PCR-RFLP as a useful tool for diagnosis of invasivemycoses in a healthcare facility in the North of Brazil

a Programa de Pos Graduação em Medicina Tropical, Universidade do Estado do Amazonas, Manaus, AM, Brasil b Laboratório de Micologia, Departamento de Pesquisa, Fundação de Medicina Tropical Dr. Heitor Vieira Dourado, Manaus, AM, Brasil c Laboratório de Micologia, Departamento de Pesquisa, Fundação Oswaldo Cruz, Rio de Janeiro, RJ, Brasil d Water Sciences Institute, Cranfield University, Cranfield, Bedfordshire, United Kingdom e Laboratório de Micologia, Coordenação de Sociedade, Ambiente e Saúde, Instituto Nacional de Pesquisas da Amazônia, Manaus, AM, Brasil


Introduction
In recent years, the incidence of invasive mycoses has globally increased rendering in an important cause of morbidity and mortality, especially in immunocompromised patients, such as those having acquired AIDS, transplant patients using corticosteroids and/or patients undergoing chemotherapy in intensive care units [1,2,3]. Due to the high morbidity and mortality, and a tendency to rapidly spread to other organs, early diagnosis and a correct treatment of invasive fungal infections are essential with sensitive and rapid diagnostics in urgent need [4].
Currently, in developing countries including Brasil, the identification of invasive fungal infections comprises conventional culture, biochemical methods, microscopic determination of micromorphology and immunological assays. However, these traditional methods (culture and direct) are most often used to detect agents that cause mycoses. The MALDI-TOF technique is being used as an alternative to precocious identification of fungal species; however, it is still not a reality in Brasil [5,6]. To overcome these limitations, molecular approaches can be used for the detection and identification of pathogenic fungi [7,8].
Methods employing polymerase chain reaction (PCR) are among the most promising alternatives and are increasingly being applied in routine detection of pathogenic microorganisms. The greatest advantages of PCR can be seen in its sensitivity, specificity and speed with test results being available within hours [9,10]. Real-Time PCR and microarray assays have been investigated for the detection and identification of pathogenic fungi [11,12,13]. However, these methods are the most of the time, unavailable in the routine of mycology laboratories of developing countries.
Santos et al. [14] presented a PCR-RFLP assay targeting the ITS region from the fungi rDNA that was able to differentiate the most important agents from invasive mycosis. The present work is the continuation of the assessment of this study by applying it in the routine of a hospital with patients at risk of being affected by invasive mycosis. The Fundação de Medicina Tropical Dr. Heitor Vieira Dourado (FMT-HVD) is a reference hospital for the treatment of tropical diseases in the North of Brasil. Its wards are mostly occupied by immunocompromised patients with AIDS, who are highly susceptible to infections with opportunistic pathogens.

Materials and methods
The study was conducted in three stages: a) PCR-RFLP (NL4-ITS5, DdeI digestion) of reference strains; b) PCR-RFLP for identification of 90 cultures of clinical isolates comparing the results with conventional techniques; and c) determination of accuracy of PCR for detection and RFLP for the identification of causative agents of invasive mycoses in biological samples from 120 patients with clinical suspicion of invasive mycoses, prospectively. For the analysis of these latter, data was supplemented with epidemiological characteristics of patients with fungal infection.

Biological samples
120 biological samples were investigated including whole blood (7), blood culture (33), cerebrospinal fluid (CFS) (70) and bone marrow aspirate (10). Samples were taken from patients with suspected invasive mycosis who had been referred to the Mycology Laboratory of the FMT-HVD between September and December 2012, to the realization the conventional PCR and RFLP. Anonymized information of patients (sex, age, place of residence, clinical specimen investigated, HIV serology, viral load of HIV and CD4 + cells/mm 3 ) was obtained from the computerized system "iDoctor hospital" used in FMT-HVD.

