By Marilyn C. Welsh, Ph.D., George Kelley, Ph.D., and Zung Vu Tran, Ph.D.
Phenylketonuria (PKU) is a well-known genetic cause of mental retardation, affecting one in 10,000 to 20,000 live births (Stanbury, Wyndgaarden, Friedrickson, 1983). The disorder is the consequence of mutations in the gene that codes for the enzyme phenylalanine hydroxylase (PAH). This enzyme is essential for hydroxylation of dietary phenylalanine (Phe) to tyrosine in the liver (Güttler, Lou, 1986). Given normal intake of Phe, severe mental handicap results (Jervis, 1939; cited in Knox, 1972). For the past four decades, however, newborn screening programs have identified individuals with PKU (Knox, 1972) so that a diet low in Phe can be initiated early in life. If dietary treatment is begun early and controlled consistently during childhood, mental retardation is averted.
Even with early dietary treatment of PKU, research suggests that impairments in the intellectual, cognitive, and behavioral domains occur. For example, whereas early treatment appears to result in intellectual levels in the normal range, the intelligence quotient (IQ) scores of children with PKU often do not achieve the levels predicted from the IQ scores of unaffected parents and siblings (Williamson, Dobson, Koch, 1977). In addition, studies that have found that IQ declines with age during the school years suggest that such declines can be linked to elevated Phe caused by poor or absent dietary control (Waisbren, Schnell, Levy, 1980). Research in the 1970s and 1980s on the cognitive skills of early-treated children with PKU indicated a pattern of cognitive function characterized by intact speech and language skills but impaired visual-spatial, perceptual-motor, and problem-solving abilities (Welsh, Pennington, 2000). With regard to behavior, early-treated children do not present a consistent clinical profile; however, behavioral problems tend to cluster in the areas of hyperactivity, impulsivity, poor planning, and less task persistence (Welsh, Pennington, 2000).
Although the specific neuropathology of PKU has not been explicated fully, one hypothesis proposes that PKU produces a prefrontal cortex dysfunction (Chamove, Molinaro, 1978; Pennington, van Doorninck, McCabe, et al., 1985; Welsh, 1996), and several lines of evidence are consistent with this hypothesis. First, the biochemical alterations resulting from genetic mutation cause a disruption in catecholamine biosynthesis. One of these catecholamines, dopamine, is essential for prefrontal cortical function. There is empirical evidence of lower levels of central dopamine (McKean, 1972) and an inverse association between Phe levels and dopamine levels in individuals with PKU (Krause, Halminski, McDonald, et al., 1985). Impaired prefrontal cortical function due to diminished levels of dopamine is consistent with decades of cognitive and behavioral research demonstrating a particular profile of deficits in early-treated PKU. The profile includes lower nonverbal intelligence, impairments in novel problem-solving, and impulsive behavior lacking in planning and goal orientation. Moreover, studies designed to explore the prefrontal dysfunction model of PKU have found evidence of specific impairments in a cognitive domain (executive function) that has been linked to this brain region (Diamond, Prevor, Callender, et al., 1997; Welsh, Pennington, Ozonoff, et al., 1990).
Our current understanding of the intellectual, cognitive, and behavioral characteristics of early-treated PKU, as well as the efficacy of diverse dietary regimens with regard to these characteristics, is the product of several qualitative reviews (e.g., Waisbren, Schnell, Levy, 1980; Welsh, Pennington, 2000). In what follows, an alternative systematic and quantitative approach to reviewing the empirical research on PKU will be described, and its potential value for making sense of the vast literature on early-treated PKU will be discussed.
The Meta-Analysis Approach
Glass (1976) developed and advocated an approach to research integration referred to as "meta-analysis." According to Glass, McGraw, and Smith (1981), "it is nothing more than the attitude of data analysis applied to quantitative summaries of individual experiments. By recording the properties of studies and their findings in quantitative terms, the meta-analysis of research invites one who would integrate numerous and diverse findings to apply the full power of statistical methods to the task. Thus, it is not a technique; rather, it is a perspective that uses many techniques of measurement and statistical analysis." Meta-analysis can also be viewed "as a more explicit approach to literature review, complementary to narrative review, not in opposition to it" (L’Abbe, Detsky, O’Rourke, 1987). Thus, meta-analysis does not exclude the need for traditional reviews. Rather, it should be viewed as a statistical tool to summarize research findings.
At a time of great proliferation in published scientific research, meta-analyses are a vital necessity. It has been estimated that general practice physicians would need to read 19 journal articles a day, 365 days a year, just to keep their knowledge current (Anonymous, 1995). Given this large volume of research to be assimilated, the narrative method of research review—that is, studies chronologically and/or categorically arranged and described—is inadequate by itself to summarize this accumulated research.
Meta-Analysis of PKU Outcome Research
In other words, qualitative and quantitative reviews of the literature are complementary approaches to addressing research questions in a comprehensive and coherent way. The following characteristics of the empirical literature on early-treated PKU provide a particularly compelling argument for the use of meta-analysis:
- Number of articles. Thousands of articles have been published in the past few decades on neuropsychological sequelae and molecular genetics. Typically, these studies present small sample sizes, heterogeneous samples, small effect sizes, and large variability in outcome. Each of these characteristics reduces the statistical power of an individual study (i.e., reduces the possibility of statistically significant results). However, when many "low power" studies are combined and incorporated into a single analysis or small set of analyses, statistical power is substantially increased.
