Article Images (0)
×![]()
featured_play_listSave article to folder
X
Create new folder
Full Text
PDF
Table of Contents
Abstract
Introduction
Methods
DNA Source Material
Panel Development And Composition
PCR Amplification
Statistical Analysis
Results
Discussion
Supplementary Material
Acknowledgments
Declaration Of Interest
AC, MC, MCE, CH, RG, FH, GJ, KJ, SK, CL, ML, LM, CM, COM,
Abstract
A wide array of PCR tests has been developed to aid the diagnosis of invasive aspergillosis (IA), providing technical diversity but limiting standardisation and acceptance. Methodological recommendations for testing blood samples using PCR exist, based on achieving optimal assay sensitivity to help exclude IA. Conversely, when testing more invasive samples (BAL, biopsy, CSF) emphasis is placed on confirming disease, so analytical specificity is paramount. This multicenter study examined the analytical specificity of PCR methods for detecting IA by blind testing a panel of DNA extracted from a various fungal species to explore the range of Aspergillus species that could be detected, but also potential cross reactivity with other fungal species. Positivity rates were calculated and regression analysis was performed to determine any associations between technical specifications and performance. The accuracy of Aspergillus genus specific assays was 71.8%, significantly greater (P < .0001) than assays specific for individual Aspergillus species (47.2%). For genus specific assays the most often missed species were A. lentulus (25.0%), A. versicolor (24.1%), A. terreus (16.1%), A. flavus (15.2%), A. niger (13.4%), and A. fumigatus (6.2%). There was a significant positive association between accuracy and using an Aspergillus genus PCR assay targeting the rRNA genes (P = .0011). Conversely, there was a significant association between rRNA PCR targets and false positivity (P = .0032). To conclude current Aspergillus PCR assays are better suited for detecting A. fumigatus, with inferior detection of most other Aspergillus species. The use of an Aspergillus genus specific PCR assay targeting the rRNA genes is preferential.
A wide array of PCR tests has been developed to aid the diagnosis of invasive aspergillosis (IA), providing technical diversity but limiting standardisation and acceptance. Methodological recommendations for testing blood samples using PCR exist, based on achieving optimal assay sensitivity to help exclude IA. Conversely, when testing more invasive samples (BAL, biopsy, CSF) emphasis is placed on confirming disease, so analytical specificity is paramount. This multicenter study examined the analytical specificity of PCR methods for detecting IA by blind testing a panel of DNA extracted from a various fungal species to explore the range of Aspergillus species that could be detected, but also potential cross reactivity with other fungal species. Positivity rates were calculated and regression analysis was performed to determine any associations between technical specifications and performance. The accuracy of Aspergillus genus specific assays was 71.8%, significantly greater (P < .0001) than assays specific for individual Aspergillus species (47.2%). For genus specific assays the most often missed species were A. lentulus (25.0%), A. versicolor (24.1%), A. terreus (16.1%), A. flavus (15.2%), A. niger (13.4%), and A. fumigatus (6.2%). There was a significant positive association between accuracy and using an Aspergillus genus PCR assay targeting the rRNA genes (P = .0011). Conversely, there was a significant association between rRNA PCR targets and false positivity (P = .0032). To conclude current Aspergillus PCR assays are better suited for detecting A. fumigatus, with inferior detection of most other Aspergillus species. The use of an Aspergillus genus specific PCR assay targeting the rRNA genes is preferential. Key words: Aspergillus PCR, analytical specificity, cross reactivity, detection range.
Introduction
The European Aspergillus PCR Initiative (EAPCRI) has provided recommendations for successfully detecting Aspergillus fumigatus DNA in whole blood, serum and plasma.1–4 Together with the MIQE guidelines, these studies provided a roadmap for establishing a standard for qPCR to aid diagnosis of invasive aspergillosis (IA).5 In line with the relatively low incidence of IA initial EAPCRI research focused on standardising the methodology required for testing samples that permitted screening strategies to be employed. Consequently, standardization aimed at achieving optimal sensitivity, and the subsequent negative predictive value required to exclude IA. Now the focus is on testing samples necessarily obtained by more invasive procedures (e.g., broncho-alveolar lavage fluid, cerebrospinal fluid and tissue biopsy), to confirm a diagnosis of IA through optimal specificity and by employing the positive predictive value. While research specific to the sample types continues, it was also decided to concentrate on the analytical specificity of the individual PCR assays currently in use, with particular emphasis on the range of Aspergillus species that an assay can detect and potential cross-reactivity with other fungal species. Since IA is most commonly caused by A. fumigatus, previous EAPCRI studies concentrated on this species but this may not reflect the global epidemiology of IA.6 Aspergillus species other than A. fumigatus are increasing in clinical importance and invasive fungal disease caused by other fungal genera may require specific anti-fungal therapies. It is important that any Aspergillus PCR assay be analytically specific, capable of detecting, and possibly differentiating, a range of Aspergillus species, without cross-reacting with other moulds. Conversely, it may be necessary to account for all these opportunistic pathogens by developing panfungal assays to detect a broad range of fungi, including Aspergillus spp. To address this the EAPCRI distributed a panel of genomic DNA from a range of Aspergillus species together with other fungi to determine the analytical specificity of assays from centres using molecular assays to aid in the diagnosis of IA. D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024
Methods
DNA source material
Conidia were harvested from a sporulating culture of each mould using a dry cotton swab. For Candida albicans a sterile 10 µl loop was used to harvest biomass from an overnight culture. All fungi were resuspended into 3 ml of sterile water containing a drop of Tween 20 to prevent clumping. A 1 ml volume of each fungal suspension was harvested by centrifugation at 10 000 g for 5 minutes. The supernatant was discarded, and biomass was exposed to mechanical disruption using the equivalent of 20 µl of ceramic beads (Roche, Burgess Hill, UK) and 30 seconds bead-beating using a mini-bead-beater (Biospec Products, Bartlesville, OK, USA). After pulse centrifugation, the beads were washed with 200 µl molecular grade water and DNA was extracted using the Qiagen EZ1 Advance XL Tissue kit (Qiagen, Hilden, Germany) as per manufacturer’s instructions using an elution volume of 100 µl. The DNA was quantified in triplicate by nanophotometer (Implen P300, Geneflow, Lichfield, UK) and diluted in TrisEDTA buffer (Sigma-Aldrich, Dorset, UK) to the desired concentration.
