European Science Foundation

Report on a Workshop on Aspergillus and Aspergillosis (Diagnosis)

16^th - 17 ^th September, 1999

David W Denning : University of Manchester

Jean-Paul Latgé : Institut Pasteur

Correspondence to: David W Denning

ESF Workshop on Aspergillus and Aspergillosis (Diagnosis)

The European Science Foundation awarded a grant to hold a workshop on Aspergillus and aspergillosis which was held in September, 1999 in Manchester. This is the report from that workshop.

Objectives

• To overview the ESF network grant on Aspergillus and aspergillosis.
• To establish new collaborations.
• To finalise the scientific approaches behind the diagnostic component of the GARDIFI application to FP5

Program

Objectives - finalise pre-circulated protocols:

For PCR group
a) which sample type

1. which sample preparation
2. an agreed comparative primer pair standard
3. Real time PCR method
4. QC
5. Interlaboratory comparisons
   
   

For serology Group
a) Frequency of sampling in different patient populations

(PCR/Serology)

b) Handling and Storage of samples

1. Assaying samples - which, when, where.
2. QC
3. Inter-laboratory comparisons
4. Inter-test comparisons

Introduction

Dr Denning welcomed the participants. The purpose of the 3 year network on Aspergillus and aspergillosis was outlined by Dr Denning.

There are three components to the network:

1. Diagnostics
2. Resistance
3. Genomic sequencing and functional genomics

GARDIFI program

Dr Denning then went on to outline the scientific and organisational elements of the GARDIFI program (Genetic Risk, Antifungal Resistance and Diagnosis of Invasive Fungal Infections) which is an application being made to the European Union Framework V Program under Key Action Infectious Diseases. This application would combine the patient resources of 32 hospitals, dealing with solid organ and stem cell transplant patients, patients with acute leukaemia and those in intensive care units. There would be 14 susceptibility testing laboratories affiliated, covering 16 European countries, 6 PCR diagnostic laboratories and 2 Central serology laboratories. The application was due in mid-November and part of the purpose of the ESF meeting on Diagnostics was to develop a consensus on questions and means of addressing questions on the diagnosis of aspergillosis (and invasive candidiasis). The outcome of the application would be known in March/April and the program would start in July 2000.

Criteria diagnosis of invasive fungal infections

Dr Meis outlined the new criteria that had been developed jointly by the EORTC invasive fungal infections co-operative group and the NIAID Mycoses study group. The criteria (which are available on the Aspergillus website) separate proven infections from probable, and probable from possible. Proven infection are split into those caused by moulds and those caused by yeasts and if the species is known then that should be appended. In addition there were categories for bloodstream infection and endemic mycoses. Probable infections would require a component of host risk, microbiological criteria and clinical/radiological criteria for the diagnosis to be established. With respect to probable cases and the criteria were designed for cancer and bone marrow transplant patients, and were not intended for AIDS, solid organ transplants or other types of patients. Dr Meis pointed out that PCR was considered as a possible microbiological criterion, but rejected because of the lack of evidence to support its utility other than in some bone marrow transplant patients in a limited number of centres. The possible category was not intended to be used for clinical studies, but rather for epidemiological surveys or empiric therapy studies.

Decision analysis

Dr Verweij stood in for Hans Severens who was unable to join the meeting. He discussed the principles of decision analysis and the diagnostic strategy decision analytic model that can be used to evaluate the cost effectiveness of different diagnostic strategies. In addition the number of diagnostic tests that are used to reach a final diagnosis could be minimised for some groups of patients so that rather than doing CT scans, bronchoscopy, antigen, antibody acute phase reactants, PCR and lung biopsy on all patients, a sequential approach to decision making can be adopted. One of the key points in comparing decision models is establishing the treatment threshold when given a set of clinical radiological and laboratory results. The key final objective of decision analysis would be to ascertain whether treatment can be safely withheld or stopped after negative results are received in at risk and ill patients. Finally different diagnostic approaches may lead to different patient outcomes, which of course is critically important.

Dr Paul Verweij illustrated his talk with studies that are published on urinary tract infections and on different diagnostic strategies for managing invasive aspergillosis.

