Introduction

Bartonella bacilliformis is the etiological agent of Carrion’s disease, an overlooked illness with a lethal febrile stage and a warty phase. Its endemicity is restricted to Peru, Ecuador and Colombia, with some cases having been described in Bolivia and Chile. The transmission is by a sandfly of the genera Luyzomyia, mostly Lutzomyia verrucarum [1]. The human is the only reservoir known, and in endemic areas about 40% of asymptomatic carriers have been described [2]. In addition, Carrion’s disease-like syndromes have been related to two other Bartonella species: Bartonella rochalimae and Bartonella ancashensis [3–5]. Although its relevance remains uncertain, these species may be an explanation for the Carrion’s disease cases sporadically reported in distant areas such as Guatemala or Thailand [1]. In fact, B. rochalimae has been isolated worldwide [6,7].

Although the warty phase is easy to diagnose by the clinic manifestations, the initial febrile stage as well as asymptomatic carriers, are often misdiagnosed or non-diagnosed leading to perpetuation of the illness. Correct diagnosis of both acute and asymptomatic carriers is extremely important and adequate treatment is imperative to save lives. In endemic areas the diagnosis is usually made by thin blood smear and/or by clinical data. Despite having a specificity of microscopy of 96%, a low sensitivity of 36% has been described [8]. Moreover, other diseases such as malaria, dengue or tuberculosis that are also present, should be taken into account, since the first symptoms are common and may lead to misdiagnosis and erroneous treatments. All these factors are of enormous relevance since the mortality rates of Carrion’s disease are of 40–85% without treatment [9]. Furthermore, even despite receiving correct treatment the mortality rate is of 10% [10]. A more reliable method is blood culture but this is cumbersome, time-consuming and contaminations have been described in the 7–20% of the cultures [11]. Serologic tests have also been described and show a higher specificity of about 85% for both IgM ELISA and indirect fluorescence antibody test, but are difficult for routine practice [1]. Molecular diagnosis by PCR is probably the easiest way to achieve a more accurate diagnosis in endemic areas, as the equipment required is not as sophisticated or expensive, may be installed in different Health Regional Centers which may provide diagnosis to more peripheral patients, and the personnel may be easily trained in technique management. Several PCR approaches have been described in the literature in the last years [12,13]. However, these studies do not generally involve a large number of samples and additionally, as occurs with the remaining diagnostic tools, they are hampered by the lack of a standard case definition. In any case, PCR approaches have been showed as more effective that optical microscopic [12], being able to diagnostic Carrion's disease patients in acute phase previously classified as negatives by thin blood smear. Nonetheless, a critical issue is the detection limit of these techniques, raising doubts about its usefulness in the detection of low-bacteraemia carriers.

Dried blood spot (DBS) is used for the diagnosis of several infectious diseases [14,15], and has been proposed for use as easy method to transfer blood samples from endemic areas to reference centers in order to carry out molecular techniques for the diagnosis of Carrion’s disease [13]. Therefore, since this illness principally affects children in rural areas, DBS may be an easy solution to both the transportation of samples and for small blood volume collection in the pediatric setting.

The aim of this study was to evaluate the detection limit of three PCR approaches designed to detect B. bacilliformis, both in blood and filter papers to test their potential use for transferring samples from endemic areas to reference centers.

Materials and Methods

Results

When DNA was directly extracted from the blood, the detection limit was 5 CFU/μL for both the Bartonella-specific 16S rRNA and the fla genes. Meanwhile, a limit of 500 CFU/μL was obtained on amplification of the its region. In the case of DBS, the Bartonella-specific 16S rRNA PCR approach showed the lowest detection limit, which was also of 5 CFU/μL. Concerning dry blood, despite the detection limit being the same for 16S rRNA and its, the sensitivity decreased for fla when the detection limit dropped to 500 CFU/μL compared with 5 CFU/μL obtained directly from blood (Table 1). It was of note that fainter bands were always obtained with DBS.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Detection limit for the 3 PCR approaches studied both for blood samples and dried blood spots.

https://doi.org/10.1371/journal.pntd.0004529.t001

Regarding specificity, the 16S rRNA gene amplifies for all Bartonella species (either in vivo or in silico) but a positive result was also obtained when tested B. melitensis. The its amplification assay was specific for Bartonella spp., and no other of the tested microorganisms had a positive PCR. Moreover, the its scheme might allow to distinguish between different Bartonella spp. by the different amplified size. The fla gene amplification was also specific for Bartonella species and differentiates between Bartonella spp. causing Carrion’s disease (B. bacilliformis, B. rochalimae and B. ancashensis) and the remaining Bartonella causing human disease (Table 2) once no amplification was obtained or predicted for the last ones.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Amplification sizes in different Bartonella spp. for each of the three PCR approaches in study.

https://doi.org/10.1371/journal.pntd.0004529.t002

Discussion

Carrion’s disease is an overlooked and restricted disease that affects the poorest populations living in remote rural areas, which badly communicated, without equipped laboratories, and with many other illnesses with a common symptomatology [1]. Thus, correct diagnosis of Carrion’s disease is essential, particularly since misdiagnosis is frequent [12, 19]. PCR techniques rank among the most rapid techniques to diagnose B. bacilliformis. For this reason, the determination of the detection limit of these techniques is extremely important. For this study we have chosen three approaches, the amplification of 16S rRNA, the hypervariable intergenic transcribed spacer 16S-23S rRNA and the fla gene which codes for the flagelin protein of B. bacilliformis. The amplification of 16S rRNA has been proposed for Carrion’s disease diagnostic in Peru [12]. All tested Bartonella had an amplified product of 438 bp. Moreover, the in silico analysis showed that these primers are able to amplify all Bartonella spp. Then, this PCR approach may be also useful in other environments to detect and identify other Bartonella spp. either combining with sequencing or RFLP.