Panel development and composition
The panel was designed to determine the analytical specificity of each assay, namely, the capacity to detect various species of Aspergillus (true positives) plus potential cross reactivity with other fungal species (false positives). The panel was composed of genomic DNA extracted from the following fungal cultures: A. fumigatus, A. flavus, A. lentulus, A. niger, A. terreus, A. versicolor, C. albicans, Lichtheimia (formerly Absidia) corymbifera, Cunninghamella sp., Fusarium oxysporum, Scedosporium apiospermum, and a Penicillium sp. All were phenotypically and microscopically identified clinical isolates that were stored in the culture repository of the Public Health Wales Mycology Reference Laboratory. For the six Aspergillus species two concentrations of DNA were included for each species, representing strong positives (≈100 fg/µl), potentially encountered when testing respiratory specimens and weak positives (1–10 fg/µl), more typically seen when testing blood based samples. For the additional six genera strong positives (100–500 fg/µl) were incorporated to challenge the specificity of the assay and represent the scenario potentially encountered when testing respiratory samples. The panel was composed of 18 × 1.5 ml screw cap microfuge tubes containing 30 µl of each DNA extract, the contents of each tube was made anonymous to every participant. After development the panel was quality assured by performing genus specific qPCR assays.7–9 and Fusarium and Scedosporium PCR unpublished data. After assembly all panels were frozen at −80◦C with no freeze/thaw intervals. The panels were distributed to 27 participating centres on dry ice for next day delivery by courier. To mimic the duration of distribution the distributing centre retained an additional panel at −80◦C until all panels had been safely delivered. All 28 participating centers were asked to confirm receipt, comment on the state of the panel (frozen or thawed) and keep specimens frozen at −80 ◦C until testing.
PCR amplification
Participating centers were asked to perform their current PCR assay used to detect Aspergillus species in duplicate. Results were recorded as detected, not detected or inhibitory (as determined by local internal control PCR), with centers requested to return the quantification cycle (Cq) for positive results. In addition they were asked to provide the following technical details: PCR platform used, target gene, detection range of assay, whether an internal control PCR was performed, the DNA template input volume and final PCR reaction volume, within a designated time frame.
Statistical analysis
For analysis PCR positivity associated with DNA from Aspergillus species was considered true positivity, whereas PCR positivity associated with DNA from other fungal genera was considered false positivity. Analytical sensitivity (the proportion of true positives containing Aspergillus sp. DNA detected by the PCR assay, N = 12), analytical specificity (the proportion of potential false positives containing other fungal DNA not detected by the PCR assay, N = 6) and overall accuracy (the proportion of samples containing Aspergillus DNA detected plus the proportion of samples containing other fungal DNA not detected, N = 18) were determined by the generation of 2 × 2 tables. Individual sample positivity was calculated considering all replicates tested. To minimize the influence of process-borne false positivity when calculating analytical sensitivity, analytical specificity and accuracy of each test >50% of replicates were required to be positive before the individual sample was considered positive. Ninety-five percent confidence intervals were generated for each proportionate value and, where required, two-tailed P values (Fisher exact test; P = .05) to determine the significance of the difference between rates.10 To determine any associations between the technical factors provided and performance linear and D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 logistic regression analysis was performed with significance defined by a P value of .05.
Results
A total of 28 centers participated in the study returning 33 data sets (five centres returned two data sets: three centers used a combined testing strategy utilizing A. fumigatus specific and Aspergillus genus specific PCR assays, while one center evaluated a commercial Candida/Aspergillus specific PCR versus an “in-house” Aspergillus genus specific PCR assays, and one center evaluated two different inhouse Aspergillus genus specific PCR assays) comprising 54 replicates for the 18 samples. Thirty assays were performed on qPCR platforms from six different manufacturers and three assays were performed on conventional block based platforms (Table 1). Twelve, nine, eight, and four assays targeted the 28S rRNA gene, internal transcribed sequence (ITS), 18S rRNA gene and other genes, respectively. Twenty-five assays were designed to detect the Aspergillus genus, seven were specific to A. fumigatus, and one was specific to A. fumigatus and A. terreus. Although one of the Aspergillus sp. assays and one of the A. fumigatus specific assays were performed in combination with a panfungal PCR only the front-line Aspergillus assays were included in analysis. The median DNA template input and final PCR reaction volumes were 7.5 µl (range: 1–20 µl) and 25 µl (range: 10–50 µl), respectively, and the median input/final volume ratio was 0.25 (range: 0.1–0.5) (Table 1).
