Appendix C Session IV: Treatment Regimens & Their Effectiveness: A Meta-Analytic Review - Technical Report

The content in this publication was accurate at the time it was published, but it is not being updated. The item is provided for historical purposes only.​

By Marilyn Welsh, George Kelley, and Zung Vu Tran

Introduction

II. Method

III. Results

Introduction

Phenylketonuria

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, and this enzyme is essential for hydroxylation of dietary phenylalanine (Phe) to tyrosine in the liver (e.g., Guttler & Lou, 1986). Given normal intake of the amino acid Phe, severe mental handicap results (Jervis, 1939; as cited in Knox, 1972); however, for the past four decades, newborn screening programs have identified individuals with PKU (Knox, 1972) so that a diet low in Phe may be initiated early in development. If this 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 intellectual, cognitive, and behavioral domains do exist in this population of children and adults. For example, whereas early treatment appears to result in intellectual levels in the normal range, the IQ scores of children with PKU often do not achieve the levels predicted from the scores of unaffected parents and siblings (e.g., Williamson, Dobson, & Koch, 1977). In addition, some studies that have found that IQ declines with age during the school years suggest that such declines can be linked to elevations in Phe caused by poor or absent dietary control (e.g., Waisbren, Schnell, & Levy, 1980). Research exploring the cognitive skills of early-treated PKU children conducted in the 1970s and 1980s indicate 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 with 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 dysfunction (e.g., Chamove & Molinaro, 1978; Pennington, van Doorninck, McCabe, & McCabe, 1985; Welsh, 1996), and several lines of evidence are consistent with this hypothesis. First, the biochemical alterations resulting from the 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 PKU individuals (e.g., Krause, Halminski, McDonald, Dembure, Salvo, Fiedes, & Elsas, 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: lower nonverbal intelligence, impairments in novel problem solving, and impulsive behavior lacking in planning and goal-orientation. Moreover, studies specifically 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 (e.g., Diamond, Prevor, Callender, & Druin, 1997; Welsh, Pennington, Ozonoff, Rouse, & McCabe, 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 outcomes, has been the product of several comprehensive qualitative reviews (e.g., Waisbren et al., 1980; Welsh & Pennington, 2000). In what follows, an alternative, systematic and quantitative approach to reviewing the empirical research will be described and it’s potential value for understanding the vast literature examining early-treated PKU will be discussed.

Back to Top

Meta-Analysis

Glass (1976) developed and advocated an approach to research integration referred to as "meta-analysis." According to Glass et al., "... 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" (Glass et al., 1981; p.217). Meta-analysis may also be viewed "... as a more explicit approach to literature review, complementary to narrative review, not in opposition to it." (L’Abbe et al., 1987, p.224). Thus, meta-analysis does not exclude the need for the traditional narrative reviews. Rather, it should be viewed as a complementary statistical tool to summarize research findings.

At a time when there is such a proliferation of research published in many fields of science, meta-analyses are a vital necessity. It has been estimated that a general practice physician would need to read 19 original 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, by itself, inadequate to summarize and interpret this accumulated research knowledge.

When faced with dozens, hundreds, or even thousands of studies, it is impossible to integrate objectively all the research information without the aid of statistical methods. Investigators can no more understand the findings of 50 studies than they can, by inspection, make sense of 50 pieces of data from an experimental study. Thus, the essential character of meta-analysis is the application of statistical methods to the summary findings (outcomes) of many empirical studies.

Back to Top

Meta-Analysis of PKU Outcome Research

The preceding description of the meta-analytic perspective makes it clear that both qualitative and quantitative reviews of the literature are complementary approaches to achieving the goal of addressing research questions in a comprehensive, coherent, and explicit manner. The following characteristics of the empirical literature exploring early-treated phenylketonuria provide a particularly compelling argument for the use of meta-analysis:

  • There have been thousands of published articles in the past few decades examining such diverse issues as neuropsychological sequelae and molecular genetics.
  • Typically these studies involve small sample sizes, heterogeneous samples (e.g., in age), small effect sizes, and large variability in outcome. Each of these characteristics reduces the statistical power of an individual study (i.e., less chance of finding 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.
  • There exist 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 that one carefully and explicitly code each study with regard to the specific operational definitions of these variables. Studies then can be appropriately grouped (i.e., based on common definitions) for statistical analysis.