The its amplification permits to differentiate between B. bacilliformis and B. ancashensis from the main pathogenic Bartonella spp [20]. In fact the its region has been used in different studies of Bartonella spp [7].

Regarding fla, this gene permit to distinguish between the three Bartonella causing Carrion’s disease: B. bacilliformis (940 pb), B. rochalimae (974 bp) and B. ancashensis (937 bp) from the remaining Bartonella spp. with clinical interest. However, one exception is B. clarridgeiae (997 bp). Additionally, and in silico analysis showed that B. schoenbuchensis will also results in a positive fragment of 1008 bp.

Our results show that the Bartonella-specific 16S rRNA PCR seems to be the best of the techniques analyzed to detect the presence of B. bacilliformis in blood samples (5 CFU/μL) since the lowest detection limit was achieved on comparison with fla and its PCRs. These results are in accordance with Angkasekwinai et al. [21], who reported a detection limit of 1 and 10 copies/μL in a loop-mediated isothermal amplification when the detection limit was determined using bacterial genomic DNA alone or in the presence of human plasma respectively. This sensitivity might allow diagnosing the acute cases of Carrion’s disease, in which the mean percentage of infected RBCs is 61% (ranging from 2 to 100%) [22]. Nonetheless, the concomitant use of these PCR approaches will provide information about other Bartonella spp. infections.

Filter paper may be an alternative for easy transportation of samples from endemic areas to reference laboratories but the decreasing sensitivity of the results must been taken into account which may lead to the non-detection of cases with a low bacteremia. Although the same detection limit was obtained for 16S rRNA PCR both directly from blood and filter papers, the bands were fainter in the latter. It is true that 1 week delay in the sample processing could affect the PCR by increasing the detection limit. Nonetheless, in rural settings the transfer of samples to reference centers is associated with bad communications ways, resulting in some days from sample collection to molecular determinations.

None of the non-Bartonella microorganisms included in the study were positive when its or fla PCRs were performed. Nonetheless, when Brucella spp. was tested, amplification was obtained to 16S rRNA PCR. Although this is a limitation, it is need to take into account that a diagnostic should to be performed both in the adequate clinic context and in parallel with other diagnostic tools such as differential PCR for Brucella diagnostic when needed [23].

The prevalence of asymptomatic people in endemic areas has already been described by PCR being 0.5% [1]. However, the number of inhabitants previous exposed increases to around 40% when serologic techniques like ELISA or IFA are performed [1]. It is need to take into account that B. bacilliformis possess tropism for both erythrocytes and endothelial cells, being then present a non-blood circulating bacterial. In the chronic illness stage (verrucuous patients) the sensitivity of the microscopical techniques decreases from the 36% described in the acute phase to less than 10% [24], highlighting the lower blood bacterial carriage and a possible transient bacteremia. Those facts might results in false PCR-negative when the technique is applied in the detection of both verrucous patients and asymptomatic carriers.

It is important to remark that in the last years 2 more sensitive PCR techniques have been described in the literature: qPCR [13] in which 24.6% of DBS samples are positive, as well as a loop-mediated isothermal amplification [21] that achieves good results on analysing Lutzomyia samples. However, qPCR requires the expertise of trained personnel and is more expensive and difficult to be implemented. Meanwhile the usefulness of loop-mediated isothermal amplification remains to be validated to detect the presence of B. bacilliformis in human clinical samples. Enrichment of the sample before conventional PCR has been proposed to increase the positivity by 55% when compared with the original blood samples [25]. However, this enrichment technique results in a 14-days delay in sample processing thereby making it unaffordable for diagnostic purposes.

To conclude, here we show that 16S rRNA PCR have low cfu detection limit and should be used with special attention to test samples from individuals with clinical suspicion of Carrion’s disease since the applicability to detect healthy carriers is not clear. The use of DBS could facilitate the transfer of samples from rural endemic areas to health facilities, despite the possibility of a small decrease in positivity. It is critical to develop rapid, sensitive and specific techniques which may be applied in endemic rural areas to avoid misdiagnosis and to facilitate the detection of asymptomatic carriers and thereby the decrease the number of B. bacilliformis cases.

Acknowledgments

We are indebted to the Citomics core facility of the IDIBAPS for the technical help. Additionally, the authors thank Dr. S. Valdezate and Dr. I. Sanfeliu for kindly provide DNA of Brucella melitensis and a clinical isolate of Bartonella henselae, respectively.

The following strain was obtained through BEI Resources, NIAID, NIH; Bartonella tamiae, Strain Th307, NR-31059.