American Association for Clinical Chemistry
Better health through laboratory medicine


Excerpts from the Literature

Articles of interest compiled by the editorial board. Please welcome our new member of the editorial board, Joely Straseski, Ph.D, DABCC, FACB.

Powdered blend of polyunsaturated fatty acids (PUFAs) supplementation normalizes docosahexaenoic acid (DHA) and arachidonic acide (AA) levels in patients with PKU.

Jans JJ etal, Mol Gen Metabol 2013, 109:121-124. (UG)

In untreated or poorly treated PKU patients, in addition to neurotoxicity by phenylalanine, it has been hypothesized that limited availability of docosahexaenoic acid (DHA, 22:6n-3) and arachidonic acid (AA, 20:4n-6) in the brain attribute to mental retardation and microcephaly. DHA and AA are structural components of neuronal cell membranes and are pivotal for brain development and retinal function. A typical PKU diet provides low saturated and polyunsaturated fat intake and have a poor LC-PUFA status in plasma and erythrocytes as compared to healthy controls. In addition, phenyalanine metabolites phenylpyruvate and phenyllactate inhibit the endogenous synthesis of DHA and AA, and thus lowering their plasma levels in PKU patients.

In this study, the investigators measured plasma and erythrocyte DHA and AA levels in 54 patients with PKU, after supplementation with fish oil or the fatty acid supplement KeyOmega (a powdered blend of DHA and AA).  In PKU patients concentrations of DHA and AA in erythrocytes and plasma were below the reference range. Although, supplementation with fish oil, DHA levels reached normal range, the AA concentrations did not reach to the normal values in these patients. In contrast, both DHA and AA levels increased and reached reference values upon supplementation with KeyOmega. The authors concluded that KeyOmega offers additional benefit over fish oil since both AA and DHA status are normalized in PKU patients supplemented with KeyOmega.

A systematic review of statistical methods used in constructing pediatric reference intervals.

Daly CH, Liu X, Grey VL, Hamid JS. Clin Biochem (2013). (JS)

Establishing reference intervals is often an arduous but critical step in new test development or validation. Pediatric populations present unique challenges that can make this task even more complicated. This article by Daly et al. discusses the need for a standardized statistical approach when establishing pediatric reference intervals. They review the current literature and investigate the different statistical methods used to construct pediatric reference intervals. The authors examine common reporting methods for the detection of outliers, partitioning into relevant groups, and statistical approaches to determining upper and lower reference limits. They also discuss concerns regarding sample size (more or less than 120), parametric vs. non-parametric statistics, and the importance of outlier detection and confidence limits.

Inclusion/exclusion criteria resulted in 22 original articles for their review. Striking variation was observed in the statistical methods used in these articles. Examples of methods include parametric, non-parametric, robust, parametric fractional polynomials and visual assessment of means and standard deviations. Fifty-nine percent of these studies used statistical methods that were outlined in the CLSI guidelines for reference interval determination. The vast majority of studies performed partitioning (96%), but not all of these tested for differences between the partitions (76%) or collapsed insignificant partitions (69%). Twenty four percent of studies that performed partitioning did not apply statistical methods to test whether the partitions were appropriate. The majority of studies did not perform tests to identify outliers in their data sets (59%). Less than 20% reported confidence intervals for their ranges.

The authors discuss the importance of providing confidence intervals and properly determining appropriate partitions and detecting outliers. Overall, they recommend the bootstrap method to determine confidence intervals by re-sampling of the data and the robust method for small sample sizes.
This topic is extremely important and the field will benefit from a discussion of, and guidelines for, the standardization of interval determinations in specific populations such as pediatrics. As large collections of pediatric reference interval data continue to be collected (e.g., CALIPER, CHILDx, KiGGS and the National Children’s Study), establishing standardized criteria specific to this challenging patient population will be critical to interpretation across studies.