1. Introduction
Quantitative PCR (qPCR) is an adaptation of standard PCR permitting the detection and real-time quantification of specific target amplification products. A target DNA sequence is selectively amplified using sequence-specific primers, a reporter and quencher labeled dual fluorochrome, an oligonucleotide hybridization probe and a Taq DNA polymerase enzyme. During amplification, the probe specifically hybridizes the accumulating product, and the endonuclease activity of the Taq DNA polymerase cleaves reporter-labeled nucleotides resulting in detectable fluorescence. Reactions are characterized by the time duration during the standard 40 cycles of qPCR amplification where a threshold of baseline fluorescence (C
qs) is exceeded. TaqMan qPCR is established as a useful method for the detection and identification of
Aspergillus species in clinical samples including the airway [
1,
2,
3,
4,
5]. Using these and other next-generation sequencing approaches, our group has demonstrated high levels of airway
Aspergillus in patients with bronchiectasis where higher qPCR-derived
A. fumigatus and
A. terreus is associated with poorer clinical outcome [
1,
3,
6]. Importantly, however, to determine absolute quantification of
Aspergillus 18S rRNA, a serial dilution of
plasmid containing 18S DNA is necessary for the generation of a standard curve on each plate, a time consuming and costly process limiting the specimens that can be studied. In addition, optimization of the employed standard curves is required, which in itself, demonstrates dynamic and differing ranges for the absolute quantification of
Aspergillus species [
7]. The results of even standard and test specimens may vary based on reaction efficiencies and differences in specimen content including the presence of inhibitors [
8,
9]. For all these reasons, an improved and alternative method may be beneficial.
Recently, droplet digital PCR (ddPCR) has been developed and could potentially circumvent issues associated with qPCR [
10,
11,
12]. This technique, based on partitioning the PCR reaction mix into a thousand-fold magnitudes smaller and segregated reaction droplets allows amplification of the respective target(s) within each individual droplet which is then quantified by a target-dependent fluorescence signal (
Figure 1). The digital aspect of this approach relies on distributing the target gene into a significant number of partitions (or droplets) such that each receives a number of genes (i.e., 0, 1, 2, etc.). Performing PCR on such partitions results in the amplification being labeled positive (in those containing the target) or negative (no amplification). As positive readouts potentially contain more than a single gene copy of the target molecule, a simple summing of the number of positives will not yield the correct number of target molecules that may be present. Therefore, Poisson statistics are applied in ddPCR to estimate the total number of target molecules present within an interrogated specimen and avoids the need for reference to a standard curve [
10,
11,
12]. As ddPCR represents an end-point PCR reaction, data are unaffected by variations to reaction efficiency and the absolute copy number of the target genes can be determined with confidence so long as the fluorescence readout is correctly partitioned to positive and negative droplets. The high precision and accuracy of ddPCR further reduces the need for technical replicates which improves experimental throughput, saves time, and effectively permits accurate quantification of targets in low volume human specimens such as that from the airway [
10,
11,
12].
Table 1 summarizes the comparisons between qPCR and ddPCR.
In this study, we evaluate the accuracy and sensitivity of ddPCR (as compared to qPCR) in detecting Aspergillus species in the normal and diseased airway from respiratory specimens. To allow concurrent detection of A. fumigatus and A. terreus in our samples, we developed duplex primers and probes that allow interrogation of both species by ddPCR.
3. Discussion
The study of airway fungal communities, including next-generation sequencing of the pulmonary mycobiome, is gaining clinical and academic interest, particularly in the setting of chronic respiratory disease states such as asthma, COPD, cystic fibrosis and bronchiectasis [
1,
6,
16,
17,
18,
19,
20,
21,
22,
23]. Fungi, even in diseased states where dysbiosis is prevalent, demonstrate relatively low abundance at most body sites including the gastrointestinal and respiratory tracts where fungi typically make up less than 5% of the resident microbial community [
1,
3,
24,
25,
26,
27,
28,
29,
30,
31]. This low fungal abundance makes 18S ITS targeted amplicon sequencing and qPCR methodological approaches attractive, however, identifying and quantifying fungal load in this manner imposes a lower limit of detection, problematic in cases of ultra-low fungal burdens, and also makes accurate and reliable quantification challenging in the clinical setting. Here, for the first time, we evaluated ddPCR as an alternative to standard qPCR for the detection of
Aspergillus species from the human airway. While both methods can detect and quantify
Aspergillus species reliably, ddPCR demonstrates greater sensitivity for
A. terreus detection and is more resistant to PCR inhibition making it an attractive alternative for the detection of microbes, such as fungi, that occur in low abundance in respiratory specimens but have clinical relevance.
The clinical relevance of low abundance microbes in the airway is well demonstrated in chronic respiratory disease states complicated by the infection such as cystic fibrosis (CF) and bronchiectasis [
20,
32,
33,
34,
35]. In these settings, the acquisition of a new organism or confirmation that an eradication regime has been effective are clinically critical features for which standard microbiological approaches are lacking. While standard qPCR is useful, it lacks sensitivity below its lower limit of detectable thresholds for individual organisms. In CF and bronchiectasis, the airway ecology is a complex milieu of multiple co-existing organisms from a variety of kingdoms, and hence the ability to detect organisms at ultra-low abundance has value. A clinical example is the identification of allergic bronchopulmonary aspergillosis (ABPA) in CF where likely ultra-low amounts of fungi exist in the airway but are currently either missed or undetected through standard approaches. ddPCR may, therefore, offer an attractive alternative to qPCR in selected clinical settings [
20,
33,
34,
35,
36,
37,
38].