Sample positivity rates
At both concentrations the detection of A. fumigatus was superior, generating true positivity rates of 98.2% (53/54, 95% CI: 90.2–99.7) and 75.9 (41/54, 95% CI: 63.1–85.4) for 100 and 10 fg/µl of DNA, respectively (Fig. 1). The positivity rate for A. fumigatus was significantly greater than the overall combined true positivity rates at both concentrations (100 fg/µl 226/324: 69.8%, 95% CI: 64.5–74.5; 10 fg/µl 134/324: 41.4%, 95% CI: 36.1–46.8) (P < .0001). At the higher concentration, detection of A. lentulus was not significantly inferior to that of A. fumigatus (47/54: 87.0%, 95% CI: 75.6–93.6; P = .06), and it yielded the second highest positivity rate. The detection of all other Aspergillus species at the higher concentration was significantly less than both A. fumigatus (P < .0001) and A. lentulus (P = .0359). The lowest true positivity rate at the higher concentration was seen for the detection of A. versicolor (22/54: 40.7%, 95% CI: 28.7–54.0). At the lower concentration there was a reduction in positivity for all Aspergillus species tested (Fig. 1). This was greatest for the detection of A. lentulus, where there was 68.5% reduction (P < .0001). There were also significant reductions for A. niger (33.4%, P = .0009), A. flavus (27.8%, P = .0063) and A. fumigatus (22.2%, P = .009), but not for A. versicolor (14.8%, P = .1526) and A. terreus (11.1%, P = .3356). The positivity rate for detecting the lower concentration of A. fumigatus DNA was significantly greater than that of all other species tested (P < .0052). The second highest rate at the lower concentration was associated with A. terreus (25/54: 46.3%, 95% CI: 33.7–59.4), and the lowest rate was associated with A. lentulus (10/54: 18.5%, 95% CI: 10.4–30.8). The overall false positivity rate was 14.5% (47/324, 95% CI: 11.1–18.8) and was significantly less than the combined true positivity rate when detecting Aspergillus DNA (P < .0001). The highest rate was associated with cross reactivity with Penicillium DNA (17/54: 31.5%, 95% CI: 20.7–44.7), followed by cross reactivity with Fusarium DNA (10/54: 18.5%, 95% CI: 10.3–30.8) (Fig. 1).
Assay performance
The combined performance as determined by accuracy, analytical sensitivity and specificity for each individual assay is shown in Table 1, and a breakdown of performance with respect to the individual samples is shown in Table S1. The median accuracy was 61.1% (range: 33.3–94.4, 25% percentile: 50.0, 75% percentile: 83.3). The median analytical sensitivity was 50.0% (range: 0.0–91.7, 25% percentile: 25.0, 75% percentile: 83.3). The median analytical specificity was 100% (range: 66.7–100, 25% percentile: 83.3, 75% percentile: 100). The mean analytical sensitivity and specificity were 52.5% (208/396; 95% CI: 47.6–57.4) and 92.4% (183/198: 95% CI: 87.9–95.4). Consequently, the generation of false positive results was not as important as the generation of false negatives (P < .0001). Overall accuracy was significantly greater (P < .0001) for the assays designed to be Aspergillus genus specific (71.8%) compared to the assays specific to individual species (47.2%) (Table 2). Given the design of the panel, including six Aspergillus species, this was not unexpected and was driven by the analytical sensitivity, and the mean for the Aspergillus genus specific assays was 62.3%. For the Aspergillus genus specific assays the most regularly missed species were A. lentulus (25.0% of total false negative results), A. versicolor (24.1%), A. terreus (16.1%), A. flavus (15.2%), A. niger (13.4%), and A. fumigatus (6.2%). The mean analytical sensitivity for assays specific for individual Aspergillus species was 21.9%, significantly less than that of the Aspergillus genus specific assays (P < .0001) (Table 2). However, the sensitivity when detecting the D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 Ta b le 1. C h ar ac te ri st ic s an d p er fo rm an ce o f th e P C R as sa ys u se d in ea ch ce n te r. C en te r A ss ay N um be r A ss ay V ol um e T em pl at e V ol um e PC R Pl at fo rm In te rn al G en e A ss ay A cc ur ac y Se ns it iv it y Sp ec ifi ci ty N um be r (I H or C ) (µ l) (µ l) C on tr ol T ar ge t T ar ge t (% ) (% ) (% ) 1 1 (I H ) 50 15 R ot or ge ne Q + 28 S A sp er gi llu s sp p. 83 .3 83 .3 83 .3 1 2 (C ) 50 10 A B I V er it i + 28 S A sp er gi llu s sp p. 77 .8 75 .0 83 .3 2 3 (I H ) 50 10 A B I 75 00 + 18 S A sp er gi llu s sp p. 83 .3 83 .3 83 .3 3 4 (I H ) 21 10 B io R ad St ep O ne pl us + IT S A sp er gi llu s sp p. a 77 .8 66 .7 10 0 4 5 (I H ) 50 10 R oc he L C 48 0 + 28 S A sp er gi llu s sp p. 83 .3 75 .0 10 0 5 6 (I H ) 20 8- 10 R oc he L C 2. 