Towards a molecular diagnosis of invasive aspergillosis and disseminated candidosis Free

Abstract

A lot of in-house polymerase chain reaction assays have been reported for diagnosis of invasive aspergillosis and disseminated candidosis. Encouraging results have been published to anticipate the diagnosis over the conventional microbiological methods. However, the absence of standardized methods has led to diverging results. As a consequence, these tests are not recognized as consensual diagnostic criteria, in contrast with some antigenemia detection kits. The major breakthrough for improving the results of these methods is the emergence of real-time technologies. This markedly improves the reliability of the PCR results by dramatically decreasing the risk of false positive results due to PCR products carryover. Moreover, using the quantitative results provided by this technique, this allows to rapidly compare the efficiency of primers, probes, and DNA extraction methods. Therefore, the hope is to identify the more specific and sensitive parameters to implement comparative studies. Automated DNA extraction should also be useful to achieve this goal.

Whatever sophisticated technology is used, we still have to define the meaning of detecting nucleic acids in a given clinical sample. This seems simple in normally sterile anatomical sites but less obvious for example in respiratory specimens for invasive aspergillosis or in blood for candidosis in heavily colonized patients. Additional studies of the kinetics of fungal DNA are needed. The development of real-time technology should improve our knowledge in order to give the clinicians informative clues for making a decision.

1 Introduction

The use of polymerase chain reaction has increased to the point where it is now accepted as the standard method for detecting nucleic acids from a number of microorganisms in clinical samples. However, though it is accepted for some viruses and bacteria, it has not reached the same acceptance in mycology, more specifically for the two main fungal infections, which are invasive aspergillosis and disseminated candidosis. Several conventional PCR assays have been reported and shown to be interesting in the hands of their authors. However, all publications dealing with in-house PCR assays and their results are very diverse. As a consequence, PCR is not included in the diagnostic criteria of the international consensus on the definitions of opportunistic invasive fungal infections in immunocompromised patients with cancer and hematopoietic stem cell transplants [1].

The main reason for the non-acceptance of PCR as a diagnostic tool for fungal infections is the absence of a consensual technique. The sensitivity and the specificity of a PCR assay is very dependent on every step of the amplification, including the type and the preparation of the sample, the DNA extraction method, the choice of the DNA target and the primers, the use of a hot-start method and of an enzymatic prevention of contamination, and the means used to check the specificity of the amplified products [2,3]. In the absence of a commercial kit, there is little hope that a consensual technique will be accepted and evaluated with patients at risk of invasive fungal infections in random and prospective studies in different centers.

Hope is coming from another important step in the evolution of molecular diagnostics: the advent of real-time detection of PCR products [4]. The development of fully automated platforms, that comprise nucleic acids extraction and product detection through a series of linked instruments, will have a tremendous impact on the implementation of molecular testing in the specialized laboratory, and as a consequence, within the routine clinical diagnostic laboratory. The absence of post-PCR processing after amplification has many advantages over the use of acrylamid or agarose gel electrophoresis, with or without Southern-blot hybridization, or enzyme-linked immunoassays to check the specificity of PCR products. With real-time PCR, the only source of contamination after sample preparation is pipetting the prepared specimen into the reaction mixture and loading the instrument. Therefore, the main cause of false positive reactions, i.e., the previously amplified products potentially aerolized in the environment, is deleted. This increases the reliability of the results and therefore the trust of the clinicians. Moreover, all currently available real-time PCRs can give the results in less than 2 h, a requirement for clinical decision making.

We shall now focus on real-time PCR assays developed for the diagnosis of invasive aspergillosis and disseminated candidosis. Reviews are available for the ins and outs of so called conventional PCR in mycology [2,3]. Real-time PCR assays also have numerous other applications, such as identification, strains typing, or detection of mutations responsible for drug resistance in microorganisms. However, for these applications the DNA source is the colonies obtained after culture. The technical issues are therefore slightly different from those encountered in the positive diagnosis where the main question is how to detect the minute fungal DNA among the huge quantity of irrelevant human DNA. To control contamination for the false positive results and to control the yield of the amplification reaction to exclude false negative results is then crucial for the reliability of the technique.