One key advantage of ddPCR over traditional TaqMan qPCR is the direct quantification of a target microbe without the need for a standard curve or controls. This leads to improved reproducibility and better accuracy by eliminating reliance on quantitative reference materials whose quantification, source, batch, storage and handling conditions can all influence qPCR results for biological specimens [
39]. “Known” quantities of the target, used in TaqMan qPCR standards, as measured by UV spectrophotometry using Nanodrop, employs an indirect quantification approach which potentially affects reliability and accuracy because of quantification assumption uncertainty (i.e., gene copy number per cell and conversions of measured absorption to copy numbers, etc.) [
40]. Spectrophotometry quantifies all nucleic acids that absorb at 260 nm which include quantifying DNA and RNA impurities within the standard itself which, when used to quantify the target in samples of unknown quantity, may lead inadvertently to an overestimation.
Our detected higher precision for quantifying a target of ultra-low abundances such as
A. terreus and the higher run-to-run reproducibility observed with ddPCR is consistent with the binary nature of digital PCR quantification and the findings of others [
10,
39,
40,
41,
42]. ddPCR quantifies by counting frequencies of positive endpoint PCRs based on a Poisson distribution, whose quantification is not dependent on variability in PCR amplification efficiency, an issue with TaqMan qPCR. Substrate competition and PCR amplification efficiency are likely explanations for the observed variability between qPCR and ddPCR results for
A. fumigatus and
A. terreus, respectively, where no superiority of one technique was present for
A. fumigatus but the latter (ddPCR) better for
A. terreus. A combined effect of the target number, abundance and amplicon length on reaction mix consumption is also greater for multiplex compared to single-plex reactions. Therefore, if any reaction component is limiting, multiplex reactions can show either significant Cqs delays or even total loss of PCR products particularly for targets of lowest abundance. The improved precision attained through ddPCR is a key consideration for respiratory specimens of ultra-low concentrations or where the target is undetectable by TaqMan qPCR.
Our detected tolerance to PCR inhibitors with ddPCR is consistent with its binary nature [
43]. PCR inhibitors function either through DNA sequestration or by reducing PCR amplification efficiency, both of which increase C
qs values and lead to underestimation with a TaqMan qPCR approach. Despite reduced amplification (i.e., lower fluorescence intensity due to PCR inhibition), quantification by ddPCR depends on end-point droplet fluorescence which remains higher than the background fluorescence readings. Biological samples including respiratory specimens are often complex and contain PCR inhibitors; therefore, the robust and resistant nature of ddPCR against such inhibitors may be useful in the clinical setting alleviating cost and inconsistent recovery efficiency associated with DNA purification procedures. ddPCR, however, is not fully immune to PCR inhibition, and where severe inhibition exists, ddPCR can experience “total molecular drop-out” where the target remains unamplified [
44].
ddPCR permits multiplexing, an additional advantage where clinical material is scarce.
The majority of available Aspergillus-related diagnostic kits do not specifically detect A. fumigatus and A. terreus in biological samples and usually consist of primers and probes that employ a pan- Aspergillus approach which cross-react across several Aspergillus species. While duplexing is more challenging in a TaqMan qPCR approach, largely due to substrate competition and the need for internal positive controls, simultaneous measurements of A. fumigatus and A. terreus are possible with ddPCR due to the individual amplification approach of the generated droplets. ddPCR duplexing while reducing labor, improving logistic arrangements and optimizing data quality through limiting accumulated pipetting errors and the lack of a need for standards does come with the added costs of reagents and consumables unique to ddPCR.
One key drawback, however, of ddPCR is its poor performance with high abundance samples containing >10
5 gene copy numbers. This is due to the partitioning aspect of this technology, the number of droplets generated and the Poisson distribution algorithm employed for the accurate determination of the number of DNA copies per sample [
11]. Therefore, in such circumstances, sample dilution may be required, which poses an additional experimental variability, or one can simply revert to TaqMan qPCR. Sample processing, droplet generation, thermal cycling and droplet analysis further adds additional processing time (~2 h) to the overall process compared to qPCR. The balance between achieving good sensitivity and high accuracy for any test is critically important. Sensitivity is the proportion of true positives that are correctly identified by a diagnostic test while accuracy is the overall proportion of true results, whether true positive or true negative. In our work, sensitivity was high for both qPCR and ddPCR approaches but highest using the latter. Accuracy also remained high for both the healthy and diseased specimens tested in this work however again was highest using the ddPCR approach. Therefore, ddPCR appears to be both highly sensitive and accurate, likely explained by the reaction being performed at the individual droplet level as compared to qPCR where competition for the substrate reagents occurs particularly in duplex assays and these reactions remain subject to PCR inhibition.
While several groups internationally, including ours, have developed qPCR protocols to detect
Aspergillus species in biological samples, emerging technologies such as ddPCR offer an alternate, highly sensitive and accurate quantification for samples with ultra-low microbial (fungal) burdens such as that of the airway where detection, even in small amounts has an important clinical consequence and may be potentially missed by qPCR [
1,
2,
3,
4,
5,
7,
45]. Attributes such as that offered by ddPCR may be useful in the current era, where microbiomes including the fungal mycobiome, are taking on greater importance and relevance in understanding pathogenesis, disease course and their consequence across a range of respiratory disease states [
1,
3,
27,
30,
41,
42,
46,
47]. Going beyond fungi, the promise of ddPCR will likely extend across a wide range of microorganisms, organ systems and human disease if it is appropriately applied to the right specimen, in the right setting and to resolve a specific clinical question.