sex ratio are presumably a reflection of differences in genetic constitution, we postulate that the phenotypically normal subgroup of individuals with "idiopathic autism" is genetically different from the phenotypically abnormal individuals and that differences in the sex ratio in different autism populations is one indicator of a population's genetic heterogeneity.

Occipitofrontal circumference (OFC) is one of the few physical findings in autism that varies significantly from the norm and is distinct and measurable. As part of a study of genetic heterogeneity of autism, we scrutinized data from a large sample of patients with idiopathic autism (N = 137), using OFC as the categorizing variable. The OFC standard deviation (OFCSD) values of the autistic propositi (0.61+/-1.6) varied significantly from that of the normal population (0.0+/-1.0), (P<0.001). Comparison of the macrocephalic (OFCSD > 2.0, N = 32) with the normocephalic individuals (-2 SD < OFCSD < +2 SD, N = 95) showed no significant differences in sex ratio, morphological status, IQ, seizure prevalence, or recurrence risks. The macrocephalic individuals were slightly less apt than those with normocephaly to have a family history of Attention Deficit Hyperactivity Disorder (ADHD) (P<0.05). Each clinical subgroup of autism propositi, defined on the basis of phenotypic status, type of onset, seizure history, or IQ, had a higher than normal mean OFC indicating that macrocephaly is an independent clinical trait in autism. As in the non-autistic population, macrocephaly was highly familial with 45% of the macrocephalic and 37% of the normocephalic propositi having at least one macrocephalic parent. Microcephaly, however, was an independent significant variable that predicted the presence of other phenotypic or genetic traits and outcome. The microcephalic patients were more likely to have abnormal physical morphology, structural brain malformations, lower IQ, and seizures. Their sex ratio was closer to normal, and their relatives had a higher incidence of seizures.