The meta-analytic approach to the review of 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, (5) evaluation and interpretation of the results in the form of a written report.

This particular meta-analytic project for the Consensus Development Conference on PKU, scheduled for October 16-18, 2000 begins with stages one and two already complete. Regarding stage one, four specific aims were to be addressed in this meta-analytic project:

  • What is the effect of treatment on cognitive and behavior outcomes?
  • What is known about age at 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 phenylalanine and tyrosine, and cognition and behavior outcomes at different age levels?

With reference to stage two, literature search, the NIH provided the bibliography compiled by Karen Patrias ( National Library of Medicine) and Felix de la Cruz ( NICHD). Thus, for the purposes of this report to the Consensus Development Conference, we have restricted the analysis to:

  • The NLM-provided bibliography.
  • 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 the correlation between some index of treatment (e.g., Phe level) and outcome (e.g., IQ).

Back to Top

Specific Questions Addressed

The four general aims of the project explore the degree to which the low-phenylalanine dietary treatment influences the psychological function of individuals diagnosed with PKU. In this regard, several behavioral outcomes, and specific processes within each outcome, were examined. These outcomes were grouped in the following manner:

  • Intelligence — Full-scale intelligence, verbal intelligence, performance intelligence
  • Executive Function — Planning, inhibition and flexibility, working memory and other unspecified executive processes
  • Attention — Sustained and selective attention
  • Cognitive and Motor Skills (other than executive function) — Memory (long-term, short-term, verbal, auditory), language, visual-spatial and perceptual-motor skills, motor skills, speed of processing
  • Behavior — Ratings of general behavior by parents/teachers, and self-concept, ratings of symptoms of psychological disorders (eg., depression, attention deficit)

Four overall aims, comprised of a total of 20 sub-questions, were the focus of this meta-analysis. Each question was addressed with regard to each of the outcome measures, above. The specific questions are listed below:

Aim One: What is the effect of treatment on cognitive and behavior outcomes?

Sub-question 1a

What are the differences between treated PKU individuals and unaffected (non-sibling) control individuals on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 1b:

What are the differences between treated PKU individuals and unaffected family control individuals (eg., siblings, parents) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 1c:

What is the association between age at diet initiation and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 1d:

What is the association between various indices of dietary control (eg., good vs poor treatment, % of phe levels above or below a certain level, etc.) and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 1e:

What are the differences between individuals with treated PKU and unaffected control individuals on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior at different ages?

Sub-question 1f:

For individuals with treated PKU only, what are the nature of the age differences (ie., cross-sectional) or age changes (ie., longitudinal) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Back to Top

Aim Two: What is known about the age at diet discontinuation on cognitive and behavioral outcome?

Sub-question 2a:

What are the differences between individuals with PKU who are currently on diet and those who are currently off diet on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 2b:

What are the differences between individuals with PKU who discontinued diet "early: (ie., prior to 8 years of age) and those who discontinued diet "late" (ie., after 8 years of age) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 2c:

What are the longitudinal changes that are observed in individuals with treated PKU, after diet termination, on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 2d:

What are the differences between individuals with early-treated PKU who are now all off diet and unaffected control individuals on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Back to Top

Aim Three: What is known about the reversibility of clinical symptoms upon the reinstitution of treatment at different ages?

Sub-question 3a:

What are the differences between individuals with early-treated PKU who are returned to diet and other individuals with early-treated PKU who remain off diet on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior? (between-subjects design)

Sub-question 3b:

What are the changes observed when diet is reinistituted for in individuals with early-treated PKU who have been off diet, on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior? (within-subjects design)

Sub-question 3c:

What are the effects of diet reinstitution at different ages for individuals with early-treated PKU on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Back to Top

Aim Four: What is known about the relationship of blood levels of phenylalanine (Phe) and tyrosine on cognition and at different ages?

