It is well known that the polymerase chain reaction (PCR) is susceptible to inhibitors [1–4] and many publications describe methods for assessing inhibition using spiked alien molecules of various types [5–9]. Inhibition in real-time PCR can be measured as the increase in threshold cycle (Ct) or crossing point (Cp) relative to an uninhibited control [10]. The presence of inhibitors has the potential to increase error, reduce assay resolution, and produce false results in both quantitative and qualitative PCR assays. Direct assessment of inhibition is not usually performed [1], but as real-time PCR analyses frequently include additional reactions to control for sample variation (normalisation) by measuring reference 'housekeeping' transcripts [11] or genomic DNA [12], an assessment of, and compensation for, inhibition is often conducted indirectly.

Using a spiked alien molecule (as an internal positive control) or reference gene to assess inhibition relies on the fundamental assumption that any inhibitor present within the sample will have an equal effect on both PCR reactions. However, there appears to be no evidence in the literature to substantiate this assumption. Intuitively, there is no fundamental reason why this assumption should be valid, and yet it underpins a significant proportion of the PCR analyses performed daily in research and diagnostic laboratories throughout the world. In this study we examine, using a model system, how a range of different reactions may be differentially affected by PCR inhibitors and discuss the implications of the unexpected findings.

For more detailed methods please refer to the additional file.

Co-purification of inhibitors of PCR during nucleic acid extraction is a well recognised phenomenon [17, 2–4] that can be caused by numerous substances [18, 19]. The present study demonstrates that these inhibitors can have different effects on different PCR reactions, and that these differential effects can be concentration dependent. The latter point is particularly relevant as the actual concentration of a co-purified inhibitor is usually unknown.

The effect of reaction specific inhibition can be relatively subtle although statistically significant, as demonstrated here using urine extracts (Figure 1). In contrast, more dramatic effects may be seen at higher concentrations of inhibitor, as illustrated in Figures 2 &3. It is possible for one PCR reaction to be unaffected by a potential inhibitor whilst another is completely suppressed. If two different PCR reactions are to be compared, or one is to be used as a reference reaction for the other, as in the 'normalisation' procedure commonly used in quantitative gene expression studies [11], it is important that the two reactions are affected by potential inhibitors to the same extent; we describe this as inhibition compatibility. Recognition of the importance of assessing inhibition compatibility should contribute to reducing error and increasing accuracy in both gene expression studies and PCR-based molecular diagnostics. Inhibition incompatibility is likely to have a major effect on recent developments in the field, like multiplexed tandem PCR [20] and PCR arrays [21], that aim to allow many PCR reactions to be performed on a single sample.

Susceptibility of a PCR reaction to inhibitors is an important factor influencing the robustness of an assay that should be considered during experimental design. If different reactions are to be compared then they need to be of similar robustness, i.e. to be inhibition compatible. These observations lead to the question of how inhibition compatibility can be measured. An initial approach would be to perform inhibition assessment for the relevant PCR reactions in the presence of different concentrations of EDTA, as described in the present study. This is a simple strategy but it assumes that all potential inhibitors will have the same effect on the respective reactions as EDTA. The results obtained here using EDTA suggest that a PCR reaction that is more susceptible to inhibition by depletion of free Mg²⁺ may also be more susceptible to inhibition by urine or urine extracts. This implies that the inhibitors present in urine might also be acting by depletion of free Mg²⁺, although this remains uncertain. Perhaps a more thorough approach would be to assess the relevant PCR reactions for inhibition using a range of concentrations of several well known inhibitors such as heparin and ethanol in addition to EDTA. Ideally, nucleic acid extracts from the sample types of interest (assuming they are known not to contain the PCR target) should also be tested for their inhibitory potential with the relevant PCR reactions but this may not always be feasible.

Ideally, it would be possible to design PCR reactions to be inhibition compatible and to minimise inhibition susceptibility. Unfortunately, the findings of this study have not revealed any primer, or amplicon parameter that is reliably and consistently associated with susceptibility to inhibition, although we can tentatively suggest that amplicon GC content and primer Tm may be significant factors. While it is unlikely that there is a single simple factor that can be manipulated in experimental design to ensure inhibitor compatibility there are a number of general measures that can be taken to minimise the problem. These measures include the careful selection of type of thermostable DNA polymerase [22], reduction in the amount of DNA template added to the reaction and the use of certain additives such as bovine serum albumin, which provide some resistance to inhibitors that may be present in blood [23]. Increasing the denaturation time can also reduce the susceptibility to inhibition in certain cases (data not shown).

The findings of this study tackle an area that is frequently overlooked when conducting PCR. Whilst we acknowledge that our experiments use an unusual approach, as unextracted urine is not routinely used in PCR reactions (although unextracted urine has recently been used for viral genome detection [24]), the results illustrate an important principle. Users should be aware that co-purified inhibitors in nucleic acid extracts may not affect all PCR reactions equally and this fact must be taken into account when considering sample choice, experimental design and data interpretation.