Prevalence and Management of Invalid GeneXpert Enterovirus Results Obtained with Cerebrospinal Fluid Samples: a 2-Year Study

Clinical samples.

CSF specimens submitted to Vanderbilt University Medical Center between 1 April 2007 and 20 September 2008 for detection of EnV by PCR were collected consecutively. Specimens with enough leftover volume and still unfrozen after the completion of diagnostic testing were included in the study.

Determination of CSF characteristics.

Each CSF sample was visually examined for visible red blood cells (RBCs), clotting, and the presence of xanthochromia. RBC visibility was graded semiquantatively on a scale using rankings of clear, 1+, 2+, 3+, and 4+. These grades were based on specimen colors and turbidities which roughly correspond to CSF RBC counts of <200/μl (clear), 200 to 5,000/μl (1+), 5,000 to 75,000/μl (2+), 75,000 to 150,000/μl (3+), and >150,000/μl (4+).

GXEA.

The GXEA kit was purchased from Cepheid (Sunnyvale, CA). The information from each kit lot was recorded. CSF samples were tested using the GXEA as previously described (6, 7). In brief, 140 μl of CSF was added to the GeneXpert cartridge and then processed automatically for the different steps of sample preparation and amplification. After the diagnostic procedure was completed, the following two additional procedures were performed within 48 h on unfrozen CSF specimens with enough leftover volume prior to the routine GXEA procedure: (i) a 1:5 dilution, in which a CSF specimen was diluted once with saline in a 1:5 ratio; and/or (ii) a freeze-thaw cycle, in which a CSF specimen was quickly frozen in dry ice and thawed in a 37°C water bath one time.

During the study period, a total of 525 CSF specimens were submitted to the diagnostic laboratories. The samples spanned two EnV seasons, with 301 collected in 2007 and 224 collected in 2008. The patients' ages ranged from less than 1 day old to 74 years old, with an average age of 8 years. The ratio of males to females was 0.56:0.44. Among the 525 CSF samples enrolled in this study, 95 were positive for EnV by the GXEA, giving a positive rate of 18.1%. Invalid GXEA results were reported for 43 (8.2%) specimens during the 2-year study period. The invalid-result rate was 9.6% in 2007 and 6.3% in 2008, with no significant change in the invalid-result rate from 2007 to 2008 (χ² = 1.96, P > 0.05). Included in the evaluation were GXEA kits with seven different lot numbers purchased during the 2-year study period, and invalid results were evenly distributed among the seven lots.

Assuming that the built-in internal control was functioning correctly to detect inhibitory or interfering substances, we next assessed potential factors associated with the occurrence of invalid results. Invalid results correlated with visible RBCs in the CSF specimens tested (Table 1). Of the 525 CSF samples tested, 388 samples were graded as clear, of which only 19 (4.9%) samples gave invalid results. The invalid-result rate but not the EnV-positive rate correlated with RBC cellularity (χ² trend = 38.01, P < 0.0001). In comparison to the results for clear specimens, CSF specimens with 1+, 2+, 3+, and 4+ RBC cellularity gave odds ratios at a 1.07-, 2.92-, 6.28-, and 8.60-fold likelihood, respectively, of encountering invalid results (Table 1). Visible RBC clots were only seen in six samples and were not related to the occurrence of invalid results (data not shown). Xanthochromia was noted in 44 CSF specimens of which 2 (4.5%) tested invalid; however, no statistical significance was discovered (χ² = 0.85, P > 0.05) in comparison to the 41 (8.5%) invalid findings in the 481 nonxanthochromic samples. These data indicated that invalid GXEA results were correlated with increased RBC cellularities of the CSF specimens tested.

CSF specimens with sufficient volume that were still unfrozen were retested to pursue potential solutions for minimizing the occurrence of invalid results. Twenty-eight CSF specimens were tested again after a 1:5 dilution. All but five diluted samples gave valid results, thus diminishing the invalid-result rate by 82.1% (Table 2). Twenty-five samples were subjected to a one-time freeze-thaw cycle before retesting, as reported recently (12), and all invalid results were resolved except for one sample, reducing the invalid-result rate by 96.0%. Taken together, we found that a one-time freeze-thaw cycle or a 1:5 dilution of the tested specimen markedly diminished the invalid-result rate, with no statistical difference indicated between the results for the two procedures (Fisher's exact test, P > 0.05).