Sub-question 4a

What is the association between concurrent (ie., with testing) Phe and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 4b:

What is the association between infant (ie., highest diagnostic level) Phe and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 4c

What is the association between average (ie., mean, median) lifetime Phe and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 4d:

What is the association between other measurements of Phe (eg., yearly, most recent, etc.) and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 4e:

What are the effects of the manipulation of Phe (ie., "phe loading") on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 4f:

What is the association between other measurements of various measurements of tyrosine and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Sub-question 4g:

What are the effects of the manipulation of tyrosine or l-dopa (ie., "supplementation") on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Back to Top

Method

Literature Searches

The articles considered for inclusion in this meta-analysis were derived from a bibliographybibliography compiled by the National Library of Medicine. The following data bases were searched for research articles dealing with PKU that were published between 1980 and July, 1999: MEDLINE, LOCATORplus (National Library of Medicine online catalog for books and audiovisuals), BIOETHICSLINE External Web Site Policy, POPLINE, EMBASE, PsycINFO, Sociological Abstracts, Social SciSearch, ERIC, AGRICOLA, Dissertation Abstracts, CINAHL, MANTIS, ExtraMED, Allied and Alternative Medicine, GPO, NTIS.

For information on the databases and resources not available here, please contact Dr. Felix de la Cruz at (301) 496-1383.

Criteria for inclusion

The inclusion criteria that guided our selection of those articles in the NLM bibliography that were relevant to this meta-analysis were as follows:

  • English language articles
  • Human participants
  • Any age participants (infancy through adulthood)
  • At least one group in the study must be diagnosed with classical PKU
  • At least one PKU group in the study must have been treated at some point in development
  • The study must include some index of dietary treatment (eg., age at initiation, age at termination, age at diet reinstitution; infant, mean lifetime, or current Phe level, effects of a Phe manipulation, etc.)
  • The study must include some behavioral outcome relevant to the overall aims of this project (ie, intelligence, executive function, attention, various cognitive and motor skills, typical and atypical behavioral characteristics).

In addition to the above inclusion criteria, several exclusion criteria allowed us to eliminate articles in the bibliography that did not provide data appropriate for this meta-analysis. These criteria were as follows:

  • Case studies (studies with N=1, where an average score on the outcome could not be calculated)
  • Letters, comments, and editorials that did not include relevant and appropriate data
  • Studies in which the only group represents individuals who are not diagnosed with classical PKU (eg., hyperpheneylalanemia, atypical PKU, individuals who heterozygous for the PKU mutation, etc)
  • Studies in which the only group is comprised of untreated PKU individuals
  • Studies with an exclusive focus on biochemistry, neurobiology, molecular genetics, neonatal screening, or epidemiology with no behavioral outcome data
  • Studies in which the only behavioral outcome is school achievement

Recorded variables

There were three categories of relevant information that was abstracted from each article and recorded in the codebook: Study Characteristics, Subject Characteristics, and Study Outcomes.

Within the Study Characterisics category, the following overall information about each study was recorded:

  • Study identification number
  • First author, journal name, and publication date
  • Country in which study was conducted
  • Total sample size
  • Gender composition of sample (all female, all male, mixed group)
  • Age group of sample (all children under 21 years, all adults over 21 years, mixed group)
  • Date coded, time to code, and coder's initials

Within the Subject Characteristics category, the following information relevant to the experimental group (one or more PKU groups) and the control group (one or more non-PKU groups) was recorded:

  • Experimental group age mean, standard deviation, and range
  • Control group age mean, standard deviation, and range
  • The mean and standard deviation of three Phe levels for the experimental group infant (ie., maximum diagnostic) level, mean lifetime level, and level concurrent with testing
  • Mean and standard deviation of the tyrosine level, and the nature of this assessment (eg., lifetime, concurrent)
  • Mean and standard deviation of the measure of economic status for both the experimental and control groups, as well as the scale used
  • Mean age of diet initiation, termination, and reinstitution for the experimental group.

For each outcome of each study included in the meta-analysis, the following Study Outcome information was recorded:

  • The specific focus of the analysis (eg., PKU group vs unaffected control group on full-scale intelligence)
  • The nature of the Phe assessment associated with that particular outcome; that is, infant, mean lifetime, concurrent, or "other" (eg., mean for the year during which testing occurred, most recent level, highest/lowest level during a period of time, etc)
  • Mean and standard deviation of the Phe level for the experimental and control groups (Note: this would be relevant for analyses in which two PKU groups are compared)
  • Level of dietary control, labeled as "best," "moderate," and "poor" based on the operational definitions of these categories by the study author(s)
  • The behavioral outcome was recorded in terms of three characteristics the general behavioral category (eg., intelligence, executive function), the more specific behavioral category (eg., IQ-full-scale intelligence, EF-planning), and the particular instrument used (eg., WISC-R, Tower of Hanoi)
  • Mean and standard deviation of the age at which diet was discontinued for the experimental (PKU) group or for the two PKU groups in the comparison
  • Discontinuation status referred to whether discontinuation was early (prior to age 8 years) or late (8 years or later)
  • The total sample size, and the sample sizes for the experimental and control groups
  • The type of value from which the effect size was calculated was recorded as mean, correlation, p-value, t-ratio, F-ratio, or percentage
  • The relevant data for that outcome means and standard deviations for the experimental and control groups, correlation, exact p-value, t-ratio, F-ratio, or percentage
  • The specific aim and specific question within that aim addressed by that outcome.