0 + Fa ct or C ge ne s A .f um ig at us /t er re us b 44 .4 16 .7 10 0 6 7 (I H ) 21 10 B io R ad C FX 96 + IT S A sp er gi llu s sp p. 72 .2 58 .3 10 0 7 8 (I H ) 20 5 R oc he L C 48 0 − 18 S A sp er gi llu s sp p. 94 .4 91 .7 10 0 8 9 (I H ) 21 7 A B I7 50 0 + IT S A .f um ig at us 50 .0 25 .0 10 0 9 10 (I H ) 20 8 R oc he L C 48 0 + IT S A sp er gi llu s sp p. 50 .0 25 .0 10 0 10 11 (I H ) 25 7. 5 A B I7 50 0 + 18 S A sp er gi llu s sp p. 83 .3 83 .3 83 .3 10 12 (I H ) 25 7. 5 A B I7 50 0 + 28 S A .f um ig at us 55 .6 41 .7 83 .3 11 13 (I H ) 20 5 R ot or ge ne 60 00 + 28 S A sp er gi llu s sp p. 61 .1 41 .7 10 0 12 14 (I H ) 20 2 R oc he L C 48 0 + IT S A .f um ig at us 50 .0 25 10 0 13 15 (I H ) 20 5 A B I7 50 0 + 28 S A sp er gi llu s sp p. 61 .1 41 .7 10 0 14 16 (I H ) 25 5 R ot or ge ne 30 00 + IT S2 A sp er gi llu s sp p. 88 .9 83 .3 10 0 15 17 (I H ) 10 1 B io R ad C FX 96 − 18 S A sp er gi llu s sp p. 61 .1 58 .3 66 .7 16 18 (I H ) 30 10 R oc he L C 48 0 + 18 S A sp er gi llu s sp p. 55 .6 50 .0 66 .7 16 19 (I H ) 30 10 R oc he L C 48 0 + 28 S A sp er gi llu s sp p. 88 .9 83 .3 10 0 17 20 (I H ) 50 9 R oc he L C 48 0 + M it o A sp er gi llu s sp p. 61 .1 41 .7 10 0 17 21 (I H ) 50 9 R oc he L C 48 0 + 18 S A .f um ig at us 50 .0 25 10 0 18 22 (I H ) 20 5 A B I7 90 0 + C yt oc hr om e B A .f um ig at us c 44 .4 16 .7 10 0 19 23 (I H ) 50 5 B lo ck - N S − 18 S A sp er gi llu s sp p. 77 .8 66 .7 10 0 20 24 (I H ) 50 5 B lo ck - N S − C yp 51 a A sp er gi llu s sp p. 33 .3 0. 0 10 0 21 25 (I H ) 20 2 R oc he L C 1. 2 − IT S2 A sp er gi llu s sp p. 66 .7 58 .3 83 .3 22 26 (I H ) 20 5 A B I7 50 0 + 28 S A .f um ig at us 44 .4 16 .7 10 0 22 27 (I H ) 20 5 A B I7 50 0 + 18 S A sp er gi llu s sp p. 83 .3 91 .7 66 .7 23 28 (I H ) 50 20 L C 48 0 + 28 S A sp er gi llu s sp p. 83 .3 83 .3 83 .3 24 29 (I H ) 30 10 A B I7 90 0H T + 28 S A sp er gi llu s sp p. 88 .9 91 .7 83 .3 25 30 (I H ) 50 20 M X 30 00 + 28 S A sp er gi llu s sp p. 55 .6 41 .7 83 .3 26 31 (I H ) 20 5 R oc he L C 2. 0 + IT S2 A sp er gi llu s sp p. d 55 .6 33 .3 10 0 D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 Ta b le 1. (c o n ti n u ed ) C en te r A ss ay N um be r A ss ay V ol um e T em pl at e V ol um e PC R Pl at fo rm In te rn al G en e A ss ay A cc ur ac y Se ns it iv it y Sp ec ifi ci ty N um be r (I H or C ) (µ l) (µ l) C on tr ol T ar ge t T ar ge t (% ) (% ) (% ) 27 32 (I H ) 10 2 Il lu m in a E co − IT S A .f um ig at us 38 .9 8. 3 10 0 28 33 (I H ) 25 10 R ot or ge ne + 28 S A sp er gi llu s sp p. 66 .7 50 10 0 a A ss ay ta rg et s A .f um ig at us ,A .fl av us ,A .t er re us . b A ss ay ta rg et s A .f um ig at us an d A .t er re us w it h in di vi du al pr ob es . c A ss ay pe rf or m ed in co m bi na ti on w it h a pa nf un ga lP C R ta rg et in g th e 18 S rR N A re gi on us in g th e A B I 79 00 .O nl y re su lt s fo r th e A sp er gi llu s sp ec ifi c PC R ha ve be en in cl ud ed in th e an al ys is . d A ll ne ga ti ve A sp er gi llu s PC R sa m pl es w er e th en te st ed by a pa nf un ga lP C R ta rg et in g th e IT S2 re gi on us in g th e R oc he L ig ht C yc le r 2. 0. O nl y re su lt s fo r th e A sp er gi llu s sp ec ifi c PC R ha ve be en in cl ud ed in th e an al ys is . K ey : 28 S: 28 S rR N A ge ne 18 S: 18 S rR N A ge ne C Y P5 1A :L an os te ro l1 4- α -d em et hy la se ge ne M it o: M it oc ho nd ri al R N A ge ne C :C om m er ci al PC R (R D L Fu ng ip le x A ss ay ) A B I: A pp lie d B io sy st em s In st ru m en t M X 30 00 :S tr at ag en e M X 30 00 qP C R in st ru m en t. IH :I nho us e PC R IT S: In te rn al tr an sc ri be d se qu en ce L C :R oc he L ig ht C yc le r B lo ck -N S: C ov en ti on al PC R in st ru m en t – no t sp ec ifi ed . individual designated species was 72.2% (13/18: 95% CI: 49.1–87.5). There was no significant difference (P = .1238) between the mean analytical specificity for the Aspergillus genus and individual Aspergillus species PCR assays, that were 90.7% and 97.9%, respectively.
