Total and Free Thyroid Hormone in Serum

Because of the diverse clinical presentations of thyroid dysfunction, initial requests for assessment of thyroid function are often made by clinicians who, while alert to the possibility of thyroid dysfunction, may not be familiar with the interpretation of laboratory assays, or with the non-specific factors (200) or drug effects (128, 201) that can compromise current assays. Clinical decisions can be assisted by comments from the laboratory, tempered by a knowledge of the limitations of current immunoassays (201a, 202). The quality of this assistance will depend on two key components: the training and experience of the reporter and the available clinical information.

6.1. Historical

The competitive binding assays that are used for thyroid diagnosis were initially developed 30-35 years ago by clinical investigators who used 'in house' assay reagents and were often closely involved in patient care. This nexus between laboratory and clinical investigation allowed the deficiencies of the early assays to be readily appreciated. However, optimal diagnostic technology was available to only a minority of practitioners and patients, with results sometimes available only after long delay. In recent years there has been a strong trend away from 'in house' reagents which have been replaced by kits that incorporate highly sophisticated standardised reagents and automated instrumentation (eg solid phase antibodies, magnetic separation systems, chemiluminescent detection systems). Assay turnaround is much faster and up-to-date techniques are widely available to almost all practitioners, only a minority of whom have specific expertise in endocrinology or laboratory methodology. Nonspecialist users of endocrine assays are most likely to benefit from laboratory assistance in the interpretation of results, but as assay automation has increased, laboratorians have become more distant from the bedside. As clinicians receive less assistance, they may provide progressively less relevant information and vice versa. Laboratorians, in turn, can become frustrated by results that are uninterpretable or ambiguous without the relevant clinical background. Potential assay imperfections may be ignored simply because clinical correlation is impossible (see below).

6.2.Diagnostic approach to discordant or apparently anomalous results

The relationship between laboratory results and clinical findings may be either concordant or discordant. With discordant results, a distinction needs to be made between a previously unsuspected diagnosis, subclinical disease and anomalous assay results. If the discrepancy is confirmed, the fundamental assumptions that underpin the diagnostic use of tropic-target hormone relationships should be considered (see above).

a.	Re-evaluation of the clinical context, with particular attention to long-term features suggestive of thyroid disease and to the medication history.
b.	Measurement of serum TSH by a method that is sufficiently sensitive to identify the degree of TSH suppression.
c.	Authentic estimation of serum free T4, avoiding one-step methods that are known to give spurious results in euthyroid hyperthyroxinemia and during critical illness.
d.	Measurement of the serum T3 with appropriate binding correction.
e.	Measurement of serum total T4 to establish whether the serum free T4 is disproportionately high or low, possibly due to a preanalytical or method-dependent artefact.
f.	Follow-up to establish whether the abnormality is transient or persistent.
g.	Evaluation of propositus and family members for evidence of unusual binding abnormalities or hormone resistance.

In some circumstances, the cause of an apparently anomalous or unusual laboratory result becomes obvious from the clinical context (table 10). By taking into account the effects of drug therapy, or acute perturbation of the pituitary-thyroid axis, confidence in assay validity can be affirmed and unnecessary further investigation is avoided. Without the relevant information, an alert laboratorian would need to consider a possible assay anomaly or error.

Table 10. Situations in which interpretation of unusual laboratory results depends on relevant clinical information
*Clinical background*	*Assay Results*
	fT4	fT3	TSH
Pregnancy	L*	N	N
Antithyroid drug treatment, initial months	H,N,L	H,N,L	U
Recent T4 commencement for hypothyroidism	N	N	H
Hypothyroidism, appropriate T4 dose	H	N	N
Hypothyroidism, intermittent compliance	H,N		H
Appropriate T4 suppressive therapy	H	N	U,L
Recombinant TSH, suppressive T4	H	N	H
Hypopituitarism	L		L,N
Phenytoin	L*		L,N
Critical illness	L*	L	L
Heparin effect in critical illness	L,N,H*	L	L
Recovery phase of critical illness	L,N		H
Drugs that inhibit T4/T3 binding to TBG	L*	L*	L,N
Amiodarone effect in euthyroid subject	H	L	N
Acute T4 overdose	HH	H,N	N
Rheumatoid arthritis	N	N	H
U undetectable; L low; N normal; H high; * effect dependent on assay method

6.3 Clinical feedback, quality assurance and cost effectiveness

Clinical feedback will remain a key aspect of quality assurance in laboratory testing. While assay precision or reproducibility can be evaluated solely in the laboratory, diagnostic accuracy requires clinical correlation. As with any diagnostic test, the non-specificity of a procedure may not become apparent until the full diversity of the non-diseased population is appreciated (88). Premarketing evaluation of commercial assays may fail to include the critical samples that probe the diagnostic accuracy of an assay, so that non-specific interference is not appreciated until procedures have been in use for some time.

In studies of "cost effectiveness" it is hard to evaluate the human and financial costs that result from inappropriate management, needless duplication and unnecessary testing (77). Diagnostic inaccuracy of immunoassays may account for substantial unnecessary expenditure on laboratory resources (202). Dilution effects and binding protein abnormalities that affect free T4 estimation, the multiple effects of medications in non-thyroidal illness and the problems associated with thyroglobulin autoantibodies will continue to challenge both clinical chemists and clinicians. Further technological development will be no substitute for collaboration between these two groups. Consultation will also be required in the selection of thyroid tests that will best serve a particular population. For example, the multiple effects of severe illness on free T4 estimates may be of minor importance for a laboratory that serves predominantly ambulatory patients. A different assay profile will be required by a hospital laboratory that needs to exclude thyroid dysfunction during critical illness or in pregnancy.