Real-time PCR could bring information to the clinicians but such claims can only be taken seriously when each newly described assay is suitably compared to its characterized predecessors with all their pros and cons for each specific infectious disease. This is particularly true for fungal diseases the diagnosis of which is often only probable and rarely definite. Fungal infections are usually opportunistic and this means that the presence of fungal DNA is not always synonymous with infection.

2 Real-time technology

The possibility that, contrary to conventional PCR, the detection of amplicon could be visualized as the amplification progresses is the base of real-time PCR. All the techniques record fluorescence in real-time as the PCR samples pass photodetection diodes within the instrument. This provides insight into the kinetics of the PCR. The role of each parameter of the reaction, from the nucleic acid extraction method to the last cycle through the hybridization of the primers and probes, can be evaluated by analyzing the amplification curves.

The monitoring of accumulating amplicon in real-time has been made possible by the labeling of primers, oligonucleotide probes or the amplicon itself with molecules capable of fluorescing. These labels produce a change in signal following direct interaction with the amplicon or hybridization to the amplicon. The signal increases as the amount of amplicon increases after each amplification cycle. The fluorescent signal can be obtained by one of several methods. The simplest method employs the SYBR Green dye that increases in fluorescenœ when bound to double strand DNA allowing quantification. The analysis of the melting curve is indicative of the nature of the amplified fragment. However, for diagnostic purposes, it is of the utmost importance to check the specificity of the amplified products. This can be achieved by using specific hybridization probes.

There is a range of chemistries currently in use. One relies upon fluorescence resonance energy transfer (FRET) between fluorogenic probes and has been developed for the LightCycler instrument (Roche Diagnostics). The Applied Biosystems apparatus uses the TaqMan technology based on the 5′ nuclease activity of the polymerase to release the two dyes of the fluorogenic probe. The choice between these two more popular methods depends on specific demands and/or local opportunities. However, other chemistries are in development such as molecular beacons [5] and scorpion probes [6]. Moreover, besides the most common oligoprobes based on acid nucleic, other methods are being developed using DNA analogues such as peptide nucleic acids (PNA) [7] or locked nucleic acids (LNA) [8]. All these technologic developments will bring opportunities for designing new probes.

As there is still a risk of breaking the tubes or capillaries when handling after amplification, and therefore a risk of contaminating the environment, an enzymatic prevention based on the substitution of dTTP for dUTP in the reagent mix and the systematic use of the enzyme uracyl-DNA-glycosylase (UDG) can be added to real-time PCR [9]. This enzymatic prevention is easy to implement in routine diagnosis and is systematically incorporated in conventional commercial kits (Amplicor®, Roche).

False negative results, due to PCR inhibitors in the clinical samples, should also be taken into account in the procedure. This can be achieved by co-amplifying an internal control. Such a control can be added before DNA extraction or before amplification and is called exogenous, as it is not present in the acid nucleic preparation. An endogenous control, such as housekeeping genes, is useful for quantification of gene expression but not for DNA detection. The internal control should ideally hybridize to the same primers, have an identical amplification efficiency, and contain a discriminating feature such as a change in its length or in its sequence to be differentiated as is already done for aspergillus [10]. However, this needs to be developed for each specific assay. Internal control that binds different primers from the target DNA, such as an internal control based on the detection of 10 pg mouse DNA added to the sample [11], has the advantage of being useable for several PCR assays. It is now possible to read quantitative results simultaneously with the new available instruments and this should make the use of internal control more popular.

Amplifying human DNA from a clinical sample known to contain huge quantities of human DNA cannot be used as an internal control. Indeed, if the efficiency of the reaction is too low to correctly amplify minute microorganism DNA, this can be efficient enough to amplify human DNA. This could lead to a false reassurance on the amplification reaction and therefore to false negative results. Moreover, the quantity of human DNA in a clinical sample is not standard, and huge differences are expected between a neutropenic patient and a healthy individual. Nevertheless, from our experience, the present commercial DNA extraction kits are efficient enough to obtain clean DNA from serum and blood if the starting volumes are respected according to the manufacturers' recommendations, and we do exceptionally observe residual PCR inhibitors.

