All methodological and analytical optimizations were carried out in accordance with the Good Laboratory Practice (GLP) guidelines and all clinical experiments were conducted according to the amended Declaration of Helsinki (DoH) guidelines. As mandatory prerequisites, ethical approvals (Approval no: A2020-0085 and A2021-0012) from the Institutional Ethics Committee (IEC) of the University Medicine Rostock (Rostock, Germany) and signed informed consent from all subjects were obtained.
In compliance with the unavoidable infection safety mandates, we have advanced our state-of-the-art experimental setups and methods. In order to reduce viral transmission risk, we have tested both mainstream and side-stream viral filters with pore size of 0.2 µm. Mainstream filters were to avoid viral contamination of the clinical environment, side-stream filters were used to protect sampling and analytical equipment. Special focus was employed upon confounding effects e.g. contribution/loss VOCs by selected filter and effects of sampling flow and humidity. Effects of normal and higher respiratory rates were also tested. Operational conditions for sample storage, transport and analysis are assessed. Recommendations on general experimental setups (see Fig. 6a and 6b) for safe and repeatable continuous real-time and offline sampling are made.
Operational conditions for direct, real-time analysis with filters by PTR-ToF-MS
Breath VOCs were measured continuously via a PTR-ToF-MS 8000 (Ionicon Analytik GmbH, Innsbruck, Austria). Continuous side-stream mode of sampling via a 6 m long heated (at 75–100° C) silico-steel transfer-line. In general, a continuous sampling flow of 20 sccm (i.e. ml/min) and a time resolution of 200 ms were applied34. For online measurements, the sampling flow was readjusted (to 50, 65 ml/min) to achieve breath-(phase)-resolved analysis beyond the side-stream filters. Drift tube temperature of 75°C, voltage of 610 V and pressure of 2.3 mbar were used. The E/N ratio resulted at 138 Td. At the end of every minute, a new data file was recorded automatically and the mass scale was also recalibrated. We used the following masses for mass calibration: 21.0226 (H3O+-Isotope), 29.9980 (NO+) and 59.049 (C3H6O).
VOCs were measured in counts per seconds (cps) and corresponding intensities were normalised onto primary ion (H3O+) counts. Raw data were processed via PTR-MS viewer software (version 3.228). As PTR-MS continuously records both exhaled breath and inhaled room-air, the ‘breath tracker’ algorithm (based on Matlab version 7.12.0.635, R2011a) was applied to identify expiratory and inspiratory phases28. Here, acetone as the tracker mass was used as an endogenous substance, which has significantly higher signal intensity in expiration than in inhalation. As the mass resolution of PTR-ToF-MS (~ 4000Δm/m) can assign volatiles upon their measured protonated mass and corresponding sum formula, compound names are used while discussing results35. VOCs were quantified via multi-component mixture of standard reference substances32.
Mainstream filters to avoid contamination of the clinical environment
Protective mainstream HEPA filter test: In all cases, we had to apply a mainstream high-efficiency particulate absorbing/HEPA filter (Ultipor BB25G Hydrophobic filter, CE 0088, PALL→, Dead space: 35 ml, Resistance of 3.5 cm H2O at 60 L/min) at the rare end of the breathing mouthpiece, in order to prevent any possible viral transmission (SARS-CoV-2 retention of > 99.999%) to room air from exhalation or vice versa. Nonetheless, we have also examined the real-time confounding effects from these mandatory filters.
Test protocol for confounding-effects
Transfer line of the PTR-ToF was connected to the sterile breathing mouthpiece in side-stream mode for continuous breath-resolved measurements of VOCs. After two minutes of normal oral breathing (by healthy volunteer in sitting position), the filter was connected to the rare end of the mainstream (i.e. after the side-stream connection of PTR transfer-line) and volunteer continued breathing through the filter for next two minutes. Similarly, room air was sampled for two minutes with and without filter for additional comparisons. Sampling attachment is presented in Fig. 6c. Test results are presented in Fig. 1a, where VOC data are normalised onto the corresponding values from the steady state of breathing without filters. The steady state of breathing is the actual comparison point for physiological and contamination effects induced by main-stream filters.
