The study population
The principal source of subjects for this study was approximately 600 families with autistic children who subscribe to the informational services of the Autism Research Institute (ARI), Lacey, Washington (USA). We recruited a convenience sample of subjects [NT, AU, and pervasive developmental disorder-not otherwise specified (PDD-NOS) children, ages 2–12 years] into the study via a flyer and sent instructions by the study coordinator at the ARI to all subscribing ARI families, informing them about the study and inviting them to participate. The flyer directed interested parents/caregivers to respond by e-mail or telephone regarding their interest in participating. The study coordinator then contacted interested parents to describe the study and obtain consent. Consenting participants were asked to complete an online enrollment form and to provide a urine sample from the child/children in their families. The estimated participation rate for the ARI was 37%.
The online enrollment form contained detailed questions pertaining to the child’s diagnosis, including diagnostic criteria, diagnosing facility, name of diagnosing clinician, month/year of diagnosis, and diagnostic procedure(s) used. Additional questions were asked regarding dietary practices, drug exposures including chelation history, dental amalgam history, and child’s inoculation history. The number of vaccinations was collected as a potential source of Hg exposure; a distinction was made between total vaccinations and vaccinations prior to the year 2002 when thimerosal, a preservative containing an organomercurial moiety, was eliminated from many vaccines. In addition, to estimate maternal exposures to Hg during the 9 months of the index pregnancy, a count of dental amalgam tooth fillings (a potential source of Hg0 exposure) and an estimate of fish meals per week (a potential source of methylmercury exposure) for the mother were obtained for this time period.
Urine samples were collected in the home, transferred to the ARI by hand, and assigned a coded identification (ID) number by the ARI. They were then sent to the University of Washington for analysis, identified only by ID number, age, sex, and diagnosis. Data derived from these studies were evaluated in relation to the diagnostic information provided online to establish mean porphyrin levels for NT children and to determine the association of urinary porphyrin concentrations with autism or related neurobehavioral disorders.
To augment the number of subjects for whom porphyrin comparisons could be made, we analyzed porphyrin concentrations in urine samples acquired from an additional 41 subjects recruited through the Center for Autism and Related Disorders (CARD) in Tarzana, California, and 24 subjects, also through the CARD, from the Rimland Center for Integrative Medicine in Lynchburg, Virginia. CARD subjects were restricted to 2- to 12-year-old NT or AU males who had never undergone chelation treatment and were without amalgam dental fillings. CARD subjects were recruited prior to the initiation of the ARI study and hence did not complete the online enrollment questionnaire used by ARI participants. The estimated participation rate for the CARD was 31%.
Methods of subject recruitment as well as timing and manner of urine collection and processing were comparable between the CARD and ARI cohorts, and preliminary analyses of mean urinary porphyrin and Hg levels by subject source indicated no significant differences within age groupings. Therefore, data from both sources were pooled for porphyrin and Hg analyses. Overall, we enrolled 278 children in the study. After 55 children who had been previously chelated were excluded, 197 children were eligible for analysis. Final statistical models were performed using males only and included 59 NT, 59 AU, and 15 PDD-NOS subjects.
Human subjects considerations
The study protocol was approved by the institutional review boards at the University of Washington and the CARD. Human subjects approval of the ARI was conferred via an individual investigator agreement between the study coordinator at the ARI and the University of Washington. All parents/caretakers gave written consent for themselves and their children prior to enrollment in the study.
For children enrolled through the ARI, diagnosis of autism or other neurodevelopmental disorder was performed by established autism diagnostic and treatment centers that included the University of Washington Autism Center, the Seattle Children’s Autism Center [formerly the Autism Spectrum Treatment and Research Center (ASTAR)], and other pediatric neurology clinics throughout the Pacific Northwest. The diagnosis of AU, PDD-NOS, or other disorder at these centers was made using a multidisciplinary approach that combines a clinical evaluation using the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision
(DSM-IV-TR) (American Psychiatric Association 2000
) criteria, along with a psychological evaluation using the Autism Diagnostic Observation Schedule (ADOS) (Lord et al. 2000
), and other established diagnostic procedures such as the Autism Diagnostic Interview-Revised (Lord et al. 1994
) or the Childhood Autism Rating Scale (Schopler et al. 1993
). Verification of AU status by further psychological testing of children enrolled through the ARI was not feasible in this study. However, comparison of dates of AU diagnosis revealed that > 90% of subjects had been diagnosed since 2003, that is, within the immediate 5-year period since inception of this study, supporting continuity in the methods and procedures used and, therefore, homogeneity in the AU diagnosis. These observations further serve to verify the distinction between diagnoses of AU and other neurodevelopmental disorders, particularly PDD-NOS, as the same testing procedures and treatment centers were employed to diagnose PDD-NOS, most also occurring since 2003. No subjects were diagnosed before 2001.