Although the PCR technology is the more popular one, there are alternatives. The isothermal nucleic acid sequence-based amplification (NASBA) relies on the concerted action of three enzymes (reverse transcriptase, RNAse-H and T7-RNA polymerase). The amplification reaction takes place at 41 °C and generates single stranded RNA molecules as the end product that can be directly targeted with probes. This technology starts being used in mycology [12,13]. The above mentioned molecular beacons are the probes used for real-time detection with NASBA. They are short nucleotide sequences consisting of a stem–loop structure containing a sequence complementary to the target sequence. The close proximity of the fluorophore and the quencher prevent fluorescence emission. When the beacon hybridizes to the target, it undergoes a conformational change. As a result, the quencher can no longer absorb photons from the fluorophore. Molecular beacons can be labeled with different fluorescent molecules, thus allowing co-amplification and simultaneous detection of different amplified products in a single tube. This is the base of a commercialized assay for HIV detection [14].

3 Automated DNA extraction

Once reliability and fast turn out are achieved with real-time techniques, the remaining limiting step is nucleic acids extraction. There is a need for a rapid, standard method for DNA extraction to decrease the work burden and to improve comparisons between laboratories. Fully automated nucleic acid extraction techniques are now possible with affordable equipment, thus eliminating the need for vacuum pumps, centrifugation, or other manual steps that may result in cross contamination. Different technologies using silica particles capture are available, either with magnetic beads or filtration, which are worth comparative studies. First trials in mycology have been reported [15,16].

4 Invasive aspergillosis

As stressed in the introduction, to evaluate a new diagnostic tool, it must be compared with previously characterized methods. The traditional diagnostics include microscopy, culture on agar media, antigenemia, and search for antibodies. All of them are limited either by poor sensitivity, narrow detection window, complex interpretation, high levels of non-specific reactions, or immunosuppression. Culture of bronchoalveolar lavage (BAL) fluids is positive in only 50% of invasive aspergillosis proven at autopsies. The assays for anti-aspergillus antibodies are not standardized and questionable in deeply immunocompromised patients such as those at risk for invasive aspergillosis. Only the galactomannan test has reached a satisfactory level of confidence to be included in the diagnostic criteria of invasive aspergillosis in hematology despite a high level of false positive results, around 10–15% [1]. Therefore, the only criterion for proven infection is the observation of hyphae in biopsies, rarely done in patients at risk for invasive aspergillosis due to the risk of bleeding. As a consequence, the incidence of invasive aspergillosis is difficult to assess and at least 30% of cases remain undiagnosed whereas this infection is increasingly reported in autopsies series [17]. This lack of standard diagnostic means must be kept in mind when evaluating a new diagnostic tool such as PCR assays.

In the past decade, numerous in-house PCR assays have been reported for the detection of aspergillus DNA. At the beginning, PCR assays were aimed at detecting aspergillus DNA in BAL fluids to improve the sensitivity of culture. With few exceptions, the rate of positivity was high, up to 35% of the samples [2]. The probable explanation was that the inhaled aspergillus spores, or conidia, ubiquitous in the air, could lead to positive reactions. The conclusion was that PCR could not distinguish between contamination by airborne conidia, surinfection of upper respiratory tract or true invasive infection. Some authors have even underlined the potential value of a negative result to exclude the diagnosis [18]. These conclusions could be reassessed with the real-time PCR assays. Indeed, one can take advantage of the quantitative results to estimate thresholds of positivity [19]. However, invasive aspergillosis is rarely disseminated in the whole lung but more frequently localized. The yield of the BAL is very depending on the quality of the endoscopy. If the lavage does not reach the lesions, the result could be falsely negative and quantification of dubious relevance.

