| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Molecular Diagnostics and Genetics |
1 Beiersdorf AG, Hamburg, Germany.
2 Med. Klinik II, Charite-Campus Charite Mitte, Berlin, Germany.
3 Robert Koch Institute, Berlin, Germany.
aAddress correspondence to this author at: Analytical Microscopy Department, Beiersdorf AG, Unnastrasse 48, D-20245 Hamburg, Germany. Fax 49-40-4909-3855; e-mail stefan.biel{at}beiersdorf.com.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We evaluated direct EM, ultracentrifugation (UC) before EM, and solid-phase immuno-EM (SPIEM). The diagnostic accuracy of EM was studied by comparison with a real-time PCR assay on 531 clinical samples.
Results: The detection rate of EM was increased by UC and SPIEM. On 531 clinical urine samples, the diagnostic sensitivity of EM was 47% (70 of 149) with a specificity of 100%. We observed a linear relationship between viral genome concentration and the proportion of urine samples positive by EM, with a 50% probability for a positive EM result for urine samples with a polyomavirus concentration of 106 genome-equivalents (GE)/mL; the probability of a positive EM result was 0% for urine samples with <103 GE/mL and 100% for urine samples containing 109 GE/mL.
Conclusions: UC/EM is rapid and highly specific for polyomavirus in urine. Unlike real-time PCR, EM has low sensitivity and cannot quantify the viral load.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
An "ideal" diagnostic facility offers all of the diagnostic techniques available to provide the best possible service, but this is very expensive and unlikely to be cost-effective. Therefore, a decision on which technique to use for a particular virus must be made. Today, some laboratories are discarding EM as a diagnostic tool and replacing it with PCR assays because the latter can generally be performed according to routine laboratory protocols, whereas EM still depends greatly on the skill and experience of the personnel using the microscope. However, there is continuing discussion concerning whether PCR is a satisfactory alternative to EM (7)(8)(9)(10)(11), and few direct comparisons between PCR and EM as diagnostic assays for one specific agent have been made (12)(13)(14). Although technical specifications, e.g., sensitivity, specificity, and detection limits, have been published for many PCR-based assays [see Ref. (15) for an example], for EM these data are difficult to codify.
We therefore evaluated the diagnostic performance of EM to provide some reliable data to compare with PCR. We did this by evaluating different EM preparation techniques [standard negative staining, ultracentrifugation (UC) before EM, and immune-specific particle enrichment] and then comparing the best of these EM assays with an up-to-date real-time PCR assay on 531 clinical samples.
For this study, we chose the detection of polyomavirus, which is considered a possible causative agent of hemorrhagic cystitis, in urine samples from bone marrow transplant patients. Different laboratories use either EM (2)(3)(9) or PCR (4)(5)(6) for this purpose, and no one technique has been recognized as the best. Moreover, the often high viral load in urine samples from transplant patients and the need for fast diagnosis (6) makes EM competitive with PCR.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
purification of polyomavirus particles from urine samples
For purification of polyomaviruses, selected urine samples were initially centrifuged for 15 min at 3500g to remove large cellular debris. The supernatant was then layered on two sucrose cushions (300 and 700 g/L) and centrifuged for 2 h at 100 000g. For further purification, the virus band between the sucrose cushions was pipetted on a CsCl solution (1.39 kg/L) and centrifuged for 45 h at 100 000g. The CsCl gradient was unloaded from the bottom and fractionated into 0.5-mL portions (16). The CsCl density of the fractions was determined by refractometry. As assessed by negative-staining EM, the fraction with a buoyant density of 1.41 kg/L contained concentrated, highly purified, intact polyomaviruses. To remove the CsCl, we dialyzed the virus-containing fraction overnight at 4 °C against 0.05 mol/L HEPES, pH 7.
negative staining with different heavy metal salt solutions
Negative-staining EM of purified polyomavirus particles was performed as described elsewhere (17). Briefly, 400-mesh copper grids covered with Pioloform F and carbon were floated on urine drops, washed twice on drops of doubly distilled water, and then contrasted on a drop of staining solution. Excess stain was drawn off with torn filter paper, and the grids were allowed to dry. The specimens were examined in a transmission electron microscope (Zeiss EM 10A) at a magnification of x40 000 for at least 15 min. Aqueous solutions of the following heavy metal salts were used for negative staining: uranyl acetate (20 g/L, pH 4), phosphotungstic acid (20 g/L, pH 7), uranyl acetate plus phosphotungstic acid (each at 5 g/L, pH 5), sodium silicotungstate (20 g/L, pH 6), ammonium molybdate (10 g/L, pH 6), and methylene aminotungstate (20 g/L, pH 7). The pH for each stain was adjusted with 1 mol/L NaOH.