Technical factors influencing performance
A summary of performance as influenced by individual technical variables is shown in Table 2. Regression analysis showed there were no significant associations between accuracy and the PCR amplification platform, DNA template volume, final reaction volume, or the ratio of these two volumes (Table 3). However, there were significant associations between accuracy and the PCR target and whether the assay was specific for the genus Aspergillus or for individual Aspergillus species (see above). For assays targeting the 18S and 28S rRNA genes (N = 20) the combined accuracy was 71.9% (259/360: 95% CI: 67.1–76.3), whereas for assays targeting the ITS region or other genes (n = 13) the combined accuracy was 56.4% (132/234: 95% CI: 50.0–62.6). Multivariate regression analysis confirmed significant positive associations of Aspergillus genus PCR targeting the 18S/28S rRNA genes with greater than average accuracy (P = .0095) and accuracy within the top 25% percentile (P = .0246). There were also significant positive associations between analytical sensitivity and the gene targeted by PCR, the use of an Aspergillus genus PCR and using larger reaction volumes, and a positive trend with the use of larger DNA template volume (Tables 2 and 3). PCR assays targeting the 18S/28S rRNA genes generated a combined analytical sensitivity of 63.8% (153/240: 95% CI: 57.5–69.6) compared to 35.3% (55/156: 95% CI: 28.2–43.0) for assays targeting the ITS region or other genes. In relation to analytical specificity there was a negative association with PCR assays targeting the 18S/28S rRNA genes (Table 3), driven by assays targeting the 18S rRNA gene (Table 2). For the eight assays targeting the 18S rRNA the combined analytical specificity was 83.3% (40/48: 95% CI: 70.4–91.3) compared to 95.8% (143/150: 95% CI: 90.7–97.7) for the 25 assays targeting other regions (P = .0818). Multivariate analysis showed that an Aspergillus genus PCR targeting the 18S/28S rRNA genes provided greater than average sensitivity (P = .0011), with sensitivity in the top 25% percentile (P = .0001). However, using an Aspergillus genus PCR targeting the 18S/28S rRNA genes had significant associations with below average specificity (P = .0067). D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 Figure 1. Summary of sample positivity for each sample tested, the mean Cq, and standard deviation values for positive assays for each sample. The values presented are the percentage of the total number of assays returning positive results for each sample, with positivity rate defined using all replicates (N = 54). This Figure is reproduced in color in the online version of Medical Mycology.
Discussion
Predictably, when assessing the detection range of Aspergillus PCR methods, the highest rates of detection were seen for A. fumigatus (Fig. 1). Interestingly, the detection of A. fumigatus and the former cryptic species A. lentulus was comparable at the higher DNA concentration, but at the lower concentration of DNA A. lentulus detection achieved the lowest of the six species tested.11 Comparison of the Cq values showed that despite A. lentulus recording the highest (latest) mean Cq value, at the higher DNA concentration 4.4 cycles after A. fumigatus, it was the only species to generate a positivity rate that was not significantly inferior to A. fumigatus. The DNA concentrations for all species were confirmed by testing multiple replicates, indeed for A. lentulus this was performed by two centres, so dilution errors are highly unlikely. Given the genome sizes for Aspergillus species range from approximately 2–4 × 107 bp, then differences in the number of genomes per DNA dilution cannot explain the differences in Cq, which equate to >1 log10. One explanation is that the rRNA copy num- ber per A. fumigatus genome can vary by approximately 1 log10.12 If this applies to A. lentulus, then it is possible that the strain used in this study possessed a very low copy number, so when diluted to the lower concentration the resultant Cq values reflected this and were in the nonreproducible range of real-time PCR detection (>39 cycles), with the affect enhanced if oligonucleotide mismatches with A. lentulus were evident. At the higher DNA concentration the detection of A. lentulus was significantly greater than the other non-fumigatus species and probably reflects the sequence similarity between A. fumigatus and A. lentulus and the fact that most Aspergillus PCR assays will be primarily designed and optimized for the detection of A. fumigatus. Aspergillus terreus was also regularly missed, which will present a challenge to those centers that have a high incidence of IA due to this organism. It would likely be beneficial if Aspergillus PCR assays possessed the ability to detect but also differentiate infections caused by A. terreus, which is potentially resistant to polyenes and requires alternative antifungal therapies. D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 Table 2. The effect of technical factors on the analytical performance of Aspergillus PCR methods. Parameter (%, n/N: 95% confidence interval) Technical variable (no. of assays) Accuracy Analytical Sensitivity Analytical Specificity PCR target 28S rRNA (12) 70.8 (153/216: 64.5–76.5) 60.4 (87/144: 52.3–68.0) 91.7 (66/72: 83.0–96.1) 18S rRNA (8) 73.6 (106/144: 65.9–80.1) 68.8 (66/96: 58.9–77.2) 83.3 (40/48: 70.4–91.3) ITS region (9) 61.1 (99/162: 53.4–68.2) 42.6 (46/108: 33.7–52.0) 98.1 (53/54: 90.2–99.7) Other regionsa (4) 45.8 (33/72: 