The Uncertainty of Hair Analysis for Trace Metals

In 1985, an article by Barrett¹ published in THE JOURNAL addressed several issues related to the use of trace metal hair analysis for assessing nutritional status. Now, in this issue of THE JOURNAL, Seidel and colleagues² show that many of same concerns raised 15 years ago by Barrett's study remain unresolved. Both studies used a self-prepared split sample that was sent to several laboratories specializing in trace metal analysis of human hair. Both noted the divergence of results obtained and the health-related claims made. Seidel et al also commented on the current regulatory environment that hair analysis laboratories now face.

Accuracy of laboratory tests is best ensured through the use of standards of known value to calibrate tests. Internationally recognized groups, such as the National Institute of Science and Technology, provide standard materials of known value, and other organizations, such as the US Environmental Protection Agency, provide specifications for these standards. Unfortunately, there are few primary standards for clinical laboratory tests. This is, in part, because of the complex interactions between the proteins (structural, functional, or immunological) and molecules of interest.³ Hence, clinical laboratory tests most commonly are standardized through the use of split sample results analyzed using compatible methods assessed through the proficiency testing (PT) process.

In 1948, Sunderman⁴ was the first to present the results of PT for common clinical analytes, and by the mid-1960s, laboratory use of PT samples to ensure standardization was commonplace. The Clinical Laboratory Improvement Act (CLIA) of 1967 required all US laboratories engaged in interstate commerce to use PT, and in 1988, the CLIA was amended to require PT for all laboratories except those doing simple (waived) testing.⁵ The widespread use of PT has helped to drive laboratories toward these interchangeable results. As noted by Seidel et al, PT does not exist for trace metal hair analysis. Under CLIA regulations, if PT is not available, the laboratory must use other methods to ensure the accuracy of the test, perhaps by splitting samples between peer laboratories at least twice a year.⁵ As noted, all of the laboratories used for the study by Seidel et al claimed CLIA certification, although one perhaps falsely, and no note was made of the methods used to verify the accuracy of test results among laboratories.

Laboratorians express the dispersion of results found in PT through the coefficient of variation (CV = SD/mean) expressed as a percentage. For most common analytes, interlaboratory CVs for PT are less than 10% and usually less than 5%.⁶ For more exotic analytes, such as follicle-stimulating hormone, human growth hormone, or luteinizing hormone, level-dependent interlaboratory CVs between 15% and 50% are common.⁷ Based on the data presented by Seidel et al, the interlaboratory CVs were calculated as 9.8% for sulfur for the lowest and 238.1% for phosphorous for the highest. The mean CV was 103.5%, indicating that a factor of 4 encompasses 95% of the results, which easily explains why the authors have little confidence in the laboratory results.

Biological specimens, such as those containing the trace metals measured in this study, frequently are subject to contamination, especially when the analyte is ubiquitous. The exposure of hair to the environment^{8 - 9} and the chemical treatment of hair for cosmetic purposes^{10 - 11} ensure that contamination is a frequent occurrence. No consensus treatment of samples has been developed for removing such contamination^{12 - 14} ; some reports indicate that the total removal of contaminants may be impossible.¹⁵ Seidel et al indicate in their study that all laboratories except one performed a variety of methods in attempting to remove external contamination, although complete details were not provided. It is likely that some of the deviations of results may be related to the effectiveness of these methods attempted by laboratories to remove the external contaminants.