Aspergillus antigen detection

Dr Verweij outlined the present position with respect to galactomannan in blood as a marker for the diagnosis of invasive aspergillosis. He referred to a recent piece of work conducted in Leuven in Belgium by Johannes Maertens on two hundred leukaemic patients with persistent fever, 75 who died and had an autopsy. The patients were sequentially sampled for Aspergillus galactomannan detected by Sanofi's Platelia ELISA (Maertens J et al. J Clin Microbiol 1999;37:3223-3228). 27 patients had invasive aspergillosis confirmed at autopsy and 25 of these had a positive serum galactomannan. 44 patients did not have invasive aspergillosis at autopsy and 42 had a negative serum galactomannan, ELISA. This yielded a sensitivity of 92.6% and a specificity of 95.4%. In addition the galactomannan was positive before the first clinical sign of invasive aspergillosis in 23 patients, before the first chest X-ray abnormality in 28 patients, before the onset of fever in 27 patients, before treatment was commenced in 31 patients and before the first culture was positive in 19 patients. This confirmed and expanded previous experience showing that the use of the Aspergillus galactomannan test diagnosed aspergillosis between 6 and 13 days earlier than other manifestations of disease. The false positives probably relate to absorption of galactomannan from food through a damaged gut. It was pointed out by Latgé who has actually determined this experimentally using gas liquid chromatography coupled to mass-spectrometry techniques on serum samples from false positive patients ( e.g.no clinical IA) that this is genuine galactomannan, not a related substance. Therefore these may be true positive results, but they do not indicate invasive aspergillosis.

Dr Verweij then went on to discuss the use of bronchoalveolavage fluid antigen testing as performed in his own centre in 8 patients with proven invasive aspergillosis. 7 had a positive PCR on BAL and all 8 had a positive ELISA test. In those with possible invasive aspergillosis (12) 41% had a positive PCR and 25 had a positive ELISA on BAL fluid. In 44 patients who did not have invasive aspergillosis the PCR was positive in 10% and the ELISA in 6%, but both were positive in zero patients.

He also described a patient with culture-proven, Aspergillus meningitis in whom galactomannan was detectable in the CSF. He spoke about other examples of this in the literature. He also described a single patient with chronic granulomatous disease who had strongly positive galactomannan ELISA test in pus from an abscess containing Aspergillus, but a negative serum antigen. This patient had a weekly positive Aspergillus antibody in the blood.

Dr Verweij then discussed Candida serology briefly and the detection of mannan and anti-mannan antibodies. The one published paper on this topic describes 12 positives, almost all in Candida albicans infections, but there is some anecdotal evidence that this test may be positive in non- albicans infections. The sensitivity of the test was 85.7%, but the specificity was poor because of too many positives. The work had only been conducted in non-neutropenic patients and the applicability to other patient groups was unclear.

Detection of glucan in body fluids

Dr Richardson described the principle of the G-test in which 1,3 -d-glucan (which circulates in the blood in some fungal infections) can be detected. He referred to a paper in Journal of Clinical Microbiology (Misutske et al, J Clin Microbiol 1996;34:18) in which 27 of 39 candidemic patients had a positive test for glucan. A cortisone-treated rat model showed that glucan was detectable in serum in this model and concentrations were reduced by treatment with amphotericin.

Different systems could be developed and Dr Richardson described in outline terms his work with a company in this area which is at an early stage of development.

Jean-Paul Latgé then described a collaborative laboratory project he had done with Prof. H. Yamaguchi in Japan (Tokyo) to compare detection of 1,3 -d-glucan with galactomannan in patients from Tuebingen with proven or probable invasive asprergillosis and control patients. Unfortunately there was no correlation between these tests (R=0.019). Therefore there is considerable doubt about the utility of glucan testing and much additional work needs to be done to compare the two tests for the detection of polysaccharides in serum.

PCR for Aspergillus

Dr Bretagne then described his PCR system for detecting Aspergillus DNA in blood. He incorporates 0.4 mMol of dUTP and 0.5 :M UNG into his system to prevent false positives. He described the work that was published recently (Bretagne S et al, Clin Infect Dis 1998; 26: 1407-12) in which Aspergillus galactomannan was compared with PCR in haematological patients with proven or probable invasive aspergillosis. In no instance was the PCR positive before the Aspergillus galactomannan ELISA and there were some patients that were never PCR positive, but were Aspergillus galactomannan positive. However, the sensitivity of this assay is not as sensitive as some of the other systems tested and so the apparent superiority of antigen could be a reflection of that.

Considerable discussion ensued on the appropriate sample type. Dr Bretagne described that in the pilot project they compared serum and plasma with white cell pellets for the detection of Aspergillus DNA using real-time PCR (Taqman). This was done with spiked Aspergillus DNA. It was clear that plasma was a substantially worse sample than the other two and there was not much to choose between serum and white cell pellets. Given that serum was a more commonly collected sample he elected to optimise his PCR system with serum. It was acknowledged that there was no uniform method for extracting DNA from serum and that this could be a limitation of his method, irrespective of PCR technology itself.