Given the high rate of positivity and the difficulties in interpreting a PCR positive result in bronchial specimens, several teams have switched to blood samples. The small risk of contamination of blood with airborne conidia should improve the specificity of a PCR positive result. However, since blood cultures are exceptionally positive for Aspergillus spp. even during proven invasive aspergillosis, the likelihood to obtain PCR positive results is low if the amplified DNA came from living microorganisms. The good news has been that it was possible to amplify aspergillus DNA from blood or serum samples and that this positivity could be correlated with infection [20,21]. However, huge discrepancies between the studies have been reported which can be explained by differences in the clinical specimen, the target DNA and the PCR techniques [2].

To explain the discrepancies between the reported results, another crucial point with the amplification of fungal DNA must be considered. As for other microorganisms of the environment, fungi can contaminate every place, including laboratories, pipettes, tubes, water, etc. Contaminations of reagents, including those used in molecular biology, have been reported [22–24]. More specific measures to control this problem should thus be undertaken such as UV irradiation of the PCR reagents [16] and multiplication of the negative controls.

If the real-time PCR assays eliminate the false positives due to amplicons, they do not eliminate the contamination of reagents or buffers. Therefore, several points must be improved. The first one is in the use of automated DNA extraction with as few reagents as possible. If the amplified DNA comes from circulating DNA, as shown by amplification from serum samples [21,25], the existing protocols with automated DNA extraction should be efficient and sufficient [15,16]. If the hypothesis is that the amplified DNA comes from the circulating conidia or hyphae, other stringent protocols to lyses the fungal wall must be considered with the risk of using enzymes contaminated with fungal DNA. The second point is to play on primer specificity. The more specific to Aspergillus spp. the primers are, the lower is the risk of amplifying DNA of other species. The strategy using panfungal primers should be balanced with this risk. That is why the use of probes to check the specificity of the PCR products is mandatory. The simple amplification with SYBR green dye and the analysis of the melting curve are not sufficient, considering the lack of knowledge concerning the melting curves of all putative contaminant DNA. On the other hand, the primers and the probes must hybridize to DNA of at least the three or four most frequent Aspergillus species involved in invasive aspergillosis and not only A. fumigatus[26] to be relevant in clinical practice.

The real-time PCR assays published to date are all different on a lot of their features (Table 1). However, the real-time PCR presents the advantage of rapidly answering the question of the more sensitive or reliable features. For instance, we have shown that with blood or serum samples artificially seeded with A. fumigatus DNA, the yield was equivalent and superior to plasma [27]. Similar experiments can be performed with every parameter of the PCR reaction and comparing the yield with the amplification of a single copy gene for instance [27]. We can expect more consensual techniques in a near future.

Whatever the reliability of more standardized real-time PCR assays, it seems likely that an isolated positive result will be difficult to interpret. As for antigenemia tests, serial samples will be necessary. Real-time PCR tests could then be performed in parallel with antigenemia testing. Although several publications underline the higher sensitivity of the galactomannan test over PCR assays both in patients [21,28] and in animal models [29], the association of these two tests targeted at two completely different molecules should be complementary. Moreover, as for antigenemia test, quantitative results can be obtained. However, until now, we have not observed a parallel increase in galactomannan titers and in the amount of aspergillus DNA in serum [16], in contrast with what we have observed for other microorganisms [30].

Another interesting field of investigation is the development of molecular tools for identification of filamentous fungi in biopsy specimens. Indeed, because the histological appearances of certain fungi can be non-specific, definitive diagnosis typically requires culturing the fungus. Unfortunately, the culture is often not performed or negative. The identification of the fungus can be achieved either using real-time PCR tests [31,32] or using in situ hybridization [33]. In clinical practice, the DNA extraction has often to be made from paraffin-embedded tissue specimens and the quality of the DNA extracted must be checked before giving a negative result.

5 Invasive candidosis

Most of the questions on PCR assays for invasive aspergillosis can be raised for disseminated candidosis. The aim of several reports using PCR assays was to demonstrate the superiority of PCR assays over blood cultures. The latter are known to be poorly sensitive with less than 50% of disseminated candidosis with positive blood cultures. The reasons for this low sensitivity are not due to the poor performances of the current blood culture techniques. The more probable explanation is the low rate of circulating yeasts, with a low probability to catch them during blood sampling. Therefore, to amplify candida DNA from blood raises the question where is the DNA coming from: living Candida or circulating DNA?

