Polyethylene glycol precipitation: proceed with care

  1. David Halsall2

  1. 1Department of Clinical Biochemistry, Southend Hospital, Westcliff-on-Sea, Essex SS0 0RY, UK

  2. 2Department of Clinical Biochemistry, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK
  1. Corresponding author: Mr M Fahie-Wilson. Email: mfahie-wilson{at}southend.nhs.uk

Two articles in this volume of the Annals1,2 demonstrate the utility of polyethylene glycol (PEG) precipitation for the detection of serum macro-analytes. In these cases,
elevated aspartate aminotransferase (AST) activity was due to the enzyme being sequestered in serum as a high molecular mass
(macro) enzyme complex rather than as a result of increased AST release by damaged tissue. PEG precipitation is now widely
used in UK laboratories to detect elevated serum prolactin concentration due to macroprolactin3 and familiarity with the technique and ready availability of the reagent may encourage laboratories to use this approach
for the investigation of other macro-analytes which can cause similar diagnostic confusion and patient mismanagement. PEG
precipitation undoubtedly has the potential for wider routine application and has been used to study interference from macro-analytes
or immunoglobulins in assays for various cytokines, C-terminal telopeptide of type I collagen, cardiac troponin I, α-fetoprotein, C-reactive protein, IgA, IgM, insulin, thyroid-stimulating hormone (TSH), parathyroid hormone, prolactin, luteinizing
hormone, follicle-stimulating hormone (FSH) growth hormone, total triiodothyronine, creatine kinase (CK), amylase, alkaline
phosphatase (ALP), AST, alanine aminotransferase (ALT) lactate dehydrogenase (LDH) and gamma-glutamyl transpeptidase (GGT).
However, the PEG precipitation test has limitations and the characteristics of the technique, particularly the specificity
and the capacity of PEG to interfere in some assays, must be appreciated if the test is to be applied appropriately and the
results interpreted correctly.

PEG precipitation is a crude and non-specific technique, which separates proteins by virtue of their solubility. PEG acts
as an inert solvent sponge, reducing solvent availability. With increasing concentration of PEG the effective protein concentration
is increased until solubility is exceeded and precipitation occurs.4 When applied to serum PEG precipitation is relatively specific for Igs and Ig complexes. However, it should be noted that
precipitation of IgA is only partial5 and a preliminary report indicates that the rare cases of macroprolactinaemia with an IgA macroprolactin may be missed.6 Macroamylase is predominantly an IgA complex but similar problems of incomplete precipitation by PEG have not been reported.

Unfortunately, PEG precipitation is not entirely specific for Igs and Ig complexes in serum. When used to detect macro-analytes
a proportion of the free analyte is invariably precipitated by PEG and this varies considerably between analytes and also
for any given analyte. For example, the upper limit of the reference range for PEG precipitated activity (PPA) of seven commonly
measured serum enzymes7 varies from 36% (ALP) to 76% (ALT) and the reference range for PPA for LDH is 12–70%. While the precipitation of free, monomeric
prolactin has been shown to be influenced by the concentration of serum gamma globulins8 this does not account for all of the variability for this analyte and other, as yet unknown factors must be present. Attempts
to optimize the concentration of PEG used to selectively precipitate macro-analytes are often thwarted by proportionate effects
on the solubility of the uncomplexed analyte as demonstrated for CK in serum.9 While PEG is relatively inert and the precipitated protein can be re-dissolved for further study, a further limitation of
the method is that PEG can interfere with some immunoassays.10 All of the effects discussed above necessitate the determination of appropriate analyte and method specific reference ranges
and the examination of a wide range of serum samples to exclude potential matrix effects. Furthermore, as the elevation in
activity or concentration caused by macro-analytes is most frequently modest9,11 the cut-offs used with PEG precipitation to detect such cases inevitably sacrifice specificity for sensitivity and confirmatory
tests are required. These are usually more complex and expensive and may not be readily available in the routine laboratory.

When PEG precipitation has been applied to the detection of macroenzymes, results have usually been reported as the percentage
of enzyme activity precipitated, which is directly related to the proportion of macroenzyme present. When applied to detection
of macro-hormones, particularly macroprolactin, the results have largely been reported as percentage recovery of hormone,
which is inversely related to the proportion of macro-hormone present. Recently, it has been argued that the priority for
the laboratory should be to determine whether the bioactive monomeric prolactin is elevated rather than simply detect the
presence of macroprolactin and it has been demonstrated that interpretation of results in terms of percentage recovery can
be misleading when macroprolactin is present and the monomeric prolactin concentration is elevated.11 It has been proposed that the prolactin concentration after PEG precipitation be taken as a measure of monomeric prolactin
and, because of the partial, non-specific precipitation of monomeric prolactin by PEG, this should be compared with reference
ranges determined by PEG precipitation in normal subjects.11 The same arguments may also be relevant to the interpretation of PEG precipitation applied to the investigation of other
macro-analytes.