34.8–57.2) 18.8 (9/48: 10.2–31.9) 100 (24/24: 86.2–100) PCR volumes Input >Median (16) 70.1 (202/288: 64.6–75.1) 59.4 (114/192: 52.3–66.1) 91.7 (88/96: 84.4–95.7) Input ≤Median (17) 61.8 (189/306: 56.2–67.0) 46.1 (94/204: 39.4–52.9) 93.1 (95/102: 86.5–96.6) Total >Median (13) 70.9 (166/234: 64.8–76.4) 61.5 (96/156: 53.7–68.8) 89.7 (70/78: 81.1–94.7) Total ≤Median (20) 62.5 (225/360: 57.4–67.3) 46.7 (112/240: 40.5–53.0) 94.2 (113/120: 88.5–97.2) Ratio >Median (14) 68.3 (172/252: 62.3–73.7) 57.1 (96/168: 49.6–64.4) 90.5 (76/84: 82.3–95.1) Ratio ≤Median (19) 64.0 (219/342: 58.8–68.9) 49.1 (112/228: 42.7–55.6) 93.9 (107/114: 87.9–97.0) PCR platforms Rotorgeneb (4) 75 (54/72: 63.9–83.6) 64.6 (31/48: 50.4–76.6) 95.8 (23/24: 79.8–99.3) Roche LCc (12) 65.3 (141/216: 58.7–71.3) 50.7 (73/144: 42.6–58.7) 94.4 (68/72: 86.6–97.8) ABId (9) 66.0 (107/162: 58.5–72.9) 54.6 (59/108: 45.2–63.7) 82.8 (48/58: 71.1–90.4) Othere (8) 61.8 (89/144: 53.7–69.3) 46.9 (45/96: 37.2–56.8) 91.7 (44/48: 80.5–96.7) PCR range Asp genus (25) 71.8 (323/450: 67.5–75.7) 62.3 (187/300: 56.7–67.6) 90.7 (136/150: 84.9–94.4) A. fumi/terr (8) 47.2 (68/144: 39.2–55.3) 21.9 (21/96: 14.8–31.1) 97.9 (47/48: 89.1–99.6) Note: All variables were analysed independently of other covariables. aother regions include the lanosterol 14-α-demethylase gene, mitochondrial RNA gene, cytochrome B and Factor C genes. bIncludes Rotorgene 3000, 6000, and Q models. cIncludes LC1.2, 2.0 and 480 models. dIncludes ABI7500 and 7900 models. eIncludes Illumina ECO, Stratagene MX3000, BioRad Stepone plus and CFX96 models and non-specific conventional PCR assays Key: 28S rRNA: 28S ribosomal RNA gene 18S rRNA: 18S ribosomal RNA gene ITS: internal transcribed sequence Asp Genus: PCR targets the genus Aspergillus A. fumi/terr: PCR targets A. fumigatus (N = 7) or A. fumigatus and A. terreus (N = 1). Table 3. Associations between technical variables and Aspergillus PCR performance as determined by logistic regression analysis. Performance Parameter Accuracy Sensitivity Specificity Technical Parameter Coefficient P value Coefficient P value Coefficient P value Platforma 1.2321 .1182 1.1451 .2935 −1.1104 .1690 PCR targetb 2.2842 .0439 2.0794 .0301 −2.6856 .0032 PCR rangec NAd .003 NAd .003 −1.7047 .1357 DNA template volume (A)e 0.1413 .1077 0.0478 .0923 −0.0669 .3507 Final reaction volume (B)e 0.1866 .2000 0.1142 .0483 −0.0789 .7291 Ratio of volumes A/Be 0.0006 .6016 0.003 .6747 0.0006 .7429 aWhen determining associations with accuracy and sensitivity the performance of the Rotorgene qPCR platforms (the platform with the highest accuracy and sensitivity (Table 2)) was compared to the other platforms. For specificity the performance of the ABI qPCR platforms (the platform with the lowest specificity (Table 2)) was compared to the other qPCR platforms. bWhen determining associations with all the performance parameters PCR assays targeting the rRNA genes (the gene targets with highest accuracy/sensitivity but lowest specificity (Table 2)) were compared to other gene targets. cWhen determining associations with all the performance parameters Asergillus genus specific PCR assays (the assays with the highest accuracy and sensitivity (Table 2)) were compared to the PCR assays specific to individual species. dNA: not applicable. As a result of perfect predictor model where all non-Aspergillus genus specific PCR assays were associated with below average accuracy and sensitivity the regression coefficient values are not accurate and the analysis has been replaced with a Fisher exact P value determined for odds ratio. eWhen determining associations with all the performance parameters volumes that were greater than the designated median value were compared with volumes equal to, or lesser than the median value. D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 Figure 2. Neighbour-joining tree generated from a nucleotide alignment (clustal, http://www.ebi.ac.uk/Tools/msa/) of ITS regions from the fungi used in this study. ITS sequences were selected by searching the NCBI nucleotide database using the species name and selecting sequences that spanned the ITS1, 5S, and ITS2 regions. The groupings indicate that any pan-Aspergillus PCR assay targeting the ITS region could potentially cross-react with members of the genus Penicillium, as illustrated by the three representative species. There was a cost associated with targeting multiple species as these assays were prone to false positives, particularly if the 18S rRNA gene is targeted. DNA from Penicillium spp. and Fusarium spp. were also frequently detected (Fig. 1). These were strong positive results with mean Cq values of 35.7 and 31.2, respectively. While Fusarium and Aspergillus spp. should be easily differentiated at the genetic level, in this study most cross reactivity with Fusarium spp. was associated with assays targeting the 18S or 28S rRNA genes. While regions within these genes are genetically dis- tinct large sequences are conserved across fungi, and even eukaryotes, where inappropriate oligonucleotide sequence selection can result in false positivity, including cross reactivity with human DNA.13 Obviously, this problem applies equally to Penicillium sp. but is further compounded by sequence similarities across the ITS regions (Fig. 2). This indicates that there is a need to design assays to deal with the problem of analytical specificity. There is a clear choice to be made in PCR-based diagnosis of IA, either target specific fungi or employ a panfungal assay. D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024 