heavy metal shadowing
To assess viral morphology without the influence of interacting heavy metal salts, purified virus particles were shadowed in a BAF 060 coating unit (BAL-TEC AG) with a platinum/carbon layer of 
2 nm thickness at an angle of 45 degrees. The samples were then examined by transmission EM at a magnification of x40 000.
solid-phase immuno-em
For virus enrichment on the grid, we performed solid-phase immuno-EM (SPIEM) as described elsewhere (17). In brief, the grid was coated with 5 g of protein A (Sigma) per liter of HEPES buffer (0.05 mol/L, pH 7) for 15 min, and the wet grid was placed on a drop of human polyomavirus BK (BKV)-specific rabbit antiserum (1:1000 diluted in HEPES buffer; kindly provided by G. Noss, Rehlingen-Siersburg, Germany). The conditioned grid was incubated with the untreated urine sample for 30 min, washed three times in a beaker with distilled water, and negatively stained as described above.
To determine the optimum SPIEM conditions, we first examined different antibodies directed against human polyomaviruses [directed against BKV, human polyomavirus JC (JCV), or simian virus 40 (SV40)] for their ability to aggregate BKV by mixing the different antibodies with urine samples. Briefly, for each antiserum three different dilutions (1:5, 1:50, and 1:500 in HEPES) were mixed 1:1 with polyomavirus-containing urine and incubated for 1 h at room temperature. These samples were then negatively stained, and the number and sizes of virus aggregates were assessed by EM.
In parallel, we compared the IgG-binding efficacy of different protein A concentrations (1, 5, 10, and 50 g/L) by incubating fluorescein isothiocyanate-coupled IgG antibodies (Sigma) with the grids and determining the fluorescence intensities of the antibodies bound to the grids with a fluorescence microscope (Zeiss Axiovert 100). Finally, we compared different dilutions of the most potent antibody (1:10, 1:100, 1:1000, and 1:10 000) by counting the virus particles bound to the EM grids.
uc of urine samples for em analysis
We ultracentrifuged 5 mL of untreated urine at 100 000g for 60 min. The supernatant was discarded, and the pellet was resuspended in 50 µL of sterile doubly distilled water (18). The samples were then negatively stained as described above.
pcr
Polyomavirus PCR was performed as described recently (4). Untreated urine (5 µL; diluted 1:10 in distilled water) was examined with a quantitative TaqMan PCR assay that detects BKV as well as JCV. Each 50-µL PCR assay contained 5 µL of template, 500 nM each of the primers PVTMfor and Pvback, 100 nM PVProbe exonuclease probe (for sequence data see Table 1
), 100 µM each of the deoxynucleotide triphosphates (Gibco), 5 µL of 10x amplification buffer, 1 µM ROX, and 2 U of Platinum Taq DNA polymerase (Gibco). After initial denaturation for 3 min, the sample was subjected to 45 cycles of 94 °C for 30 s and 62 °C for 30 s. Fluorescence intensity was read automatically during PCR cycling in an ABI Prism 7700 SDS.
|
Polyomavirus-positive urine samples were subjected to seminested PCR to differentiate BKV and JCV. Each 30-µL PCR mixture contained 5 µL of 1:10-diluted urine, 333 nM each of the primers PVsnfor and PVback (first PCR) or BKfor, JCfor, and PVback (second PCR; for sequence data see Table 1
), 50 µM each of the deoxynucleotide triphosphates (Gibco), 3 µL of 10x amplification buffer, and 1 U of Platinum Taq DNA polymerase (Gibco). After initial denaturation for 3 min, the sample was subjected to 30 cycles of 94 °C for 20 s and 53 °C for 20 s, with a final extension for 5 min at 72 °C for the first amplification round. One microliter of the resulting PCR product was used as template in the second amplification round. Conditions for the second round were the same as for the first, except that the annealing temperature was increased to 58 °C.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Whereas heavy metal shadowing clearly revealed the well-known regular capsid structure of polyomavirus particles (Fig. 1A
), the use of different negative stains revealed different appearances (Fig. 1
, B–H). However, all of the negative stains tested displayed the viral particles at sufficient quality for a clear diagnosis.
|
UC.
For the enrichment of virus particles from urine, samples were subjected to UC before negative staining and the results were compared with those obtained with direct EM, i.e., without enrichment procedures. The number of virus particles on the grid was clearly higher after UC (Fig. 2B
) compared with the same urine sample without enrichment (Fig. 2A
). In addition, some urine samples negative after direct EM were positive after UC enrichment (Table 2
, columns dEM and UC).
|
|
spiem
Immuno-specific enrichment of polyomaviruses had an effect similar to that of UC enrichment: The number of particles bound to the grid was increased (Fig. 2
, A and C), and the total number of positive urine samples was higher after SPIEM (Table 2
, columns dEM and SPIEM). Although SPIEM provided equal or better enrichment than UC, one urine sample remained negative after SPIEM, although it was positive after UC alone (Table 2
, sample 29).