Overall the non-specific nature of PEG precipitation is a major limitation of the technique but in some applications it is
an advantage. Most macro-analytes consist of Ig complexes. However PEG also precipitates the rare non-Ig containing macro
forms of prolactin, and macro forms of CK (macro-CK type II) and GGT that do not contain Ig. PEG precipitation has been recommended
as the best of a number of techniques for detecting macroprolactin12 and one reason for this is that, in addition to macroprolactin, PEG precipitates the big-prolactin (50–60 kDa) component
of total serum immunoreactive prolactin which also lacks biological activity in vivo.

Ig containing macro-analytes can be considered a subset of a more general class of assay interferences caused by antibodies
which include autoantibodies directed against analytes and antibodies directed against animal-derived assay reagents, which
can cause cross-linking or blocking. The antibodies concerned can be specific, as in macro-analytes and anti-animal antibodies
or polyspecific antibodies against either class of epitopes (heterophile antibodies). Interference from endogenous antibodies
in immunoassys continues to be a major concern for clinical chemists. Boscato and Stuart13 found that heterophile antibodies could be detected in the serum of 40% of normal samples and it has been suggested that
a radical solution to the problem of interference in immunoassay by endogenous antibodies would be to remove all Ig from the
sample before assay.5 PEG precipitation has been proposed for this purpose but is far from ideal given the limitations of the technique discussed
above. Nevertheless, PEG precipitation has the potential to identify all forms of assay interference due to antibodies. Together
with dilution tests, treatment with commercial heterophile antibody blocking tubes and confirmation by an alternative assay,
PEG precipitation has been recommended recently in the UK guidelines for the use of thyroid function tests14 for the investigation of assay interference in cases with unusual combinations of TSH and thyroid hormone results. As with
PEG precipitation, all the methods recommended require rigorous reference ranges and careful interpretation.10,15 At the moment, little information is available regarding reference ranges nor on the relative sensitivity and specificity
of these tests in the investigation of antibody interference in immunoassays. Because of the varied nature of antibody interference,
a battery of tests will be more sensitive than any one1618 but without further information it is not possible to determine the best strategy for using the tests. Furthermore, it should
be borne in mind that while these investigations may indicate antibody interference none will necessarily give a definitive
result for the analyte in question and a reference technique may be required.

While screening programmes using PEG precipitation to detect macroprolactin are well established in UK laboratories3 and screening for macro-CK has been advocated9 it is unlikely that any of the techniques mentioned above can be used to screen all samples arriving in the clinical biochemistry
laboratory in the near future. Detection of interference remains largely dependent on awareness of the problem and is usually
suspected when results from biochemical investigations and other clinical investigations are incongruous. Clinical validation
is considered to be a core function of clinical biochemists19 and this process should generate many more investigations for assay interference than is currently the case. A recent study
by Ismail et al.17 using rigorous clinical validation to detect potential assay interference in thyroid function tests and gonadotrophins showed
that as many as one in 200 results may warrant further investigation and that this approach was fruitful (28 of 59 suspect
results were found to be affected by interference). It is noteworthy that the clinical biochemist was instrumental in the
detection of macro-AST at an early stage in the cases reported in this volume of the Annals.1,2

PEG precipitation technique can be recommended as a simple method for the detection of assay interference provided carefully
defined protocols and reference ranges are used. However, we would emphasize the need for a variety of other screening methods,
with rigorously defined reference ranges, and confirmatory techniques, that may, unfortunately limit thorough, local investigation
of immunoassay interference. We reiterate the message from previous publications that the first step in identifying assay
interference is an awareness of its existence.

  • Accepted January 10, 2008.

References

  1. Abstract/FREE Full Text
  2. Abstract/FREE Full Text
  3. Abstract/FREE Full Text
  4. Abstract/FREE Full Text
  5. FREE Full Text
  6. Abstract/FREE Full Text
  7. Abstract/FREE Full Text
  8. Abstract/FREE Full Text
  9. Abstract/FREE Full Text
  10. FREE Full Text
  11. Abstract/FREE Full Text
  12. Abstract/FREE Full Text
  13. Abstract/FREE Full Text
  14. CrossRefMedline
  15. Abstract/FREE Full Text
  16. Abstract/FREE Full Text
  17. FREE Full Text
  18. FREE Full Text