Detection and identification by conventional methods
Cultivation of biological samples was performed on Agar Sabouraud (BD Difco, Sparks, USA) and Agar Mycosel (BD Difco, Sparks, USA). Colonies growing on Sabouraud agar were identified by standard laboratory methods, including micro morphological and physiological tests (germ tube production, growth on Agar Niger and CHROMagar™ Candida (Becton-Dickinson, Sparks, USA)), incubation at 42°C, and testing for carbohydrate and nitrogen assimilation and carbohydrate fermentation.

Detection and identification by PCR-RFLP
Generation of PCR products and subsequent digestion was performed as described by Santos et al. [14]. DNA was extracted from samples (200 μL of fungal biomass or biological sample) using the QIAamp Blood and Tissue kit (Qiagen, Hilden, Germany) following the recommendations of the manufacturer. The DNA was quantified by absorbance at 260 nm using Genequant (Eppendorf, Hamburg, Germany) and 20 ng of the extracted DNA served as a template for   [15]), 50 μM dNTPs, and 1 U Ampli-Taq DNA polymerase. PCR was performed using a thermocycler (Kyratec's SuperCycle, Republic of Korea) under the following PCR conditions: initial denaturation of 5 min at 94°C, 40 cycles of 30 s at 94°C (denaturation), 30 s at 50°C (annealing), for 30 s and 1 min at 72°C (extension), followed by a final extension for 10 min at 72°C. PCR products were visualized by electrophoresis on a 2% agarose gel and stained with SYBR® Green (SYBR Safe DNA Gel Stain, Invitrogen, Carlsbad, USA). A DNA Ladder Mix of 100 bp (SM0331, MBI Fermentas, St. Leon-Rot, Germany) served as size marker. For RFLP, PCR products from individual isolates were digested with 10 U of the restriction enzyme DdeI (Thermo Fisher Scientific, Vilnius, Lithuania) for 3 h at 37°C, and subjected to electrophoresis as described above. The sizes of the PCR products and restriction fragments generated from the isolates were compared with the corresponding nucleotide sequences deposited in the GenBank database (Table 1).

PCR-RFLP's profiles obtained with the reference strains
Eight reference strains were submitted to PCR-RFLP in order to investigate if the profiles described in silico (Table 1) were similar to that obtained experimentally. As shown in Fig. 1, the PCR-RFLP profiles were correctly predicted by the in silico investigation. In addition, RFLP patterns distinct all chosen reference strains allowing their unequivocal identification.

PCR-RFLP's profiles obtained with cultures from clinical isolates
PCR-RFLP was evaluated for identifying cultures obtained in the routine of a healthcare facility in the North of Brasil, 90 clinical cultures of fungi previously isolated from clinical samples were investigated. These same cultures were also submitted for identification by conventional methods (phenotypic assays). It was observed 100% accordance with the conventional and PCR-RFLP identification. Both methods take, on average, 5 and 2 d, respectively.

PCR (NL4-ITS5) sensibility/specificity and PCR-RFLP's identification of fungal agents in biological samples
PCR-RFLP was also evaluated for the detection/identification of fungal agents directly in biological samples. It was carried out an accuracy study comparing both the PCR-RFLP and conventional methods. 120 biological samples were assessed, including whole blood (7), blood culture (33), CFS (70) and bone marrow aspirate (10).
Seven biological samples were considered positive by conventional methods. None of these culture-positive results was PCR-negative corresponding to a 100% sensibility ( Table 2). On the other hand, 12 samples were considered positive only by PCR but tested negative by conventional methods. In addition, to the high sensitivity, PCR-RFLP took only two days while the conventional identification took between 3 and 25 d (with an average of 5 d), for the fungal agents identification.
All the 12 patients considered positive only by PCR had been previously diagnosed with cryptococcosis and submitted to an antifungal treatment against cryptococcal meningitis ( Table 3). The digestion of the PCR products allowed the identification of C. neoformans in all these 12 samples.
The seven patients diagnosed by both (conventional and PCR methods) were mostly men (n = 5); age range between 18 and 58 and the majority of them lived in the East Zone of Manaus. Five of them were HIV positive presenting viral loads between b50 to 170,660 copies/mL and lymphocytes CD4 + between 5 and 491 cells/mL. The RFLP patterns allowed identifying their fungal agents as C. neoformans (n = 3) in CSF samples, C. gattii (n = 1) in CSF sample, H. capsulatum (n = 1) in a sample of bone marrow and C. albicans (n = 2) in blood cultures. The conventional assays and PCR-RFLP showed identical species identifications (Table 4).