- Operational definitions. There is a wide variety of operational definitions of both the "independent variable" (i.e., treatment) and the dependent variables (e.g., intelligence, executive function, behavior). Meta-analysis requires the analyst to carefully and explicitly code each study with regard to the operational definitions of these variables. Studies then can be appropriately grouped for statistical analysis.
The meta-analytic approach to the literature typically involves five stages: (1) formulation of the questions to be answered, or specific aims; (2) execution of a complete literature search; (3) collection, classification, and coding of studies meeting inclusion criteria; (4) application of statistical techniques for pooling and analyzing the compiled data; and (5) evaluation and interpretation of the results in a written report.
The meta-analytic approach to the literature on PKU for the Consensus Development Conference on PKU has two stages. Regarding stage one, the specific questions to be addressed are as follows:
- What is the effect of treatment on cognitive and behavioral outcomes?
- What is known about the effects of age at time of diet discontinuation on outcome?
- What is known about the reversibility of clinical symptoms upon reinstitution of treatment at different ages?
- What is known about the relationship of blood levels of Phe and tyrosine to cognition and behavioral outcomes at different ages?
For the purposes of this report, we have restricted our analysis to the following:
- A bibliography compiled by the National Library of Medicine (NLM).
- Studies of IQ, cognitive processes, neuropsychological sequelae, executive function, and behavior.
- Studies that report sufficient data for the dependent variable so that a standardized difference effect size can be calculated (means, standard deviations, correlations, exact p values, etc.).
- Studies that report a correlation between some index of treatment (e.g., Phe level) and outcome (e.g., IQ).
Screening of the 2,346 abstracts listed in the NLM bibliography is under way. It is estimated that approximately 200 of them will meet inclusion criteria. Our objective is to apply meta-analytic methodology to summarize what is known and, when possible, to arrive at clear, definitive answers to issues of intellectual, cognitive, and neuropsychological performance in individuals with early-treated PKU. The relation of Phe level, dietary control, dietary termination, and dietary reinstitution to behavioral outcomes will also be evaluated.
- Anonymous. On the need for evidence-based medicine. EBM 1995;1:5-6.
- Chamove AS, Molinaro TJ. Monkey retardate learning analysis. J Ment Defic Res1978;22:37-48.
- Diamond A, Prevor MB, Callender G, Druin DP. Prefrontal cortex cognitive deficits in children treated early and continuously for PKU. Monogr Soc Res Child 1997;62(4, Serial No. 252).
- Glass GV. Primary, secondary and meta-analysis of research. ER 1976;5:3-8.
- Glass GV, McGraw B, Smith ML. Meta-analysis in social research. Thousand Oaks, CA: Sage Publications; 1981.
- Güttler F, Lou H. Dietary problems of phenylketonuria: effect on CNS transmitters and their possible role in behavior and neuropsychological function. J Inherit Metab Dis 1986;9:168-72.
- Knox WE. Phenylketonuria. In: Stanbury JB, Wyngaarden JB, Fredrickson DS, editors. The metabolic basis of inherited disease. New York: McGraw-Hill; 1972. p. 266-95.
- Krause WL, Halminski M, McDonald L, Dembure P, Salvo R, Friedes D, et al. Biochemical and neuropsychological effects of elevated plasma phenylalanine in patients with treated phenylketonuria: a model for the study of phenylalanine and brain function in man. J Clin Invest 1985;75:40-8.
- L’Abbe KA, Detsky AS, O’Rourke K. Meta-analysis in clinical research. Ann Intern Med 1987;107:224-33.
- McKean CM. The effects of high phenylalanine concentrations on serotonin and catecholamine metabolism in the human brain. Brain Res 1972;47:469-76.
- Pennington BF, van Doorninck WJ, McCabe LL, McCabe ER. Neuropsychological deficits in early treated phenylketonuric children. Am J Ment Defic 1985;89:467-74.
- Stanbury JB, Wyndgaarden JB, Friedrickson DS. The metabolic basis of inherited disease. 2nd ed. New York: McGraw-Hill; 1983.
- Waisbren SE, Schnell RR, Levy HL. Diet termination in children with phenylketonuria: a review of psychological assessments used to determine outcome. J Inherit Metab Dis 1980;3:149-53.
- Welsh MC. A prefrontal dysfunction model of early-treated phenylketonuria. Eur J Pediatr 1996;155(Suppl. 1):S87-9.
- Welsh MC, Pennington BF. Phenylketonuria. In: Yeates KO, Ris, MD,Taylor HG, editors. Pediatric neuropsychology: research, theory and practice. New York: Guilford; 2000. p. 275-99.
- Welsh MC, Pennington BF, Ozonoff S, Rouse B, McCabe ER. Neuropsychology of early-treated phenylketonuria: specific executive function deficits. Child Dev 1990;61:1697-713.
- Williamson ML, Dobson JC, Koch R. Collaborative study of children treated for phenylketonuria: study design. Pediatrics 1977;60:815-21.