It is a logical aim for diagnostic centres to try to develop a truly pan-Aspergillus PCR assay and overcome the variations in performance seen in this study. However, the genus Aspergillus is a large taxonomic group and under dual naming contained at least eight telemorph genera including Neosartorya (A. fumigatus), Petromyces (A. flavus), and Emericella (A. nidulans).14,15 The “one species one name” system masks this diversity.16,17 An analysis of the internal transcribed spacer (ITS) regions of the fungi used in this study illustrates the issue of heterogeneity in Aspergillus, showing that Aspergillus spp. do not group as a single coherent cluster (Fig. 2). For example, A. versicolor groups more closely with the other fungi and A. flavus more closely with the representative Penicillium spp. Although the ITS regions can be used to identify fungal species to a high degree of accuracy allowing fungal barcoding, the qPCR assays used for rapid detection of IA can only focus on short sequences thereby reducing specificity compared to sequencing and resultant cross-reactivity with, in particular, the genus Penicillium.18 Panfungal assays that combine PCR amplification and sequencing are a potential solution but the time required for sequencing would delay diagnosis which could postpone initiation of effective treatment, thereby increasing patient mortality.19 Panfungal assays can lack sensitivity, particularly if more than one fungal species is present in a specimen (e.g., from the respiratory tract) when fungal colonisation is present. Rapid Aspergillus qPCR diagnosis requires careful assay design and needs to be targeted to a limited range of closely-related species or a single species, alternatively it may necessary to accept cross-reactivity with other species (e.g., Penicillium spp.) when using broad range pan-Aspergillus PCR. Indeed, in this study the generation of false positives through cross-reactivity was less evident than false negativity when detecting the lower, still clinically relevant, concentrations of Aspergillus DNA. Given the design of the panel including a range of Aspergillus species assays designed to only detect A. fumigatus were unlikely to perform optimally, but this remains a clinical limitation of this type of assay. Subsequently, the accuracy of the Aspergillus genus specific assays was significantly greater than those aimed at individual species, and this was driven by the greater detection range (analytical sensitivity) of the genus specific assays. There were positive associations between analytical sensitivity and Aspergillus genus PCR assays targeting the ribosomal RNA genes, likely as a result of the multi-copy nature of these genes. Although there was a trend for assays targeting the 18S rRNA to be associated with false positivity excluding it from the model removed the positive association between targeting the rRNA genes and sensitivity. Consequently, it was retained as the positive association with sensitivity outweighed the negative association with specificity, benefitting overall assay accuracy. Limitations of the study include the absence of no template controls, lacking any genomic DNA, to monitor for environmental contamination entering during the molecular process. In this study it would not be possible to differentiate this from potential cross reactivity with DNA from non-Aspergillus genera. However, given this study only required PCR amplification to be performed the opportunity for environmental contamination was limited. Furthermore, the previous studies of the EAPCRI did contain no template controls and monitored the entire molecular process; consequently, specificity values from these studies are more indicative of environmental false positivity.1,3,20 A further limitation is that inter-laboratory variance associated with nucleic acid extraction was not investigated, and it was assumed that extraction from different species would be equally efficient across platforms. However, the focus of this investigation was on the accuracy of PCR amplification and it was decided that removal of the extraction variable was beneficial to achieving this goal. In conclusion, current Aspergillus PCR assays are better suited to detecting A. fumigatus with detection of most other Aspergillus species appearing markedly inferior. From a clinical perspective the reduction in detection rates for non-fumigatus Aspergillus species is concerning, particularly in areas where the rates of infection by non-fumigatus species predominate (e.g., A. terreus in Tyrol, Austria) and assays optimized for the detection of A. fumigatus could lead to false negative results.21 Conversely, false positives derived through cross reactivity with non-Aspergillus species can occur, although in this study high DNA burdens were selected to stringently test analytical specificity. It should also be considered that for low incidence diseases, such as IA, screening strategies utilising high sensitivity are optimal and false negatives become more problematic when excluding disease. However, from a diagnostic perspective PCR positivity should not be the sole driver for therapy, and positivity should be considered in the clinical context with reference to supporting radiological and mycological evidence. The use of Aspergillus genus-specific PCR assays targeting the rRNA genes are preferred to species-specific assays, although the development of genus-specific assays may be compromised by the genetic diversity within the genus Aspergillus.
