Wrong biochemistry results

The success of analytical methods in clinical chemistry has led to a sense of security in the value of laboratory results. This is largely justified, as evidenced by the quality of laboratory performance assessed by external assurance schemes. Nevertheless, it is not widely recognised among clinicians that some biochemical tests are more prone to interference from unusual serum constituents than others—and that quality assurance schemes can do little about this.

An important example of this is tests carried out by immunoassays based on the recognition of molecules by antibodies. The antibodies are largely derived from animal sources and are typically used for measuring hormones, tumour markers, cardiac troponins, and therapeutic drugs and for viral serology.

The design of assays has evolved enormously since the discovery of immunoassay by Berson et al in 1956,¹ and it is now a major analytical tool in clinical laboratories worldwide, allowing relatively minute (picomole (10^-12)) amounts of analytes to be measured with unrivalled precision.

Interference in immunoassays by antibodies is a recognised phenomenon. For example, endogenous antithyroglobulin antibodies invalidate thyroglobulin measurements, and exogenously administered antibodies used to treat digoxin toxicity prevent measurement of plasma digoxin. However, there is more insidious interference due to the presence of unsuspected abnormal binding protein(s) in the patient.^2–4 These mainly include heterophilic antibodies such as rheumatoid factor, anti-animal antibodies (anti-mouse, anti-rabbit, anti-sheep, etc), and anti-idiotypic antibodies (antibodies elicited by an idiotope on another antibody molecule during the course of an immune response). In some cases these antibodies in patients' sera may interfere with the analytical reaction between the analyte being measured and the antibodies used in the immunoassay's cocktail. The exact effect of such interference will depend on the site where they interfere with the reaction, leading to falsely raised or lowered measurements. These interferences are specific to each patient, so only that patient's data will be affected, while quality assurance criteria for the assay will have been passed.

Examples of this type of interference that has had serious clinical consequences include human chorionic gonadotrophin assays.⁵ As a result of wrongly interpreted results six young non-pregnant women were aggressively treated with chemotherapy and surgery for non-existent “occult” trophoblastic disease.⁶ In our experience at a national reference steroid laboratory, samples with consistent and substantial increases in steroids using routine direct immunoassays have raised the possibility of disease but have subsequently been found to be normal after reassay using more robust techniques involving extraction procedures. In one case a raised oestradiol value led to a patient having a hysterectomy and bilateral oophorectomy, and only when no fall in oestradiol was seen postoperatively was the sample further analysed and the original result found to be wrong because of immunoassay interference. Similar problems are also noted in other steroid assays, such as testosterone in women.⁷ False positive interference in troponin assays in patients with chest pain due to acute coronary syndrome has led to prolonged inpatient stays and invasive investigation.⁸,⁹ False negative results are equally important in that they lead to underinvestigation.¹⁰

The presence of interfering antibodies is surprisingly common, affecting 30-40% of the population.³ They probably arise from mundane activities such as keeping pets, ingesting animal antigens, vaccination, infection, or even blood transfusion. Most analytical assays currently in use can neutralise and block low concentrations of these interfering proteins (μg to mg/l) with no or minimum impact on analytical accuracy. Larger amounts of interfering proteins, which may be as high as grams per litre, or proteins with high binding affinity can, however, overwhelm the analytical system, leading to assay interference and erroneous results. The number of these extreme cases is not known, though specific types of interference, such as heterophilic and anti-murine antibodies, in the order of 0.05% have been reported.⁴,¹¹ Our experience suggests that interfering antibodies of various types affect about 0.5% immunoassays (A Ismail, J Barth, unpublished data), though others have reported higher percentages.¹² Even the lowest prevalence quoted should be seen in the context of the total number of immunoassays—many millions a year in British hospital laboratories alone. Thus many thousands of patients in the United Kingdom might be affected. Furthermore, this problem is likely to worsen owing to the wider use of biotechnologies such as monoclonal antibodies and T cells for diagnostic and therapeutic purposes.¹³

Since these interferences are relatively uncommon, clinicians need to be aware of them and alert to the mismatch of clinical and biochemical data. They should not discard clinical evidence in favour of a numerical value. Moreover, this form of interference is not specific to a single analyte and may affect other immunoassays performed on the same patient in a different clinical setting. Thus patients who have such interference detected should have this fact documented in their clinical records, so that the results of future immunoassays can be viewed with caution.