Screening and Coding of the Studies

The initial NLM bibliography included 2,346 titles/abstracts of articles published between 1980 and 1999. Three trained research assistants screened approximately 750 abstracts each with the objective of identifying all articles in which there appeared to be some behavioral assessment of individuals diagnosed with PKU and treated with the dietary regimen. This screening was done in a "liberal" manner, such that relevant articles would not be missed (eg., if only a title was given in the original bibliography, the full article was checked prior to elimination). This screening phase resulted in approximately 700 articles identified for further inspection. The research assistants retrieved these articles from local libraries and inter-library loan services, and made the decision whether each would be retained or eliminated based on the inclusion and exclusion criteria listed earlier in this report. This inspection phase culminated in the identification of 143 articles that were coded for the meta-analysis.

The coding phase involved 143 studies and 3,826 outcomes; however, 45 studies and 2,383 outcomes were excluded prior to the analysis due to insufficient data to calculate an effect size. This "insufficient data" typically involved studies in which means were reported without standard deviations (eg., in a figure) or studies in which there was no control data for comparison with the data from PKU individuals. Therefore, the meta-analyses were derived from 98 studies that yielded 1,543 effects sizes for 19 different outcomes and 20 research questions.

Effect size calculation

The individual effect sizes were calculated using of the following approaches as suggested by Hedges and Olkin (1985):

  • Taking the difference between the experimental and control group means divided by the pooled standard deviation
  • Converting the correlation (r) to g
  • Converting exact p value to g
  • Converting t-ratio to g
  • Converting F-ratio to g
  • Converting proportions to g

All data were corrected for small sample sizes (d) as suggested by Hedges and Olkin (1985); our final metric for the effect size was Hedges d, instead of g. The variance for each effect size was also calculated using procedures suggested by Hedges and Olkin.

Back to Top

Results

Data Analyses

A total of 356 analyses were conducted in which each of the four aims and their accompanying sub-questions were addressed with regard to each of the behavioral outcomes of interest. Of these analyses, 236 were eliminated due to an insufficient number of effect sizes; ie., less than 5. Three types of analyses were conducted, depending on the nature of the specific question addressed: summary analysis, ANOVA, and simple bivariate regression.

Summary analyses were conducted using the random-effects model described by Hedges and Olkin (1985) and was appropriate for two-group comparisons (eg., PKU vs unaffected controls). The relevant data provided by this analysis were mean effect sizes (d) and a 95% confidence interval for each effect size. The ANOVA approach to analysis was utilized for multiple group comparisons (eg., comparison of different levels of dietary control) and provided a between-groups effect, as well as the mean effect size for each set of comparisons. The simple regression analyses addressed questions regarding the association between an index of dietary treatment (eg., mean Phe level) and a behavioral outcome and yielded the following statistics: r, r-squared, and an exact p value.

Back to Top

Aim One Analyses

What is the effect of treatment on cognitive and behavior outcomes

Sub-question 1a:

What are the differences between treated PKU individuals and unaffected (non-sibling) control individuals on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Table 1 displays the total number of outcomes (N), mean effect size (d), and 95% confidence interval for the mean effect size relevant to this question. The vast majority of behavioral outcomes could be analyzed, and only two outcomes, the working memory sub-category of Executive Function and the language sub-category of the Other Cognitive and Motor Skills, were excluded. These cognitive outcomes were not analyzed due to an insufficient number of effect sizes (ie., less than 5).

These analyses explore the difference between individuals diagnosed with PKU and treated early in life with the low-Phe diet and unaffected, non-familial control participants on a variety of behavioral and cognitive outcomes. In these cases, a positive effect size reflects a result in which the PKU group performed more poorly than the control group, on average.