For subjects enrolled through the CARD, all diagnosed children were fully evaluated by a trained psychologist and met the International Classification of Diseases, 9th Revision
(World Health Organization 1975
) and DSM-IV-TR (American Psychiatric Association 2000
) criteria for autism. All diagnoses were verified by obtaining copies of the diagnoses and subsequently validated by additional evaluations at the CARD using the ADOS and other diagnostic procedures cited above.
Participants recruited through the ARI were invited to enroll children with a previous diagnosis of autism or other neurodevelopmental disorder as well as their typically developing siblings. Children whose parents responded “No” to the question “Is this child diagnosed with any neurodevelopmental disorder?” were designated NT. However, verification of NT status through further psychological testing of children enrolled through this process was not conducted. NT subjects enrolled through the CARD were children of CARD employees, all of whom were trained observers and aware of their children’s development. In this respect, all children designated as NT met all developmental milestones, had no symptoms of ASD, ADHD (attention deficit hyperactivity disorder), or (other) learning disabilities, and were seen to be performing successfully in school or preschool with normal peer play. Opportunities for misclassification, therefore, were minimal. The possibility exists that siblings may differ from unrelated controls from the same source population in their genetic contribution to specific inherited disorders such as those affecting porphyrin metabolism. We note in this regard that genetic variation in porphyrin metabolism, particularly that affecting urinary porphyrin excretion, is exceedingly rare especially within the U.S. population, affecting, in the case of the most prevalent form, < 1 in 100,000 individuals (0.001%) (Health Grades, Inc. 2010
). Although the absence of differences in genetic variance in related and unrelated NT subjects in this study was not verified, it is unlikely that siblings and unrelated controls differed significantly in this respect.
Procedures for urine collection and measurement of urinary porphyrins, Hg, and creatinine concentrations
Urine samples (~ 50 mL, first or second morning voids when possible) were collected by parents/caregivers in clean glass containers and then transferred to Nalgene Nunc 60-mL, wide-mouth polyethylene bottles with screw-on lids (Item 2106–0002; Fisher Scientific, Seattle, WA). Samples were delivered frozen to the ARI, where they were logged, assigned an ID number, and shipped in batch in frozen ice packs by overnight express service to the University of Washington. A comparable protocol was followed by the CARD. For analyses, a 10-mL aliquot was removed and acidified with 1 N HCl for Hg analysis by continuous-flow, cold-vapor spectrofluorometry (Pingree et al. 2001b
). Porphyrins were quantified in the remaining unacidified portion of the urine sample by HPLC-spectrofluorometric analysis, as previously described (Bowers et al. 1992
; Woods et al. 1991
). Urinary creatinine concentrations were also measured in unacidified urine using a standard colorimetric procedure (Sigma, St. Louis, MO, USA). Urinary porphyrin concentrations were first creatinine adjusted (nanomoles per gram) and then transformed using the natural logarithm because of the wide variation and skewed distribution. Hg values below the detection limit (LOD) (0.02 μg/L) were assigned
Statistical analyses were conducted using PASW Statistics 17.0 (formerly SPSS) (Chicago, IL). Descriptive assessments first eliminated statistical outliers (values ≥ 3 SD in both directions) and then used cross-tabulations and one-way analysis of variance (ANOVA) procedures to compare nonchelated children from the three confirmed diagnostic groups [AU (n = 64), PDD-NOS (n = 19), and NT (n = 114)] with regard to sex, age, potential sources of Hg exposure, and mean (± SD) urinary Hg and porphyrin levels. The small number of five females in the AU group precluded subsequent statistical analyses for each sex. Thus, we examined potential determinants of diagnostic status among only the 133 male children (59 AU, 15 PDD-NOS, and 59 NT).
In males, logistic regression models that controlled for age initially tested potential associations between diagnosis and sources of exposure to Hg from the number of dental amalgam tooth fillings in the child and the mother, the number of vaccines that the child was reported to have received, the number of fish meals per month, and urinary Hg concentrations. The mean (± SD) of each porphyrin was also stratified by diagnosis, age, and sex, where an ANOVA F-test was applied separately for each sex to identify statistically significant differences between diagnostic groups.
Logistic regression analyses were also used to evaluate potential associations among males between porphyrins and the risk of having a diagnosis of AU or PDD-NOS, using NT as controls. Statistical measures included regression coefficients, their SDs, and estimates of the strength of association expressed as odds ratios (ORs) and 95% confidence intervals (CIs) for each porphyrin model. A statistically significant association was accepted if p < 0.05. Apart from the expected effect of age and age-squared, which were retained in final models, the only covariate that approached statistical significance was a restricted diet (p < 0.065). This variable was more reasonably attributed to response to diagnosis rather than to etiology or causal association and therefore was not retained in the analyses. The analyses also evaluated the combination of the three lesser carboxyl porphyrins (hexa-, penta-, and copro-) for potential association with AU or PDD-NOS. Urinary porphyrin concentrations corrected for creatinine (nanomoles per gram), along with the natural logs of these values, were tested in the analyses.