Supplementary Material
Supplementary data are available at MMYCOL online. D ow nloaded from https://academ ic.oup.com /m m y/article/55/4/402/2628999 by guest on 22 M ay 2024
Acknowledgments
We acknowledge the European Aspergillus Initiative of the ISHAM. The EAPCRI Steering Group consists of the following members: J. Peter Donnelly, chair of foundation, Radboud University Medical Centre, Nijmegen, The Netherlands; Juergen Loeffler, Secretary, Wuerzburg University, Wuerzburg, Germany; Rosemary A Barnes, treasurer, Cardiff University, Cardiff, United Kingdom. The EAPCRI Laboratory Working Party consists of the following members: Juergen Loeffler, Wuerzburg University, Wuerzburg, Germany. P. Lewis White, Public Health Wales Microbiology Cardiff, UK. Lena Klingspor, Karolinska University Hospital, Stockholm, Sweden. C. Oliver Morton, Western Sydney University, Sydney, Australia. Manuel Cuenca-Estrella, Spanish National Centre for Microbiology, Instituto de Salud Carlos III, Madrid, Spain. Katrien Lagrou, Laboratory of Clinical Bacteriology and Mycology, University Hospital Leuven, Belgium. Stephane Bretagne, Paris Diderot, Sorbonne Paris Cité University, Faculty of Medicine, Paris, France. Willem Melchers, Radboud University Nijmegen Medical Centre, Nijmegen, Netherlands. Carlo Mengoli, University of Padua, Padua, Italy. We thank the following members of the participating laboratories: Ferry Hagen and Jacques Meis, Department of Medical Microbiology and Infectious Diseases, Canisius-Wilhelmina Hospital, Nijmegen. The Netherlands; Kathleen Harvey-Wood, Southern General Hospital, Glasgow, Scotland, United Kingdom; Laurence Millon, Laboratoire de Parasitologie-Mycologie Centre Hospitalier Universitaire, Bescancon, France; Markus Ruhnke, Charité Medical School, University of Berlin, Berlin, Germany; Melinda Paholcsek, University of Debrecen Medical and Health Science Center, Human genetics Department, Life Science Building, Debrecen, Hungary; Massimo Cogliati, Dip. Scienze Biomediche per la Salute, Università degli Studi di Milano, Milan, Italy; Yvette Debets-Ossenkopp. Dept. Medical Microbiology and Infection control. VU University Medical Centre, Amsterdam, The Netherlands; Petr Hamal, Department of Microbiology, Faculty of Medicine and Dentistry, Palacky University Olomouc, Czech Republic; Wolfgang Mutschlechner and Cornelia Lass-Florl, Division of Hygiene and Medical Microbiology Innsbruck Medical University, Innsbruck, Austria; Catriona Halliday and Sue Sleiman, Clinical Mycology Reference Laboratory, Pathology West, Westmead, Australia; Chris Linton and Elizabeth Johnson, UK Mycology Reference Lab, Public Health England, Bristol, United Kingdom; Rebecca Gorton and Chris Kibbler, Royal Free Hospital, London, United Kingdom; Martina Lengerova, Department of Internal Medicine - Hematology and Oncology, University Hospital Brno, Brno, Czech Republic, Czech Republic; Alida Fe Talento, Katie Dunne and Tom Rogers, Department of Clinical Microbiology, Trinity College, Dublin, Ireland; Deborah Abdul-Ali Emory University, Atlanta, GA, USA; Gemma Johnson, Blizard Institute of Cell and Molecular Science, Queen Mary University of London, London, United Kingdom; Maria J. Buitrago and Leticia Bernal-Martinez, Spanish National Centre for Microbiology, Madrid, Spain; Katia Jaton, Phillippe Hauser and Jaques Billes, Institute of Microbiology, University Hospital of Lausanne, Lausanne, Switzerland; Birgit Willinger, Division of Clinical Microbiology, Medical University of Vienna, Vienna, Austria; C. Orla Morrissey, Alfred Health and Monash University, Melbourne, Australia; Sarah Kidd, National Mycology Reference Centre, Microbiology and Infectious Diseases, SA Pathology, Adelaide, Australia.
Declaration of interest
AC, MC, MCE, CH, RG, FH, GJ, KJ, SK, CL, ML, LM, CM, COM,
COrM, WM, YDO MP, MR, BW, KHW: No conflicts delcared AFT has received funding from Pfizer. KL has received a research grant from MSD, received travel support from MSD, Pfizer and Gilead and received lecture honoraria from Gilead, MSD, and Pfizer. PLW is a founding member of the EAPCRI, received project funding from Myconostica, Luminex, and Renishaw diagnostics, was sponsored by Myconostica, MSD and Gilead Sciences to attend international meetings, on a speaker’s bureau for Gilead Sciences, and provided consultancy for Renishaw Diagnostics Limited. RAB is a founding member of the EAPCRI, received an educational grant and scientific fellowship award from Gilead Sciences and Pfizer, is a member of the advisory board and speaker bureau for Gilead Sciences, MSD, Astellas, and Pfizer, and was sponsored by Gilead Sciences and Pfizer to attend international meetings. JL is a founding member of the EAPCRI, received an educational grant and scientific fellowship award from Pfizer, and was sponsored by Astellas to attend international meetings. JPD is a founding member of the EAPCRI, has received grants from Astellas, Gliead, MSD, and Pfizer, provided consultancy for Gilead, Pfizer and Viamet and received honoraria for lectures from Gilead, MSD, Pfizer and Basilea. SB is a founding member of the EAPCRI, received project funding from Renishaw diagnostics. Is a member of Gilead speaker’s bureau, has received honorarium for educational programs from Astellas, for congress symposium from Gilead and Bio-Rad, and received travel grants from Pfizer. PH research funded by university grant RVO: 61989592 CLF has received grant support from Austrian Science Fund, MFF Tirol, Astellas Pharma, Gilead Sciences, Pfizer, Schering Plough and MSD. She has been an advisor/consultant to Gilead Sciences, MSD, Pfizer, Schering Plough and Basilea. She has been paid for talks by and received travel/accommodations/meeting expenses from Gilead Sciences, MSD, Pfizer, Astellas Pharma, Schering Plough and Basilea. LK grant from Gilead and has been an adviser to Astellas, Gilead, Schering-Plough, has received research grants from Gilead, Merck, Sharpe & Dohme, Schering-Plough and has received honoraria for educational lectures from Gilead, Pfizer, Merck, Sharpe & Dohme, Schering-Plough and Janssen.
Article Images (0)
×![]()
Bioz is the world’s most comprehensive AI search engine for scientific experimentation. The Bioz search engine offers researchers billions of data-driven product, technique, and protocol recommendations. Bioz taps into the latest advances in Artificial Intelligence (AI) – including Natural Language Processing (NLP), Machine Learning (ML) and Generative AI to mine and structure hundreds of millions of pages of complex and unstructured scientific papers. By harnessing the power of AI, Bioz provides researchers with an unprecedented amount of summarized scientific experimentation knowledge right at their fingertips, ultimately helping to speed up drug discovery and the rate of success in finding cures for diseases. Bioz is used by over 15 Million researchers from over 17,000 different universities and companies in 196 countries.