Although the pH values of the urine samples typically were between 6 and 8, sample 29 had a relatively low pH of 5. For further examination, 10 selected urine samples (original pH values between 5.0 and 7.8) were adjusted to five different pH values ranging from 5 to 9 by titration with 1 mol/L HCl or 1 mol/L NaOH, respectively. The samples were then reexamined by SPIEM (see Table 3
). The negative SPIEM result for sample 29 became positive after the pH was increased to 
6, suggesting that the number of particles bound to the antibody-coated grid depends strongly on the pH of the urine sample, with optimal enrichment at physiologic pH.
|
Defining the routine protocol for em detection of polyomaviruses in urine.
The comparison of different negative stains revealed that all tested stains are, in principle, suitable for EM detection of polyomavirus and that a positive result is not dependent on the stain chosen. Because we felt most comfortable with the uranyl acetate staining results, especially in samples with large aggregates of debris and/or virus particles (Fig. 1H
), we decided to use this stain for the following experiments.
Comparison of direct EM with UC and SPIEM enrichment on 40 urine samples showed the benefit of enrichment procedures (see Fig. 2
and Table 2
). Nevertheless, when we compared UC and SPIEM, the superior enrichment achieved with SPIEM did not give better sensitivity: None of the urine samples negative after UC enrichment were positive after SPIEM (Table 2
). In addition, the SPIEM protocol was more sensitive to slight variations during preparation (e.g., pH; see Table 3
), took more time compared with UC, and depended totally on the availability of (limited) potent antiserum. Therefore, we chose UC for routine polyomavirus enrichment.
comparison of em and pcr
UC followed by negative staining with uranyl acetate was applied to 531 urine samples. The samples were read "blind", with the operator unaware of a patient’s identity to avoid biasing of the results. Qualitative and quantitative PCRs were performed on the same samples in parallel, and the results by EM and PCR were compared. With PCR, 149 urine samples were positive for human polyomaviruses BK and/or JC. With EM after UC, only 70 samples were positive (EM sensitivity, 47%). However, all of the samples positive by EM were also positive by PCR (EM specificity, 100%). Compared with PCR, the positive and negative predictive values of EM for polyomavirus were 100% and 89%, respectively. Interestingly, none of the urine samples PCR-positive for JCV was positive by EM.
The genome concentration was linearly related to the proportion of urine samples positive by EM (r2 = 0.9), as shown in Fig. 3
.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The evaluation of different heavy metal salts for negative staining showed that the quality of EM as a diagnostic method does not depend on the negative stain used. Although the particles looked different with different negative stains, the important morphologic features of the polyomavirus particles, i.e., size and shape, were preserved with all stains tested [see also Ref. (19)]. However, we chose to use one particular stain because it is better not to switch among stains because sensitivity might be lost while adapting to a new stain.
Our comparison of direct EM with UC or SPIEM enrichment clearly showed increased sensitivity after enrichment. Although enrichment after SPIEM was 10-fold higher than that after UC, no PCR-positive urine samples negative after EM/UC were positive after SPIEM. In addition, the SPIEM protocol takes more time and seems to be more susceptible to variability than UC enrichment. Therefore, we chose EM after UC enrichment for detecting polyomavirus particles in urine. However, if unambiguous identification of BKV is necessary instead of just detection of polyomavirus particles, SPIEM could be added.
Comparison of EM after UC with an up-to-date quantitative PCR assay revealed a remarkably good correlation between EM and PCR: The probability of a positive EM result increased proportionally with the DNA content of the urine samples. A definite detection limit for EM cannot be determined because a single virus particle with definite morphology will allow the diagnosis to be made. Because positive EM detection is not impossible at low particle concentrations but only less probable, EM is not necessarily limited to samples with high viral loads, as is often assumed (10). On the other hand, urine samples can be negative by EM despite containing numerous virus particles. These findings may be partly attributable to interindividual differences in the production of morphologically complete virus particles (positive by EM) compared with viral genomes (positive by PCR). However, the significance of large numbers of viral genomes in urine samples may be much less if no intact particles are produced.
Summing up the technical aspects of this evaluation, we examined various methods of preparing viruses for detection by EM, and the best method was compared with an up-to-date PCR assay (4). The influence of different negative stains was demonstrated as well as the need for carefully evaluating different enrichment procedures. When we assessed the diagnostic sensitivity of EM by comparison with quantitative PCR, we found that the diagnostic sensitivity of EM cannot be defined by a detection limit but rather by a probability of detection of >50% at 106 particles/mL, thus distinguishing EM from all other diagnostic assays. However, even after optimization, EM is not as sensitive as PCR for polyomaviruses in urine. This must be taken into account when looking for future diagnostic uses for EM (7)(8).