We next studied whether the two procedures influenced diagnostic sensitivities. Among all positive specimens with enough leftover volume, retesting after a 1:5 dilution and a one-time freeze-thaw cycle achieved sensitivities of 98.3% and 96.4%, respectively (Table 2). There was no statistical difference in sensitivity between the two procedures (Fisher's exact test, P > 0.05). In addition, there were three and four specimens with originally invalid results that became positive after the dilution and the freeze-thaw cycle, respectively. Together, these data indicated that using a 1:5 dilution or one-time freeze-thaw cycle on CSF specimens with RBC cellularity remarkably diminished invalid test results without significantly decreasing the test's sensitivity.

This study presents our 2-year data on EnV detection by the Cepheid GXEA. Invalid results occurred at a rate of 8.2% when original, untreated CSF specimens were used. These results were in the range of the results of other published studies, which have ranged from 6.7% to 16% (7, 12), and show that the internal control is functioning correctly to detect inhibition. We found that invalid results were highly correlated to the level of RBC cellularity, i.e., “bloody tap.” Importantly, we also found that one of two simple procedures carried out prior to testing, a 1:5 dilution or a one-time freeze-thaw cycle, remarkably diminished the frequency of invalid results without significantly decreasing the test's diagnostic sensitivity. This information may be useful for laboratories performing on-demand testing for EnVs in CSF, especially for those supporting emergency room management of meningitis cases. One limitation of this study is that the results from the GXEA were not compared with a reference method; therefore, the possibility of false-negative results, as previously indicated (12), was not addressed. However, a review of our data on negative specimens failed to show the “low positive but below cutoff” results reported by Seme et al. in any of our reported negative specimens with valid internal control results. Thus, in our experience, a false-negative result due to partial inhibition that is undetected by the internal control seems unlikely, but it is intriguing that Seme et al. showed a potential benefit in incorporating a freeze-thaw step to address this issue in their laboratory (12).

Bloody taps are not uncommon in the evaluation of pediatric or adult meningitis, so rendering these specimens evaluable would potentially be useful, especially if results need to be returned quickly. The Cepheid product insert claims that there is no inhibition in CSF samples with RBCs up to 2.5% blood (vol/vol). In our study, when 3+ or higher (>75,000/μl) RBC cellularities in CSF specimens were visualized, the invalid GXEA results increased significantly, with an odds ratio over 3. There are several potential explanations for the RBC cellularity-related invalid results. First, components of RBCs are known to inhibit in vitro nucleic acid amplification. Heme is a well-known amplification inhibitor (1, 2, 15). In our study, however, a freeze-thaw cycle prior to testing significantly diminished invalid-result frequencies. Since heme is present in these samples, it is difficult to determine whether heme played a significant role as an inhibitor in the GXEA test system. It has been reported that several commonly used one-step RNA-targeted enzymes, such as rTth, are usually not affected by heme (3).

A second explanation could be that the high level of RBCs in CSF affects the lysis of cells and virions, resulting in incomplete release of viral nucleic acids. This could be manipulated by adjustment of the volume of the CSF specimen. The additional dilution or freeze-thaw cycle can improve cell and virion lysis and viral nucleic acid release, thereby diminishing invalid-result frequencies (12). The last explanation is that, in the GXEA test cycle, the sample is pumped from one chamber to another within the microfluidic device. High levels of RBCs, with or without clot formation, could block the movement of liquids between test chambers. Both of the two explanations are supported by the fact that either a 1:5 dilution or a one-time freeze-thaw cycle remarkably diminishes the frequency of invalid results.

There have been numerous studies of specimen treatments or added reagents that act as amplification facilitators and also of possible solutions for amplification inhibitors (1, 3, 5, 13). Adding additional chemicals or steps to the lysis procedure of the GXEA could potentially enhance cell and virus lysis, thereby diminishing invalid results, but would constitute “off-label” use and require validation as performed here. In our routine diagnostic services, any CSF specimens with visible RBCs are automatically diluted to 1:5 prior to testing to keep the test turnaround time rapid without significantly decreasing test sensitivity.