Performance of panfungal PCR

Dr Hebart described the results of the panfungal PCR approach that has been adopted in Tuebingen. He made the point that there is probably a very low number of fungal cells per ml of blood and a very high sensitivity is essential. He also mentioned that there is an increasing number of different fungal pathogens and it would be desirable to detect more than one or two genera. Furthermore it may be necessary to detect different species, if this cannot be done by culture. PCR provides an opportunity to do this. There are many highly conserved genes in organisms including the 18S rRNA gene, 28S rRNA gene, mitochondrial genes and parts of the internally transcribed spacer regions. The Tuebingen approach has been to adopt amplification of the 18S rRNA gene.

He then described in detail the DNA extraction used in his laboratory. The first step is some form of cell lysis, the second protein precipitation and third, DNA extraction/precipitation. The impact of different procedures may vary from specimen to specimen and Dr Hebart distinguished blood from BAL fluid in this respect. Essentially the system that is used in Tuebingen starts with hypotonic lysis of red cells from blood and lysis of leukocytes by proteinase K in blood and BAL fluid. This is then followed by spheroplasting of fungal cells. Then protein is precipitated and DNA purified from blood with the QIAmp tissue kit (Qiagen). In contrast, for BAL fluid, protein is precipitated with SDS and kalium acetate and the DNA precipitated and purified with isopropanol and ethanol. These protocols have been modified and improved in various different ways. For example omitting the red cell lysis step results in Taq polymerase inhibition, omission of proteinase K or use of saponin or TritonX also results in Taq polymerase inhibition. If an Eppendorf microformat is used with 1ml of blood instead of 5ml, the procedures is about 10 times less sensitive. If ultrasound is used to break fungal cell walls instead of spheroplasting enzymes such as glucanases, the procedure is 100 times less sensitive. Compared with the Qiagen tissue kit, Dynabeads and DNAzol reduce the sensitivity of the assay by 100 and 1000 times respectively.

Following PCR there are differences in sensitivity of amplicon detection. Agarose gel electrophoresis and ethidium bromide staining can detect up to 1pg DNA compared with GEL Star staining which can detect 100 fg DNA, as can a Southern blot or slot/blot using digoxigenin-labelled oligoinucleotides. The PCR-ELISA format using digoxigenin-labelled amplicon and biotin labelled primer can detect 10fg DNA.

Finally Dr Hebart discussed storage time using EDTA whole blood samples spiked with 10 ⁵ Aspergillus conidia. Essentially there was no difference with the temperature the sample was stored at between 4 ^o and - 80 ^o for up to 7 days. By 28 days there was growth of Aspergillus conidia at 4 ^o so that one CFU was detectable and some loss of viability of DNA at -20C ^o. However, detection at -80 ^oC was unchanged.

Real time PCR for Candida and Aspergillus

Dr Guiver then described in detail the methodology developed at the Public Health Laboratory Service in Manchester. The initial impetus for this work came from the National Meningococcal Reference Laboratory which provides a national PCR service for meningococcal disease. Two Taqman real time PCR machines are able to service up to 100 specimens a day, providing a result in under 3 hours. The set up, amplification and detection is all done in one tube which is discarded after the test. This is simple, time efficient and reduces contamination.

Dr Guiver described the theory behind the fluorescent probes used as an integral component of the Taqman system. Continuous monitoring of fluorescence emitted provides a positive quantitative signal, as soon as it has reached a predetermined threshold. Use of up to 3 different fluorescent probes allows 3 signals to be monitored simultaneously. This has been adopted in the case of Candida infection so that a primary probe is for the Candida genes and then specific probes for C. glabrata, C. tropicalis, C. krusei, C. parapsilosis, C. guilliermondii and C. dubliniensis.

The system was initially utilised on positive cultures from agar plates. There was an excellent correlation with conventional testing procedures. A prospective study in haematology patients at the Christie Hospital (receiving antifungal prophylaxis) was then initiated. Six hundred samples of 1ml blood were obtained in 57 patients. Six patients had at least one positive sample. Eight patients had unreproducible weakly positive samples. The mean assay time was 2.5 hours.

Light-Cycler PCR

Dr Loffler then described preliminary results with the other real time PCR system, Light-Cycler. In this system, closed glass capillaries are used (up to 24 at a time) with an assay time of 45 minutes. Work with spiked samples showed a high level of sensitivity and a linear range for quantitation. The cost of a light cycler is substantially less than a Taqman system, but slightly more complex to set up and run. It has the same advantages as the Taqman system in terms of a 'closed tube' system, which should minimise false positives because of 'contamination' and by amplicon carryover.

