Introduction

Actually, there are several analytical detection approaches available for human breath investigations, including gas chromatography–mass spectrometry (GC-MS) [1–7], proton transfer reaction-mass spectrometry (PTR-MS) [8–11], selected ion flow tube-mass spectrometry (SIFT-MS) [12–18] and ion mobility spectrometry (IMS) [2, 19–28] and electronic noses [29–33] and different types of sensors [34–40]. The sampling techniques include direct sampling, partly using sample loops [41–44], Tedlar bags [45–48], SPME [4, 5, 49–51] and different adsorbents [20, 26, 52]. In all cases mentioned a non-invasive and easy method for early diagnosis or therapy monitoring should be developed by identifying disease-specific biomarkers in the breath of patients.

On the other hand, first scientific evidence for microbial VOCs has been presented early in the last century [53, 54]. More recently, the detection of different VOCs and their relations to medical questions was reported, including ion mobility spectrometry [22, 55–71]. Some of the VOCs were related to bacteria taken from headspace of cultures [58, 72, 73].

The analytical technique used in the present paper is the same basic technique used to detect explosives or chemical warfare agents and was developed in a close cooperation of ISAS (Institute for Analytical Sciences, Dortmund) and the Lung Hospital Hemer, to examine hospital patients with lung cancer and airway infections [3, 21–23, 25, 74–78].

The technique, based on ion mobility spectrometry, is able to detect effectively metabolites in human breath down to the pptv or pg/L-range. For investigations of human breath at a comparatively high level of humidity, a Multi-Capillary Column (MCC) for partly pre-separating of the analytes is used in combination with a conventional ion mobility spectrometer (IMS). An IMS coupled to a MCC allows the identification and quantification of volatile metabolites present in human breath, down to the ng/L- and pg/L-range of analytes within less than 500 s and without any pre-concentration. The IMS investigations are based on different drift times of swarms of ions from metabolites formed directly in air at ambient pressure. About 10 mL of breath is necessary to carry out a full analysis.

Patients

All patients were recruited from the Department of Pulmonology, Ruhrlandklinik, University Hospital of Essen, Germany. The diagnosis of Pseudomonas was established according to the actual guidelines. Subjects with any other respiratory disease or any concomitant malignant, heart, renal, liver or collagen disease were excluded. All patients were clinically stable (no evidence of acute exacerbation for at least 4 weeks prior to enrolment). Healthy non-smokers, all employees of the hospital, served as control group. The study was approved by the ethic committee of the University of Essen and all subjects provided an informed consent.

Method

The IMS coupled to a multi-capillary column (MCC/IMS) used was a BioScout (B&S Analytik, Dortmund, Germany), consisting of the MCC/IMS and a SpiroScout (Ganhorn Medizin Electronic, Niederlauer, Germany) as sample inlet unit. The major parameters are summarized elsewhere [2, 3, 21–24, 74, 76, 79, 80]. In this spectrometer a 550 MBq [63]Ni ß-radiation source was applied for the ionization of the carrier gas (air). It is connected to a polar multi-capillary column (MCC, type OV-5, Multichrom Ltd, Novosibirsk, Russia) used as the pre-separation unit. In this MCC, the analytes of exhaled breath were sent through 1.000 parallel capillaries, each with an inner diameter of 40 μm and a film thickness of 200 nm. The total diameter of the separation column was 3 mm. The relevant MCC parameters are listed in Table 1.

Table 1 Characteristics of ion mobility spectrometer (BioScout 2010)
Full size table

All subjects were requested to exhale through a mouth piece connected to a Teflon tube. In each case, end-tidal breath controlled by a flow sensor, was collected in a sample loop of 10 mL in volume. The sample air was collected and transferred to the multi-capillary column for a first chromatographic separation after reaching three times 10 mL above the dead volume. Using the software VOCan 1.7 (B&S Analytik, Dortmund Germany), the dead volume was adjusted and fixed in the present case to 500 mL. The expiration was controlled by a CO2-sensor element integrated in the SpiroScout and recorded for each subject.

A preliminary relation between the peak position and the identity of the analyte was obtained using the database BSIMSDB 1.4 (B&S Analytik, Dortmund, Germany), but parallel measurements using e.g. GC/MSD should be realized with respect of further confirmation.