Sub-question 1b:

What are the differences between treated PKU individuals and unaffected family control individuals (eg., siblings, parents) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Table 2 displays the total number of outcomes (N), mean effect size (d), and 95% confidence interval for the mean effect size relevant to this question. Several cognitive and behavioral outcomes could be analyzed; however, the overall and more specific levels of Executive Function and Attention could not be examined due to insufficient N (i.e, less than 5 outcomes). This was also the case for most of the sub-categories of the Other Cognitive and Motor Skills category, with the exception of visual-spatial skills.

These analyses explore the difference between individuals diagnosed with PKU and treated early in life with the low-Phe diet and unaffected, familial control participants (eg., siblings, parents) on a variety of behavioral and cognitive outcomes. In these cases, a positive effect size reflects a result in which the PKU group performed more poorly than the control group, on average.

Anaylses combining sub-question 1a and 1b:

What are the differences between treated PKU individuals and unaffected control individuals (eg., familial and non-familial controls) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

These analyses collapsed over the type of unaffected control group (familial and nonfamilial), and explored the differences between early-treated PKU individuals and this broadly defined comparison group. A subset of those analyses displayed in Table 3 (overall attention, sustained and selective attention, memory, and speed of processing) was identical to the analyses conducted for sub-question 1a. Again, a positive effect size indicates poorer performance in the early-treated PKU group relative to the unaffected control groups.

Sub-question 1c:

What is the association between age at diet initiation and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Simple bivariate regression analyses were conducted to examine the associations between age at the start of the low-phenylalanine diet and behavioral and cognitive sequelae. Table 4 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R 2, and the exact significance level ( (p) of the analysis. Only the overall category of IQ, and the specific subcategory of full-scale IQ, could be analyzed to address this question. All other cognitive and behavioral categories suffered from a small number of outcomes (N less than 5). In this table, a negative correlation indicates that as age at diet initiation increased, performance on the behavioral and cognitive measures decreased.

Sub-question 1d:

What is the association between various indices of dietary control (eg., good vs poor treatment, % of phe levels above or below a certain level, etc.) and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

These analyses were conducted using data from studies in which there was an explicit identification of the quality of dietary control for the PKU participants, although the procedures by which this identification was carried out differed across studies. In addition, the operational definitions of "best control," "moderate control," and "poor control" also varied across studies. One-Way Analysis (ANOVA) was conducted to examine differences in the effect size associated with a variety of dietary control group comparisons among the three main dietary control groups, "best", "moderate" and "poor."

Studies of IQ provided data relevant to the following four groups: (1) outcomes that addressed the effect of comparing across all three dietary control groups, (2) outcomes that examined the effect of comparing the moderate and best dietary control groups, (3) outcomes that addressed the effect of comparing the poor and moderate dietary control groups, and (4) outcomes that explored the effect of comparing the poor and best dietary control groups. A One-Way ANOVA revealed a significant between-groups, Q=12.08, probability (Chi-Square) = 0.007, suggesting a difference in effect size associated with each of the four dietary control comparisons. Table 5 displays the average effect size for each of four groups, and a positive effect size reflects the observation of lower average IQ scores in the group with a lower level of dietary control.

The same analysis of full-scale IQ scores, specifically, resulted in a very similar between-groups effect, Q=11.56, probability (Chi Square) = 0.009. As seen in Table 6, the mean effect size for each of the four comparison groups is also very similar.

With regard to sustained and selective attention processes, studies provided data relevant to two comparison groups: (1) outcomes that examined the effect of comparing a best dietary control group with unaffected controls, and (2) outcomes that addressed the effect of comparing the poor and best dietary control groups. A One-Way ANOVA indicated a significant between-groups effect, Q=5.357, probability (Chi Square) = 0.02. Table 7 illustrates the mean effect size for each of the two dietary control group comparisons, and again, a positive effect size indicates that the participants in the poor dietary control performed more poorly on attention tasks.

Sub-question 1e:

What are the differences between individuals with treated PKU and unaffected control individuals on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior at different ages?

There was too few outcomes (ie., less than five) to conduct these analyses addressing whether there are differences between treated PKU individuals and healthy, normal controls at different ages, or within small ranges of age.