Proteins are known for their ability to bind both small molecules and trace metals. For an analysis to be reproducible, all of the material must be in a form that can be measured, a factor laboratorians refer to as "recovery." For hair analysis by the 2 methods used in the study by Seidel et al, the trace metals must be free of all protein and occur in a form that can be readily ionized as required for measurement. No discussion is given as to the various digestion methods used to prepare the sample, but again, no consensus exists among laboratories, and diverse methods may yield disparate results.^{16 - 17}

Moreover, the variability of the submitted sample may influence results. The authors discuss the preparation of their sample but give no indication of homogeneity. Homogeneity of PT material is difficult to obtain¹⁸ and even more so in solid material such as hair. For solids, a well-mixed powder is generally considered the best form for distribution.¹⁹ The authors most likely chose not to distribute the samples as powder, which differs greatly from the usually submitted specimens. Also, it is unclear if the distribution of heavy metals in the hair shaft is uniform from shaft to shaft or within the shaft. It is easy to envision the scenario in which pockets of trace metals collect in areas that differ from hair to hair and within the hair shaft, and thus, results in a heterogeneous sample, again perhaps explaining the differences observed among the laboratories.^{20 - 21}

Hence, reasonable explanations exist for the observed extreme disparity in interlaboratory results that are independent of the quality of the laboratory methods. Does this help practicing clinicians in choosing a laboratory if they wish to recommend hair analysis? The answer is no. Until laboratorians reach an agreement on how to standardize both methods and materials for hair analysis, the large differences among laboratory results will persist. The call by Seidel et al for increased regulation by the Health Care Financing Administration (HCFA) will not alone resolve the problem. Regulation is most effective when the profession identifies one or more correct solutions that can be recommended. Clearly this does not exist and has not existed for several decades for hair analysis. In considering a laboratory for hair analysis, clinicians must understand that a standard method does not exist. A sample for hair analysis is best sent first to a laboratory that can validate its certification or accreditation for performing the test and that reports test characteristics, such as the hair-washing method, digestion techniques, recovery rates for the elements, internal quality control performance over time, and the minimum detection limits for each element.

In addition, the study by Seidel et al focuses on the use of hair for trace metal determination and the utility for nutritional assessments. Hair analysis may have some value in determining toxic exposure of heavy metals, especially over time, but analyses of other specimen types, such as urine, are more reliable and should be considered first.²² Hair also is becoming an accepted specimen for identifying some drugs of abuse, although some of the same issues of standardization and interpretation of results remain, but to a lesser extent.²³

Ultimately, the more important question is does hair analysis have a place in routine medical practice for assessment of nutritional status? Busy practitioners may have to address issues related to hair analysis in the short time that they have with a patient. In addition to traditional issues about appropriate test use, patients may have evaluated hundreds of Internet sites for hair analysis as well as an interpretation of their own results obtained from laboratories providing the service.

Considering that current laboratory methods make it difficult, if not impossible, to separate endogenous from exogenous material, it is currently unclear which of these each laboratory actually measures. Until laboratorians are sure they are measuring endogenous material present in the sample, it is impossible to relate measurements of individual elements, ratios of elements, or repeat testing over time to biological conditions. All results must be interpreted in relation to the laboratories' reference range for the analyte. At this time, however, it is impossible to relate results to a "normal person" since this person is difficult to describe with respect to hair analysis methods, as indicated by the diversity in reference ranges and laboratory interpretations of "normalcy," as noted by Seidel et al. Results depend on where patients live (environmental factors), the biological characteristics of their hair (such as color, thickness, age), and the cosmetic treatment of the hair (such as external coloring, type of hair spray, hair gel).

Without an appropriate reference range for a normal person's hair analysis, it is difficult to determine the clinical significance or nutritional importance of subtle changes in trace metal values in hair. Similar positions have been taken by the American Institute of Nutrition/American Society for Clinical Nutrition²⁴ and the American Medical Association.²⁵ Patients should be advised of the economic consequences of purchasing vitamin or other nutritional supplements from either the laboratory or the practitioner ordering the hair analysis test and the uncertainty of any possible medical consequences based on hair analysis laboratory values.

Physicians and other health care professionals who are considering ordering hair analysis to assess nutritional status or who are basing nutritional counseling or therapy on hair analysis results, should reconsider this approach unless and until the reliability of hair analysis value is established and evidence becomes available that clinical recommendations based on hair analysis improve patient outcomes.