If the aim of PCR assays is to improve the detection of living Candida, this should be put in parallel with the current yield of blood cultures. The first publications reported sensitivities from 1 CFU ml⁻¹[20] to 100–150 CFU ml⁻¹[34]. With the current DNA extraction kits, the initial volume is usually 200 µl. To be sure to have one microorganism in a given sample in final dilutions, one must use dilutions with a minimum of 10 microorganisms. As a consequence, to obtain a constant PCR positive result, 10 Candida must be present in the 200 µl tested. This threshold corresponds to 50 Candida ml⁻¹. In these conditions, a blood culture performed with 10 ml of blood will be probably consistently positive unless the Candida do not grow because of antifungal treatments [35]. Currently, the work burden for blood culture is far less heavy than for real-time PCR.

If the amplified DNA comes from circulating DNA as suggested by the better yield reported for serum than for blood [36–38], the sensitivity using multicopy genes as a target can obviously reach higher sensitivity than 50 Candida ml⁻¹, probably below one candida genome ml⁻¹. However, the meaning of such a finding is more complex than a positive blood culture, especially in heavily colonized patients, as seen in ICUs, or patient under antifungal therapy.

Another potential limit is the need to detect several Candida species, at least the five or six main species encountered in blood cultures [39]. For aspergillus, the problem is to detect several species without the need to identify the species, as the treatment is similar for each species, with very few exceptions such as A. terreus. But the issue is different for Candida. Indeed, we must distinguish between most of the Candida species, sensitive to fluconazole, and C. glabrata and C. krusei, the sensitivity of whose is variable or null. To detect yeasts without knowing the involved species will not change the use of a wide spectrum antifungal until yeast identification. Therefore, to overpass the traditional culture on agar media, a real-time PCR assay should enable to detect and to identify every Candida species and putative mixtures of these species. This requires the new generation of real-time PCR apparatus with the possibility to simultaneously use more than two fluorogenic probes. Another option is to associate PCR assays with a DNA chips technology to characterize the amplified DNA.

As for aspergillus, the use of real-time PCR should improve the reproducibility of the PCR tests to compare the results of several teams. The current assays are different but should be rapidly more homogenous [40–42]. However, probably because of the easiness of blood culture and the non-ambiguity of a positive result, there is less development for Candida than for Aspergillus. In parallel with the diagnosis, real-time PCR assays have been developed for identification [43] or for experimental models [44]. The combined use of several PCR assays for detecting several species in polymicrobial clinical samples such as oral rinse cultures should be evaluated in comparison with chromogenic agar media which can detect mixtures of several species [45]. As underlined above for blood culture, to detect DNA would not necessary require an antifungal treatment.

6 Conclusion

To improve the reliability of the current PCR assays, the most significant changes expected in the coming years is the introduction of real-time PCR and automated DNA extraction. Because detection of PCR products and data analysis are part of real-time instruments, there is no post-amplification handling. Thus, the risk of contamination of the environment with amplicons is sharply reduced. Therefore, the likelihood of false positive results is low and the PCR assays can be routinely implemented. This can also be applied to defining the amount of target DNA and therefore following the effects of therapeutics on the microorganism loads. Interestingly, this also allows the rapid and reliable comparison of the yield of each parameter of the PCR reaction, from the DNA extraction to the primers and the probes. The most efficient parameter can be retained, leading to more homogenous PCR assays and, as a result, to a better comparison between the published reports. This opens new opportunities for well-designed prospective studies in mycology with regular quality controls.

Future development of novel chemistries and improved real-time apparatus should promote the use of multiplex real-time PCR assays. A combination with microarrays technology may also enhance the rapidity of identification. However, the development of new sophisticated tools should not mask our ignorance about the kinetics of fungal DNA in an infected patient, and therefore the meaning of a positive PCR test. Both for Aspergillus and Candida, the combined evaluation of DNA detection by real-time PCR, RNA detection by real-time PCR or NASBA technique, and antigen detection in patients with well-defined diseases requires investigation to improve our knowledge in this field.

References