Sub-question 1f:

For individuals with treated PKU who are currently on diet, what are the nature of the age differences (ie., cross-sectional) or age changes (ie., longitudinal) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Table 8 presents the summary analysis data relevant to this question, including the total number of outcomes (N), the mean effect size (d), and the 95% confidence interval around this mean. The overall behavioral categories of IQ, Executive Function, and Other Cognitive and Motor Skills, as well as full-scale IQ and the inhibition/flexibility sub-category of Executive Function, could be analyzed to address this question. The remaining overall categories (ie., Attention and Behavior) and their associated sub-categories involved too few outcomes (ie., less than 5) to be analyzed. A positive effect size indicates improvement in the cognitive skill with age for those PKU individuals currently on diet.

What is known about the age at diet discontinuation on cognitive and behavioral outcome?

Back to Top

Aim Two Analyses

Sub-question 2a:

What are the differences between individuals with PKU who are currently on diet and those who are currently of diet on meausures of intelligence,executive function, attention, other cognitive and motor skills, and behavior?

Table 9 presents the summary analysis data releveant to this question, including the total number of outcomes (N), the mean effect size (d), and the 95% confidence interval around this mean. The overall behavioral categories of IQ, Attention, Behavior, an Other Cognitive and Motor Skills were analyzed; however, there was insufficient N (ie, less than 5 outcomes) to analyze the Executive Function category. In addition, the sub-categories of full-scale IQ, sustained and selective attention and speed of processing were analyzed. A positive effect size reflects poorer cognitive and behavioral outcomes for those PKU individuals who terminated diet as compared to those PKU individuals who are currently on diet.

Sub-question 2b:

What are the differences between individuals with PKU who discontinued diet "early: (ie., prior to 8 years of age) and those who discontinued diet "late" (ie., after 8 years of age) on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Table 10 presents the following summary analysis data relevant to this question: the total number of outcomes (N), the mean effect size (d), and the 95% confidence interval around this mean. Only one overall behavioral category, IQ, involved a sufficient number of outcomes (ie., five or greater) to be analyzed. The three sub-categories of IQ, full-scale, verbal, and performance scores, also were analyzed. In this analysis, a positive effect size reflects that PKU individuals who terminated diet prior to age 8 years exhibited a poorer outcome with regard to intelligence than those PKU individuals who terminated diet after age 8 years.

Sub-question 2c:

What are the longitudinal changes that are observed in individuals with treated PKU, after diet termination, on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Table 11 presents the following summary analysis data relevant to this question: the total number of outcomes (N), the mean effect size (d), and the 95% confidence interval around this mean. Only two overall behavioral categories, IQ and Other Cognitive and Motor Skills, entered into this analysis due to a sufficient number of outcomes (ie., five or greater). Additionally, one sub-category of IQ, full-scale IQ, was analyzed. A positive effect size represents a decrease with age in performance on intelligence or cognitive tests, after diet termination.

Sub-question 2d:

What are the differences between individuals with early-treated PKU who are now all off diet and unaffected control individuals on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

The summary analysis relevant to this question is presented in Table 12, and includes the total number of outcomes (N), the mean effect size (d), and the 95% confidence interval around this mean. The following behavioral categories involved a sufficient number of outcomes to be analyzed with respect to this question: IQ (exclusively full-scale IQ scores), the overall category of Attention, and the speed of processing sub-category of Other Cognitive and Motor Skills. A positive effect size suggests that the early-treated PKU individuals who are currently off diet exhibit poorer task performance than unaffected, control individuals.

Back to Top

Aim Three

Analyses

What is known about the reversibility of clinical symptoms upon the reinstitution of treatment at different ages?

Sub-question 3a:

What are the differences between individuals with early-treated PKU who are returned to diet and other individuals with early-treated PKU who remain off diet on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior? (between-subjects design)

There was not a sufficient number of outcomes (ie., at least 5) to conduct an analysis comparing PKU individuals returned to diet with another group of PKU individuals who remain off diet.

Sub-question 3b:

What changes are observed when off-diet PKU individuals experience diet reinstitution, with regard to measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior? (within-subjects design)

The summary analysis relevant to this question is presented in Table 13, and includes the total number of outcomes (N), the mean effect size (d), and the 95% confidence interval around this mean. Only one category, Behavior, provided a sufficient number of outcomes (ie., at least 5) for this analysis. Specifically, the sub-category of psychological disorders was the target outcome of the analsysis. A positive effect size reflected improved behavior (or less indication of a psychological disorder) after diet reinstitution.

Sub-question 3c:

What are the effects of diet reinstitution at different ages for individuals with early-treated PKU on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Given the relatively small number of studies providing outcome data on the topic of diet reinstitution, there were well below the number of outcomes needed (ie., at least 5) to address the question of the effect of diet reinstitution on behavior at particular ages.