Mainstream sampling viral filter test: We have tested the applicability of a mainstream viral filter (Gibeck 0.2 µm, Filter Small S, Ref: 19512, CE 0124, Iso-Gard→) with dead space of 26 ml, resistance of 1.6 cm H2O at 60 L/min and with filtration efficiency of > 99.9999% for bacteria and of > 99.999% for viruses.
Test protocol for estimating real-time effects
Transfer line of the PTR-ToF was connected to the sterile breathing mouthpiece in side-stream mode for continuous breath-resolved measurements of VOCs. After two minutes of normal oral breathing (by healthy volunteer in sitting position) without any filter, the filter was connected to mainstream (before the side-stream connection of PTR transfer-line) and volunteer continued breathing through the filter for next two minutes. Similarly, room air was sampled for two minutes with and without filter for additional comparisons. This test was repeated with and without the rare-end HEPA filter. Sampling attachment is presented in Fig. 6c and test results are presented in Fig. 1a.
Pair-wise multiple comparisons of observed differences (in VOC concentrations) between breathing without and with mainstream filters are tested via repeated measurements ANOVA on ranks (Dunn’s post-hoc method, p value ≤ 0.05). Results are presented in Supplementary Figure S1 online.
Side-stream filters to avoid contamination of the analytical equipment
We have tested small syringe filters (Sartorius PTFE 0.2 µm, Ref: 16596-HYK, Non-pyrogenic CE 1639, Ministar →, Diameter: 26 mm, Dead space: 0.25 ml) for side-stream applications. The HYK version is with male luer-lock i.e. suitable for our additional connections.
Test protocol for estimating analytical confounders
Transfer line of the PTR-ToF was connected to the Liquide Calibration Unit (LCU) directly and via the syringe filter for comparisons. Breath matrix adapted multicomponent VOC standard mixture were introduced from LCU in variable concentrations and with different humidity conditions. Outputs were measured continuously and in real-time with and without filters. Loss/contribution of VOCs via filter and the effects of different inlet flows were examined. VOC intensities obtained through measurements (via LCU and PTR-ToF-MS) without filters represent 100%, VOC intensities obtained through measurements with syringe-filter are presented in % in relation to measurements without filter. VOC concentration was approximately 100 ppbV and concentrations with a sampling flow rate of 20 sccm (i.e. ml/min), 50 sccm. Results are presented in Fig. 2.
Test protocol for continuous real-time sampling
Transfer line of the PTR-ToF was connected to the sterile breathing mouthpiece in side-stream mode for continuous breath-resolved measurements of VOCs. General setup is presented in Fig. 6a. After two minutes of normal oral breathing (by healthy volunteer in sitting position) without any filter, the syringe-filter was connected to side-stream (between mouthpiece and PTR transfer-line) and volunteer continued breathing through the filter for next two minutes. Similarly, room air was sampled for two minutes with and without filter for additional comparisons. Results are presented in Fig. 1b, where VOC data are normalised onto the corresponding values from the steady state of breathing without filters. The steady state is the comparison point for analytical effects induced by side-stream filters.
Pair-wise multiple comparisons of observed differences (in VOC concentrations) between breathing without and with side-stream filters are tested via repeated measurements ANOVA on ranks (Dunn’s post-hoc method, p value ≤ 0.05). Results are presented in Supplementary Fig. S1 online.
Contributions or losses of VOCs via the filters, filter’s performances and stability test
Contributions and losses of VOCs via mainstream and side-stream filters are examined under sampling flows of 20, 50 and 65 ml/min. VOC data were normalised onto the corresponding values from sampling without filters at sampling flow of 20 sscm. The steady state of breathing was compared for evaluating any contribution or loss of VOCs by the mainstream HEPA filters and side-stream syringe-filters. Results are presented in Supplementary Fig. S2 online.
