The dangers of extended PCR in the clinic

To the editor—Rearranged mitochondrial (mt) DNA molecules are associated with an increasing spectrum of human diseases1. To be pathologically important, such molecules must constitute a major fraction of total mtDNA, at least within affected tissues. Although this is easily demonstrated by Southern blot analysis, it is a relatively slow and cumbersome technique, requiring substantial amounts of intact, highly purified and high-molecular-weight DNA. PCR-based methods to detect mtDNA deletions in the clinical context seem preferable, as they are quick, inexpensive, require only very small amounts of biopsy material, are tolerant of poor sample quality and can be automated. However, PCR-based methods are notoriously prone to artefacts, and are at best semi-quantitative.

One particular technique, long-extension PCR (LX-PCR), is commonly used to detect the full range of deleted mitochondrial DNA species associated with muscle disease, such as dilated cardiomyopathy2, aging or other conditions. In contrast to conventional PCR, in which most deleted mtDNA molecules of clinical relevance cannot be detected by any given primer pair, LX-PCR uses two outwardly oriented, adjacent primers whose sequences are located within a region that is rarely, if ever, deleted. This approach leads to the amplification of full-length mtDNA (16.6 kb) plus any subgenomic, deleted molecules that may be present.

In a recent report3 we showed that LX-PCR products with characteristics of typical deleted mtDNA molecules can be synthesized even from DNA templates extracted from control tissues. We showed that the appearance of such products is dependent on the overall template DNA concentration. After serial dilution of a concentrated DNA sample in which rearranged molecules are not detectable by Southern blot analysis, these 'deletion' products decrease from being the only products detectable to being almost undetectable, with the 16.6-kb band of undeleted mtDNA becoming increasingly apparent. This is consistent with the idea that the relevant 'deleted' templates are aberrant molecules present only in very low amounts, akin to the sublimons found in plant mitochondria4. In contrast, a PCR product corresponding to a 'real' deletion—one present in pathologically meaningful amounts—does not disappear after dilution of the template, remaining the sole or main product even at the most extreme dilutions (Fig. 1a). Sublimon-type products are undetectable using rho-zero cell templates, hence are not derived from nuclear pseudogenes. However, template DNAs from different tissues (ref. 4 and unpublished data) give different sublimon-type products, indicating that they are also not the result of simple mis-priming artefacts.

Figure 1: Detection limit for deleted mtDNAs by LX-PCR.

a, Reactions using increasing amounts of template DNA from two patients with mitochondrial disorders and previously characterized single deletions representing a large fraction of the mtDNA molecules. Patient 2 DNA (from a 143B osteosarcoma cybrid cell-line containing mtDNA from a Pearson syndrome patient - 0.12, 0.6, 3, 15, 75 and 375 ng, lanes 2-7) carries an 8926 bp deletion from np 6130 to 15056, representing 50% of all mtDNA molecules in the sample, based on Southern blotting (J. Poulton, unpublished data). Patient 1 DNA (from liver of a Kearns-Sayre syndrome (KSS) patient - 0.022, 0.11, 0.56, 2.8, 14 and 70 ng, lanes 8-13) carries a 10 kb deletion in 62% of all mtDNA molecules, plus a low percentage (3%) of deletion dimers and partially duplicated molecules with the same primary re-arrangement. Lanes 1 and 14 - marker (M). b, A fixed amount (30 ng) of control DNA, from a 143B osteosarcoma cybrid cell-line containing only undeleted mtDNA, was spiked with decreasing amounts of the deletion-containing DNAs from Patient 1 (70, 14, 2.8, 0.56, 0.11, 0.022 and 0.0045 ng, lanes 2-8, left-hand panel) and Patient 2 (375, 75, 15, 3, 0.6, 0.12, 0.024 and 0.0048 ng, lanes 2-9, right-hand panel). The % deleted mtDNA in these series is thus estimated as follows: Patient 1: 46%, 21%, 5.5%, 1.2%, 0.24%, 0.048% and 0.0097%, Patient 2: 46%, 36%, 17%, 4.5%, 0.98%, 0.2%, 0.040% and 0.0080%. The use of Southern blotting on these gels (not shown) increased the detection limit for deleted molecules a further five-fold. Lanes 1 (M, both panels) - marker.

In a reconstruction experiment (Fig. 1b), we 'spiked' a constant amount of control template DNA with decreasing amounts of DNA templates from patients with mitochondrial myopathy, each carrying a single, pathological deletion in mtDNA, previously mapped and quantitated by Southern blot and sequence analysis. This demonstrated that bona fide deleted mtDNA molecules present at a concentration of only one or a few copies per cell can be detected by this method. However, sublimon-type templates give equally prominent products. Our findings imply that the background of sublimon-derived products generated from control templates makes LX-PCR unreliable as a sole diagnostic method for detecting deleted mtDNAs, except in the case of deletions representing a substantial fraction of mtDNA molecules in a given DNA preparation.

We would thus recommend routine serial dilution of all DNA samples to test for the meaningful presence of deleted mtDNA molecules when using LX-PCR, and ideally the verification of all positive findings by Southern blot analysis, before a diagnostic conclusion is reached. Published claims, based exclusively on LX-PCR analysis, that deleted mtDNAs accumulate to high levels in aging and in many disease states , need to be critically re-evaluated in the light of our findings.


  1. 1

    Fadic, R. & Johns, D.R. Clinical spectrum of mitochondrial diseases. Semin. Neurol. 16, 11–20 (1996).

    CAS  Article  Google Scholar 

  2. 2

    Melov, S., Shoffner, J.M., Kaufman, A. & Wallace, D.C. Marked increase in the number and variety of mitochondrial DNA rearrangements in aging human skeletal-muscle. Nucl. Acids Res. 23, 4122–4126 (1995).

    CAS  Article  Google Scholar 

  3. 3

    Kajander, O.A. et al. Long-extension PCR to detect deleted mitochondrial DNA molecules is compromized by technical artefacts. Biochem. Biophys. Res. Commun. 254, 507–514 (1999).

    CAS  Article  Google Scholar 

  4. 4

    Small, I.D., Isaac, P.G. &, Leaver, C.J. Stoichiometric differences in DNA molecules containing the AtpA gene suggest mechanisms for the generation of mitochondrial genome diversity in maize. EMBO J. 6, 865–869 (1987).

    CAS  Article  Google Scholar 

Download references


We thank M. Niittylahti and O.Lumme for technical assistance, and P. Rustin, I. Holt, S. Khogali and N.-G. Larsson for discussions. This work was supported by grants from the Finnish Academy, Muscular Dystrophy Group, Royal Society, Tampere University Hospital Medical Research Fund, Yrjö Jahnsson Foundation, Finnish Foundation of Alcohol Research and the Pirkanmaa Region Fund of the Finnish Cultural Foundation.

Author information



Corresponding author

Correspondence to Howard T. Jacobs.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Kajander, O., Poulton, J., Spelbrink, J. et al. The dangers of extended PCR in the clinic. Nat Med 5, 965–966 (1999).

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing