MOLECULAR DIAGNOSTICS

Theoretically, polymerase chain reaction (PCR) can amplify a single strand of target DNA a trillion times. Thus, its sensitivity as a diagnostic tool for ocular infections is of a similar magnitude – theoretically. But in the real world, PCR can be far too sensitive, detecting random, or even dead, microbes that are not responsible for observed disease, Todd P Margolis MD, PhD, of the University of California – San Francisco, US, told the American Academy of Ophthalmology annual meeting. “There’s a big disconnect between sensitivity and clinical relevance.â€

Indeed, one PCR study detected bacteria or fungi in 52 per cent of swabs exposed only to air (Kim et al. AJO 146: 714-723, 2008). Another generated 42 per cent false positives when the assay was set to detect one copy of varicella-zoster virus (VZV) DNA (Short et al AJO 123: 157-164, 1997). “That’s why you don’t set a PCR-based assay to infinite sensitivity,†said Dr Margolis, who is also medical director of a clinical ocular microbiology lab. Conversely, many factors can sharply reduce actual assay sensitivity, resulting in false negatives, Dr Margolis said. Among them are specimens containing substances that inhibit the PCR reaction, samples from partially treated eyes, diluted samples, inadequate samples and samples that are not adequately frozen or not kept cold enough in shipment.

Also, clinically relevant assay parameters differ for aqueous, vitreous and corneal samples. Worse, there are no standards for sampling corneal tissue, and the amount of host DNA differs enough in epithelial and stromal tissue that they require different assay approaches to yield clinically meaningful results, Dr Margolis said. “If you are going after stromal lesions you are probably not going to pick it up.†Generally, vitreous samples are more reliable than aqueous for most infectious retinal disease. PCR assays for cytomegalovirus (CMV) iritis/corneal endotheliitis are notoriously inaccurate, yielding false negatives for about 75 per cent of clinically evident cases in his lab, Dr Margolis noted. PCR assays are also extremely sensitive to small process variations from lab to lab, he added.

Because of these variables, PCR assay parameters must be developed specifically by pathogen and sample type at each lab. The repeatability and clinical predictive power of each parameter set must be validated for it to be useful for diagnostic purposes, Dr Margolis said. “Molecular diagnostics are not magical assays. Results should be interpreted with caution,†he advised.

PCR in practice

Nonetheless, PCR represents a new paradigm in ocular infection diagnostics, said Regis P Kowalski MS [M]ASCP, associate professor of ophthalmology at the University of Pittsburgh medical school and executive director of the ophthalmic microbiology lab at the University of Pittsburgh Medical Center (UPMC), Pennsylvania, US.

“Not everything can be cultured, especially for intraocular areas,†Prof Kowalski said. Clinically validated PCR can be highly useful and accurate in these cases. For example, in his lab, PCR detects 100 per cent of VZV vs about nine per cent for cultures, and 100 per cent of Chlamydia vs about 38 per cent for culture. However, culturing is more accurate for some pathogens, such as adenoviruses (ADV). Like other lab tests, PCR should be used to supplement other diagnostic methods, including clinical suspicion, culture isolation, microscopy and confocal microscopy.

Prof Kowalski also emphasised the importance of sending clinical samples only to clinical labs that have validated assays for specific infectious suspects, and to avoid using research labs. Clinical labs use both true positive and true negative test samples to validate assays, whereas research labs do not, he explained. Rigorous clinical validation requirements result in high reliability, Prof Kowalski said. “In more than 10 years testing, no known false positives have been reported with patient samples in our lab.â€

UPMC offers PCR in-house for herpes simplex virus (HSV) 1 & 2, VZV, ADV, CMV, Epstein-Barr virus (EBV), Chlamydia, Neisseria gonorrhoea and Acanthamoeba, Prof Kowalski said. The lab is also part of an international network that provides validated tests for toxoplasma, tuberculosis, mycobacterium leprae and rubella, with turnaround times of 24 to 48 hours. “It’s important to have a network for less common pathogens. No lab tests for everything, not even the CDC.â€

Prof Kowalski also stressed the importance of proper sampling, preservation and storage practices. He recommends several steps for surgeons interested in adding molecular diagnostics to their practices. First, find a molecular test facility and educate yourself on its capabilities and protocols, and how to take and process samples. Second, keep simple supplies on hand, including saline, viral transport, swabs and cold packs.

Third, process samples correctly and arrange appropriate transport by courier. Fourth, be reasonable in your expectations. “Don’t pan sample and expect PCR for everything. It doesn’t work, it’s quite expensive and it’s not going to happen,†Prof Kowalski said. Send undiluted samples, untreated samples, freeze samples immediately at -80 C if you can, and send samples frozen, Dr Margolis added. He cautioned that PCR results always must be put in context. “The problem is usually not the test, it is the interpretation by the individual receiving the test results.†Clinical correlation is always advised.