Polymerase Chain Reaction in the Diagnosis and Management of Central Nervous System Infections

Polymerase chain reaction (PCR) is a broadly applied laboratory test for the diagnosis of a wide variety of central nervous system (CNS) diseases, including genetic and autoimmune diseases, malignant neoplasms, and infections.^1,2With its ability to detect minute amounts of DNA or RNA contained in tissues or fluids, PCR has improved the rapidity and accuracy of diagnosis, enhanced understanding of pathogenesis, and helped identify infectious causes for diseases previously considered idiopathic. In addition, PCR can be performed on a variety of tissues preserved in different ways—even archival specimens can be used to provide important epidemiological information. By making quick and precise diagnoses, appropriate treatments can be instituted, and unnecessary or invasive investigations can be avoided.

The power of PCR results from its ability to synthesize millions of copies of a specific gene segment in vitro, starting with one or only a few template copies, an amount undetectable by other methods. The PCR technique is illustrated in Figure 1. Once the sequence of the DNA segment of interest is identified (template), 2 synthetic oligonucleotides (primers) are chosen that have a sequence complementary to each strand of the DNA to be amplified. These primers are usually between 20 and 40 bases long. The first step requires denaturing the double-stranded DNA template by heating it to 92°C to 95°C. Once this occurs, the primers will anneal to their respective complementary template strands after cooling to 55°C to 70°C. To maximize primer-template annealing vs template-template reannealing, primers are supplied in great excess; thus, there are more primers available for binding than de novo DNA strands. Binding of primers to template is only the first step. In order to build a new DNA strand complementary to the template, 2 main ingredients are necessary: nucleotides (adenine, guanine, thymidine, cytosine) and the enzyme to catalyze their linkage (DNA polymerase). By supplying nucleotides in vast molar excess and by making use of a thermostable DNA polymerase (usually Taq polymerase), a new complementary DNA sequence downstream of the annealed primer is assembled, resulting in 2 complete double-stranded DNA pairs. This completes the first cycle. Multiple cycles (usually 30-40) of denaturing, the annealing of new primers, and strand elongation ultimately yield more than 1 billion copies of the original DNA template in 2 to 3 hours. The entire reaction is completed within a self-contained thermal cycler using very small volumes (microliters) of reagents and sample.

When it is important to know not only that a gene (DNA) is present, but also that it is being expressed and transcribed into messenger RNA (mRNA), or when RNA viruses are the target of detection, a variant of PCR called reverse transcriptase PCR (RT-PCR) can be used. The general principle is identical to that outlined above, but instead of Taqpolymerase, an enzyme with reverse transcriptase activity is used that first reverse-transcribes the mRNA into complementary DNA. The complementary DNA then serves as the template from which the segment of interest is amplified. Situations in which this might be important include diseases caused by genetic defects with variable penetrance or viral infections in which both latent and lytic infection play a role in pathogenesis (eg, herpesviruses).

The amplification products of PCR can be detected using a variety of methods. The products can be loaded onto an ethidium bromide–stained agarose gel and then electrophoresed, allowing migration of the DNA product to its predicted location in the gel based on the size of the fragment generated (determined by the spacing of the 2 primers chosen). To ensure that these products are specific, however, it is necessary to perform Southern blot analysis. In this process, the target DNA is transferred and immobilized on a membrane, then tagged with a complementary probe to which a radioactive, fluorescent, or colorimetric marker is attached. An additional technique that can be used to increase specificity of products is termed nested PCR. In this technique, a second round of PCR is performed using primers internal to the original primers, thus identifying only the subset of amplification products that correspond to the target fragment. Nested PCR can also be used to increase the sensitivity of PCR. Perhaps the best standard for ensuring specificity is to sequence PCR products and compare the sequence with the known target gene sequence. However, PCR product sequencing is not typically performed as a routine procedure, and its use is largely limited to initial validation of PCR assays and as a research tool.

Synthesis of nonspecific products can result in false-positive PCR results if appropriate steps to confirm specificity are not taken. Mispriming may occur (primers binding to the wrong region of DNA), but this can be minimized by increasing the "stringency" of the reaction conditions and can be detected by failure to find amplified material using Southern blot analysis. False-positive PCR results can also occur because of contamination of reactions with nucleic acid either from products of prior reactions (carryover) or that is present in the laboratory environment. Experienced laboratories take extensive precautions to avoid sources of inadvertent contamination, including proper design of laboratory facilities and protocols for handling specimens. Negative controls are used in all reaction runs to detect inadvertent contamination.

There are numerous areas in which PCR can be directly applied in clinical practice.²In this review, we will focus on the application of PCR to the diagnosis and management of infectious neurologic diseases.^2,3

Viruses are common causative agents in certain neurologic diseases (eg, meningitis, encephalitis, and myelitis). It is difficult, often impossible, to isolate many viruses by standard cell culture techniques. Even for viruses such as enteroviruses, which are generally more amenable to culture, a prolonged period may be required for accurate identification (often up to 8 days). False-negative cultures are common and may occur in 25% to 35% of specimens, and some enterovirus serotypes do not grow at all in cell culture. In the case of herpes simplex virus (HSV) encephalitis in adults, the virus is almost never detectable in cerebrospinal fluid (CSF) by culture. As a result of these problems, prior to the widespread use of CSF PCR testing in viral diagnosis, patients were often treated with unnecessary empiric antimicrobial therapy for prolonged periods and underwent invasive diagnostic procedures, such as brain biopsy. Serologic diagnosis of CNS viral infections is limited primarily by the time required to detect high levels of specific antibody production (often 2-3 weeks from the beginning of the disease) or by the ubiquitous nature of seropositivity to certain viruses (eg, HSV). Circumstantial evidence of viral infection, such as stool or respiratory excretion, may be helpful, but it does not necessarily correlate with active CNS disease (eg, the intermittent shedding of cytomegalovirus in urine or respiratory secretions and the prolonged shedding of enteroviruses both occur in asymptomatic individuals). Polymerase chain reaction is attractive because it circumvents most of these limitations. Table 1summarizes the broad range of CNS viral infections to which PCR is clinically applicable. As more specific antiviral compounds are identified that can be used therapeutically, accurate diagnosis will become increasingly important.