Discussion
The present study demonstrated that PCR-RFLP is a useful tool for routine detection and identification of agents of invasive mycoses in a tertiary healthcare facility in the state of Amazonas-Brazil. It was observed that: a) the products of PCR and RFLP profiles obtained with the standard strains were similar to those achieved with the in silico assays; b) all cultures of 90 clinical isolates were identically identified by conventional methods and PCR-RFLP; and c) the application of PCR (ITS4-NL5) to investigate the fungi detection directly in biological samples resulted in 100% sensitivity and the RFLP profiles allowed correct identification of the causative agents. Whereas the majority of previously published works were limited to identification of standard strains and cultures of clinical isolates, this study demonstrates the effective detection and identification of fungi by PCR-RFLP directly from clinical samples of biological origin.
The PCR-RFLP assay can be useful for laboratories that do not have real-time thermocycling capability. PCR-RFLP is a simple molecular  assay, requires only standard equipment used for molecular biology and provides results suitable for the simultaneous analysis of several species [14,15].
RFLP profiles obtained with the standard strains investigated in this work (Fig. 1) were consistent with the results predicted theoretically and in practice found by Santos et al. [14], Irobi et al. [15], Fallahi et al. [16], and Kordbacheh et al. [17]. There are a vast number of other fungal species that can cause invasive mycosis and that are not considered by the method described. Therefore, data must be analyzed with caution due to the risk of misdiagnosis.
In regard to conventional identification of cultures of invasive mycoses agents, it should be mentioned that the identification of Candida spp. uses enzymes, assimilation and fermentation trials. It is expensive when performed with commercial kits and slow when performed without them. The identification of Cryptococcus spp. is possible using the culture medium canavanine glycine bromothymol blue agar (CGB), however, test results are only obtained after an average of 3 d, and are subject to error. The identification of H. capsulatum is undoubtedly the most time consuming with cultivation requiring, on average 25 d to develop visible growth. Subsequent identification is micro morphological, confirmatory assay identification (pleomorphism test) may take over 25 d [7,8,18]. An attractive possibility for the use of PCR-RFLP, therefore, consists in the identification of colonies obtained from clinical samples. The method allows the establishment of a database containing RFLP profiles from relevant microorganisms. The identification of 90 fungal cultures isolated from clinical specimens presenting results similar to the results of conventional methods demonstrates this statement.
Regarding the use of PCR (NL4-ITS5) for detecting agents that cause invasive mycoses, the data presented in this study confirms that this test has adequate sensitivity and can be used with biological materials from different sources (blood, CSF and bone marrow).
This study also revealed that PCR detected C. neoformans in CFS samples that were considered negative by the conventional methods. Interestingly, all these samples were obtained from AIDS patients having undergone antifungal therapy against cryptococcal meningitis. This fact tempts to speculate that the CFS of these patients contained dead fungal cells or residual DNA from previous infection causing positive PCR results and explaining lack of success to culture the pathogens. It remains to be assessed whether nucleic acids from therapy-killed fungal cells remain intact longer in this highly sheltered environment [19]. If so, the interpretation of PCR results obtained from this type of clinical samples has to be interpreted with care and always to observe that the control sample in this case was efficient. Elimination of this residual nucleic acid can be obtained by sample treatment with viability dye, such as propidium monoazide might overcome this limitation [20,21].