Statistical evaluation

The peaks were characterized using the software Visual Now 2.5 (B&S Analytik, Dortmund Germany), which is described elsewhere [74, 81–84]. All peaks found were characterized by their position with drift time (corresponding 1/K0-value) and retention time and their concentration represented as the peak height. For all the peaks in both of the groups, Box-and-Wisker plots were realized. The rank sum as provided by Visual Now 2.5 was used to rank the peaks with the maximum difference between both groups. The value of the rank sum is related to the Mann–Whitney–Wilcoxon U value directly (rank sum = norm U = U/n1n2), were n1 and n2 are the numbers of cases in each group.

Results

To realize a non invasive identification of pathogens the exhaled breath of 53 persons (24 patients suffering chronic or infectious on Pseudomonas and 29 healthy controls) was investigated using the ion mobility spectrometer type BioScout. In total 224 signals were found as shown in Fig. 1.

Fig. 1
figure 1

Position of the peaks within the IMS-Chromatogram

Full size image

For each of the peak the rank sum was calculated using Visual Now 2.2 to be able to find the signals with the maximum potential to discriminate between both of the groups, see Table 2.

Table 2 Position of peaks and rank sum values of the peaks with rank sum <0.2
Full size table

In total, 21 signals seem to be able to differentiate the two groups control and pseudomonas with a rank sum values less than 0.2.

The position of the peaks with the rank sum less than 0.2 is shown in Fig. 2.

Fig. 2
figure 2

Position of the peaks with the maximum potential with respect to discrimination between the group of Pseudomonas and the healthy controls within the IMS-Chromatogram. The cross line related to the single spectrum (below) and the chromatogram (right) shows the position of the peak with the lowest rank sum

Full size image

Furthermore, for all 224 signals Box-and-Wisker plots were realized. The peaks with the lowest rank sum values F (0,107) and PS0 (0,112) show rather good separation of both groups. A further discrimination between the first two peaks F and P_20, which were rather close to each other, located by 1/K0 and retention time, was not realized in the present study (see Table 2). Therefore, the first and the third peak should be considered more in detail. Surprisingly, in all cases the concentration of the analytes was significantly higher in the control group than in the Pseudomonas group.

The values of sensitivity and specificity, and the positive and negative predictive value were summarized in Table 3 as obtained from visual discrimination. The peak PS0 shows a sensitivity and a negative predictive value of 100%. The sensitivity was higher than the specificity in both cases. But, the peak PS0 was really low and near to the noise and in some cases the measurements were not carried out until 600 s. This becomes visible from null values for four cases as shown in Fig. 3.

Table 3 Sensitivity, specificity and predictive values related to the peaks F and PS0 by visual discrimination
Full size table
Fig. 3
figure 3

Box-and-Wisker plot of the signals of peak F (above) and PS0 (below) in both the groups (Control vs. Pseudomonas) with a rank sum value of 0.107 and 0.112, respectively

Full size image

In addition, a calculation of the best position of a threshold with respect to differentiation of the two groups was realized for all 224 peaks. The Table 4 compares the results for the positions 1–7 of the rank sum from Table 2.

Table 4 Calculation of the best threshold to discriminate the two groups of healthy controls and Pseudomonas
Full size table

Surprisingly, the peak with rank sum position 3 delivers the best accuracy (0,88). For Peak PS0 the sensitivity found was 100, the specificity 74, the positive predictive value 82%, the negative predictive value 100%.

Therefore, the signal F should be considered for separation between control and pseudomonas group with the highest accuracy. Generally, the finding needs further confirmation and a higher number of subjects included within the study (Fig. 4).

Fig. 4
figure 4

Comparison of single spectra of Peak F (red control group—blue Pseudomonas group, left: both groups together, center: control group, right: Pseudomonas group)

Full size image

Summary

To realize a non invasive identification of pathogens, the exhaled breath of 53 persons (24 patients suffering chronic or infectious on Pseudomonas and 29 healthy controls) was investigated using an ion mobility spectrometer type BioScout. In total 224 different signals were found in the exhaled breath. Actually, 21 different signals are able to differentiate the two groups Control and Pseudomonas with rank sum values less than 0.2. The best separation was found by peak F with rank sum 0,107. In this case, the sensitivity found was 89%, the specificity 77%, the positive and negative predictive values were 83% and 86%, respectively. Generally, the finding needs further confirmation and a higher number of subjects included within the study.