In addition to viral infections, PCR has also been used to diagnose bacterial, mycobacterial, rickettsial, and protozoal infections of the CNS (Table 2).

It is often difficult to accurately determine the sensitivity and specificity of PCR in CNS infections because in some situations the previously used standards for diagnoses appear to be less sensitive than PCR. Existing standards, such as viral CSF culture, are often both insensitive and slow. In some cases, definitive diagnosis was based on invasive procedures, such as brain biopsy, which may have been performed only in severely ill patients, potentially producing a skewed picture of the severity or nature of disease (eg, HSV encephalitis). Despite the extreme sensitivity of PCR, false-negative results do occur. In the case of most viral infections, the highest levels of nucleic acid are present acutely and typically decrease with treatment or over time. A delay in performing PCR may result in false negatives. Inhibitors of PCR may be present in body fluids or tissues and may lead to false-negative results. In CSF, the most commonly encountered inhibitors are heme products resulting from the breakdown of erythrocytes. Small numbers of red blood cells in CSF samples do not inhibit PCR, but this can be a problem with CSF samples that are grossly contaminated with blood. No inhibition has been noted in specimens with high levels of protein or high leukocyte counts. As opposed to other specimens, CSF tends to have low concentrations of endonucleases/exonucleases and proteins that could inhibit the action of the polymerase.

Polymerase chain reaction can be performed with as little as 30 µL of CSF. Polymerase chain reactions are best performed on fresh CSF specimens, although brief (days) storage of CSF at refrigerator temperature does not appear to significantly reduce diagnostic yield. Positive PCR results have been obtained from CSF specimens that have been frozen for years, but the decline in sensitivity with time has not been rigorously examined. In general, DNA appears to be more stable during storage than RNA, which is more susceptible to degradation. Finally, short courses (eg, a few days) of antimicrobial therapy do not appear to have a significant effect on PCR yields, which is a great advantage over culture techniques, in which diagnostic yields often drop precipitously following even brief periods of antimicrobial therapy.

In addition to aiding in the diagnosis of infectious diseases, PCR can be used for a variety of other purposes. Polymerase chain reaction may facilitate the differential diagnosis between recurrent viral infection, which would be expected to be PCR-positive, and postinfectious immune-mediated disease, which is PCR-negative. For example, following antiviral therapy for HSV encephalitis, some patients show clinical relapse. A positive CSF HSV PCR result suggests that this is caused by recurrent viral infection and supports the reinstitution of antiviral therapy. A negative PCR result is consistent with a postinfectious immune-mediated process, and additional antiviral therapy is rarely of value. Polymerase chain reaction can be used to identify determinants of drug resistance by the sequencing of PCR-amplified genes encoding targets for antiviral therapy. For example, PCR amplification and sequencing of selected viral genes in patients with human immunodeficiency virus (HIV) infection may help identify mutations that correlate with resistance to specific antiretroviral therapies. As PCR techniques have become more sophisticated, it is now possible in many cases to accurately quantify the amount of nucleic acid that is present in the sample, rather than merely provide qualitative detection. Quantitative PCR may prove useful in monitoring the duration and adequacy of therapy or in providing prognostic information about illness. Qualitative PCR is being used with increasing frequency to study CNS infection caused by HIV and HSV. In the case of neonatal HSV encephalitis, qualitative PCR has led to changes in the recommended dosage and duration of acyclovir therapy and has resulted in more widespread use of adjunctive oral therapy following primary intravenous antiviral therapy. In the case of HIV infection, quantitation of viral load in CSF is being actively investigated as a potential marker of the risk for and severity of HIV-associated encephalopathy. Polymerase chain reaction may also help to identify a potential role for viruses or other microbial pathogens in neurologic and psychiatric diseases of uncertain causes, including schizophrenia, multiple sclerosis, Alzheimer disease, and other neurodegenerative diseases. One of the interesting results from initial studies in this area has been the frequency with which the nucleic acid corresponding to a variety of viral pathogens can be identified from brain tissue, even in patients without known neurological diseases. By contrast, CSF PCR results are almost invariably negative in asymptomatic individuals without known neurological diseases. In the case of CSF, positive PCR results almost invariably correlate with the presence of a clinically significant disease. The significance of positive PCR results in brain tissue needs to be interpreted with great caution until the nature, significance, and frequency of "normal brain genomic flora" identified by PCR are clarified.

Accepted for publication January 8, 1999.

This study was supported by Merit Review and Research Enhancement Award Program grants from the Department of Veterans Affairs, Washington, DC; by grant DAMD17-98-8614 from the US Army, Washington, DC; and by grant 5R01AG14071 from the National Institute on Aging, Bethesda, Md (Dr Tyler).

We thank Adriana Weinberg, MD, for her thoughtful comments.

Corresponding author: Kenneth L. Tyler, MD, Department of Neurology, B-182, University of Colorado Health Sciences Center, 4200 E Ninth Ave, Denver, CO 80262 (e-mail: ken.tyler@uchsc.edu).