The Roche PCR system

Dr Karina Orle from Roche Molecular Systems then described the PCR-ELISA under development there. The 18s rRNA gene was the target and the objective was to produce one Candida-specific and one Aspergillus-specific test. Each test contains an internal PCR control of a plasmid DNA with separate Probe binding sites. Pilot work had demonstrated that the Aspergillus PCR-ELISA also detected Biverticulum and Penicillium species, but not Eupenicillium species.

The assay requires 1 or 2ml whole blood. The extraction procedure starts with an extraction reagent and proteinase K, then heating to 65 ^oC shaking and then 95 ^oC without shaking. Following this, 20Fl is lysed with magnetic beads by rolling and then separated using a magnet. DNA is then eluted and heated to 80 ^oC in an open tube. The assay as currently formatted has a sensitivity of 1 conidium/ml of blood. An alternative sample of a white cell pellet was not as good in this system. Larger volumes increase sensitivity. All the work had been conducted on samples stored at 4 ^oC.

In collaboration with the Nijmegen group, an earlier version of the assay for Aspergillus had been evaluated in BAL fluids. It was compared to culture. The results are summarised below:

Possible new serodiagnostic tests

Dr Latgé then described several avenues for new, more sensitive rapid detection systems. Possibilities for such tests could include cell wall markers such as galactosanimoglycan or recombinant 1,3 -d-glucan, or antigens galactofuran or ribonuclease. Other metabolites were also possibilities. Substantial additional development work would be necessary before clinical evaluation could commence. However, the development of a large bank of sera from proven cases in multiple patient groups, already tested against current diagnostic markers was key to speeding future better tests.

Performance of panfungal PCR in allogeneic BMT patients

Dr Hebart then briefly described his results in allogeneic bone marrow transplant recipients. PCR was compared to conventional diagnosis using radiology, culture, microscopy and histology, but not antigen detection. Some patients did not have autopsies and so the diagnosis could have been missed in some patients with positive PCR tests, but other tests negative. If a single test was taken as positive, the sensitivity and specificity were 100% and 65% respectively. If 2 positive samples were regarded as positive, the sensitivity was 100% and specificity 84%. He also described the subsequent development of invasive aspergillosis in all 7 patients in whom BAL fluid pre-transplant was positive.

Consensus from working groups

1. PCR - DW Denning

   Conclusions were:

1. Plasma is an inferior sample
2. Whole blood or serum should be directly compared
3. No good fungal DNA extraction method has been published for serum which has the same sensitivity as whole blood. Further work is necessary.
4. Very little work has been done on Candida PCR performance in non-neutropenic patients.
5. PCR results should be compared with culture in the case of Candida and both proven cases and other good diagnostic tests (such as antigen) in Aspergillus infections.
6. Sample volume unclear, but probably not less than 1ml to improve sensitivity.
7. Dr E Candolfi offered to coordinate a pilot study comparing serum extraction methods on spiked samples amongst the PCR labs represented.


2. Serology - J-P Latgé

Conclusions were:

1. Large studies in different patient populations were needed - especially allogeneic BMT, lung transplant, liver transplant and leukaemic patients with prolonged neutropenia, and for Candida tests, intensive care patients. In the last group a focus on those with pancreatitis, gastrointestinal surgery and the elderly was best to minimise the size of the study. There was also a need to evaluate Aspergillus serology tests in patients with chronic granulomatous disease. There was also a need to evaluate late episodes of possible infection in patients with chronic rejection, especially after transplantation.
2. For screening purposes - twice a week was thought most practicable.
3. For diagnostic purposes - sampling on 3 consecutive days was probably a good standard for studies.
4. Single batches of antigen test kits should be used for studies.
5. There was a need to evaluate the 'glucan' test against other diagnostic tests.
6. The tests should initially be validated – in those with proven infection and then compared with other tests in 'probable' and confirmed negative cases.
7. Interlaboratory reproducibility is important to evaluate and can be done by banking sera prospectively and then selecting a group of positive, negative and indetermants/weakly positive samples as a panel for participating labs to test.

Conclusions

Dr Denning thanked the participants and agreed to circulate a summary of the meeting to all participants prior to posting it on the Aspergillus website. Dr Candolfi agreed to coordinate a small study of optimal PCR extraction methods in serum.

Participants