Aim Four

Analyses

What is known about the relationship of blood levels of phenylalanine (Phe) and tyrosince on cognition and at difference ages?

Sub-question 4a:

What is the association between concurrent (ie., with testing) Phe and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Simple regression analyses were conducted to examine the association concurrent Phe level and behavioral and cognitive outcome. Table 14 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R 2, and the exact significance level ( p) of the analysis. The vast majority of behavioral categories and sub-categories entered into these analyses, with the exception of the sub-categories of verbal and performance IQ, and language (Other Cognitive and Motor Skills). In this table, a negative correlation reflects the inverse association between Phe level concurrent with testing and scores on the cognitive and behavioral measures.

Sub-question 4b:

What is the association between infant (ie., highest diagnostic level) Phe and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Simple regression analyses were conducted to examine the association between the highest diagnostic Phe level prior to diet intitation (i.e., infant Phe) and cognitive outcome. Table 15 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R 2, and the exact significance level (p) of the analysis. Only one outcome category, overall IQ, met the criterion of five or more effect sizes for analysis. A negative correlation reflects an inverse association between infant Phe level and IQ test scores.

Sub-question 4c:

What is the association between average (ie., mean, median) lifetime Phe and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

Simple regression analyses were conducted to examine the association between the mean or median Phe level from birth to the time of testing (ie., "lifetime") and cognitive outcome. Table 16 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R 2, and the exact significance level (p) of the analysis. The majority of behavioral categories and sub-categories could not be analyzed due to a limited number of relevant outcomes (ie., less than 5). However, the overall categories of IQ, Executive Function, and Other Cognitive and Motor Skills, as well as the sub-category of full-scale IQ, could be analyzed. For these analyses, a negative correlation reflects the inverse association between average lifetime Phe level and scores on the cognitive and behavioral measures.

Sub-question 4d:

What is the association between other measurements of Phe (eg., yearly, most recent, etc.) and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

The association between other measures of Phe level and behavioral and cognitive outcome was examined via simple regression analyses. Table 17 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R2, and the exact significance level ( p) of the analysis. All of the overall categories provided sufficient data to address this question. The following sub-categories of cognition and behavior could not be analyzed: verbal and performance IQ, the working memory sub-category of Executive Function, and the language, motor, and speed of processing sub-categories of Other Cognitive and Motor skills. For these analyses, a negative correlation reflects the inverse association between this diverse collection of measures of Phe level and scores on the cognitive and behavioral measures.

Sub-question 4e:

What are the effects of the manipulation of Phe (ie., "phe loading") on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

The question of how experimentally increasing Phe level through "Phe loading" procedures relate to changes in cognitive and behavioral outcome was addressed through simple regression analysis. Table 18 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R2, and the exact significance level ( p>) of the analysis. All four overall behavioral categories could be analyzed, as well as all sub-categories, with the exception of verbal and performance IQ, working memory (Executive Function), and language, motor, and speed of processing skills (Other Cognitive and Motor Skills). In this table, a negative correlation indicates that as Phe level increases via this manipulation, performance in these cognitive and behavioral domains decreases.

Sub-question 4f:

What is the association between other measurements of various measurements of tyrosine and measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

There were very few studies reporting tyrosine levels of the PKU participants, and thus an insufficient number of outcomes (ie., less than 5) to analyze this question.

Sub-question 4g:

What are the effects of the manipulation of tyrosine or l-dopa (ie., "supplementation") on measures of intelligence, executive function, attention, other cognitive and motor skills, and behavior?

The question of how experimentally increasing dopaminergic levels via supplementation with its neurochemical precursor, L-dopa, or with amino acids, such as tyrosine, relate to changes in cognitive and behavioral outcome was addressed through simple regression analysis. Table 19 presents the following results of the regression analyses: N (number of effect sizes), the correlation or R, the percent variance explained or R2, and the exact significance level (p) of the analysis. Three overall behavioral categories, Executive Function, Attention, and Other Cognitive and Motor Skills, as well as the inhibition/flexibility sub-category of Executive Function, involved a sufficient number of outcomes to be analyzed.. In this table, a positive correlation indicates that supplementation with dopamine precursors is associated with better performance on these cognitive tasks.

Back to Top

first | previous | next | last

top of pageBACK TO TOP