For the final appraisal of EM as a tool for BKV detection, several aspects have to be considered, i.e., speed, sensitivity, availability, and costs. Speed is one of the most commonly cited advantages of EM for diagnostic purposes (7)(20), but although it enables rapid diagnosis within 15 min, direct EM is clearly not as sensitive as the combination of EM with UC enrichment, which on the other hand takes at least 1.5 h to obtain results. This makes EM with UC enrichment slower than up-to-date real-time PCR assays, which can give results within 60 min or less (21). To speed up diagnosis based on EM, the Airfuge system could be used as a reasonable alternative to UC enrichment. This air-driven ultracentrifuge for small volumes (up to 500 µL) reaches enrichment factors similar to those for standard ultracentrifugation but within 15 min (22)(23). Thus, EM after Airfuge enrichment could lead to a diagnosis within 30 min, still enabling a rapid viral diagnosis.
The sensitivity of PCR is superior to that of EM for detection of polyomavirus. We showed that only at particle concentrations of 109 genome-equivalents/mL did EM reach a 100% probability of giving a correctly positive result. This weakens the value of a negative EM diagnosis in general, but in the diagnosis of BK viruria after BMT, the situation is a little different. It has recently been shown by Leung et al. (15) that in BMT patients with BKV-dependent hemorrhagic cystitis, the mean BKV load in the urine is 10 000 times higher than in asymptomatic patients (2 x 1012 vs 2.2 x 108 genome-equivalents/mL). Thus, in these cases the sensitivity of EM seems to be adequate for the detection of BKV; at these concentrations, EM will give a positive result even without enrichment. The increased sensitivity of PCR may help to answer questions about the pathogenesis of BKV-dependent hemorrhagic cystitis, it but does not seem to be essential for a positive diagnosis. Moreover, the extreme sensitivity of PCR may lead to the identification of BKV infections with a low viral load that may not have clinical consequences. One caveat is that these high viral loads for pathogenic BKV reactivation have been described only for BMT patients. For other immunocompromised individuals, e.g., renal transplant or AIDS patients, pathogenic BKV reactivation may already occur at much lower viral loads.
The rapid availability of EM is a prerequisite for rapid diagnosis based on EM. Depending on the size and the location of the diagnostic facility, the use of EM for viral diagnostics may be impossible because there is no such instrument. In these settings, light microscopic examination may lead to a first rapid evaluation of the urine. Although urinary cytology is not as sensitive and specific as EM, it has been shown to correctly diagnose 83% of EM-positive samples, with positive and negative predictive values of 63% and 96%, respectively (1). However, in contrast to EM detection, samples positive by urinary cytology must be additionally examined by PCR because other viral agents might be mistaken for BKV, e.g., adenoviruses.
Finally, the costs of EM are often described as too high compared with PCR. This argument seems to be correct if only the initial cost is considered, but this approach oversimplifies a complex situation because the per-assay costs of PCR are higher than those of EM. Furthermore, EM, with its "open view", is able to detect all pathogens in the urine sample (7), e.g., in addition to BKV, it also detects cytomegalovirus and adenoviruses. A more extensive discussion of the perceived costs of EM is given elsewhere (8).
In summary, we have shown that EM, despite its low sensitivity, is a useful tool for the detection of BKV in urine samples from BMT patients. It is faster than PCR, taking only 15 min (without enrichment) to 30 min (with Airfuge enrichment); moreover, the per-assay costs of EM are low, especially if an electron microscope is already available. We therefore recommend the use of EM as a diagnostic tool for BKV infection (13), to provide an initial rapid diagnosis, especially in life-threatening conditions. Because of the relative disadvantages of EM compared with PCR, i.e., relative lack of sensitivity of EM and its nonquantitative nature, quantitative PCR should be performed in parallel on the same sample (a) to provide superior sensitivity for samples with viral loads <107 genome-equivalents/mL, (b) to verify the EM result (in immunosuppressed individuals, this may also include cytomegalovirus and adenovirus detection), and (c) to monitor the course of infection under therapy (6).
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  A. Nitsche, M. Buttner, S. Wilhelm, G. Pauli, and H. Meyer Real-Time PCR Detection of Parapoxvirus DNA, Clin. Chem., February 1, 2006; 52(2): 316 - 319. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
. Linear regression of the data revealed a direct linear correlation between viral load and the proportion of urine samples positive after EM, as shown by the straight line (r2 = 0.9).