Filter’s performances were examined under higher sampling flow (up to 100 ml/min) for breath phase-resolved analysis and also under higher breathing frequency (20–30 breaths/min). Results are presented in Fig. 3. Figure 3a represents side-stream sampling at inlet flows of 20, 50 and 65 ml/min via syringe-filters under normal respiratory rate of 10–14 breaths/min and Fig. 3b represents side-stream sampling at inlet flows of 65 ml/min via syringe-filters under higher respiratory rate of 20–30 breaths/min. The effects of continuous breath-resolved measurement time (i.e. 13 min of breathing with normal respiratory rate) on the stability of a syringe-filter was examined. Results are presented in Supplementary Fig. S3 online.
Advancement of experimental setup for continuous real-time breath sampling in the screening scenario of a SARS-CoV-2 test center
As per Fig. 6a, the area of analytical instrument and its operator were separated from the patient/subject entry and sitting area via a transparent polycarbonate flexi-glass wall (0.5 cm of thickness). Only 1 m of the PTR-transfer-line (i.e. 6 m long and heated at 75–100° C – in order to degrade any trace-passed viral entity) entered towards patient’s side via a custom-made entry hole (circular and tightly-fitting) of the flexi-glass wall. The 1 m (i.e. at the patient side) of the transfer-line was enclosed by a sterile and disposable polyethene cover and it was changed after each participant. After entering the sampling area patient sat on a chair, remove his/her face masks and orally breathe in and out through the sterile and disposable mouthpiece (i.e. placed via operator before the subject/patient’s entry), which has a HEPA filter connected at the rare end. PTR-transfer line was connected to the mouthpiece in side stream mode just after our recommended side-stream syringe filter (pore size: 0.2 µm, with luer-lock). The mouthpiece (+ PTR transfer line) height was adjusted and fixed via stationary metal-clamp strands at the face height of subjects (according to his/her sitting position). Thus, participants did not touch any instrumental parts with hands and they were only allowed to encircle the sterile mouthpiece (of the customised sampling attachment) by their lips for oral breathing. As per our standard sampling manoeuvre2, after a minute of metronome controlled paced breathing (with respiratory rate of 12/min), volunteers continued spontaneous breathing (i.e. without the metronome) for next minutes. Used mouthpieces, viral filters and disposable polyethene cover etc. were disposed immediately after each measurement and the sampling area was disinfected and adequately ventilated before and after each participation. Our setup was successfully and safely applied for breath VOCs screening in > 700 individuals at the SARS-CoV-2 test centre of our university hospital during the surge of the delta variant33. Besides COVID-19 patients, we could safely profile breath VOCs from patients infected and/or co-infected by other infectious respiratory pathogens e.g. Haemophilus influenza, Streptococcus pneumonia and Rhinovirus etc. within our setup.
Method optimization for manual offline breath sampling under COVID-19/similar conditions
Experiment for sampling applicability of glass-syringes: In order to check the PoC feasibility of manual CO2 controlled alveolar breath sampling in glass-syringes (volume 50 mL), we have connected the glass syringe behind our shortlisted filter i.e. connected at side-stream to the sampling mouthpiece. Real-time breath VOC concentrations were measured by PTR-ToF-MS in parallel. A small portable capnograph (EMMA™ PN 3639, Ref: 605102, Masimo→ Sweden AB, Danderyd, Sweden) was attached at the mainstream for visual control of the exhaled alveolar plateau (i.e. pET-CO2 controlled) and just after that the PTR transfer-line was also connected to the mouthpiece in side-stream mode for immediate comparison of actual breath vs. syringe sample. The end point of the mouthpiece was connected to a mainstream HEPA filter. Please refer to Fig. 6c.
Test measurements
After two normal exhalations by the healthy volunteer, an alveolar breath (pET-CO2 controlled) was sampled in the glass-syringe. Thereafter, the volunteer sealed the breathing end of the mouthpiece and then within the same minute, the sampled syringe was injected back to the mouthpiece and measured as a breath by the connected PTR-ToF (see Fig. 6c). Thus, the VOC concentrations during syringe sampling can be compared directly to the injected syringe sample.
In order to evaluate any dilution effects even from the tiny dead space of the glass syringe and syringe-filter, we have compared samples taken from same individuals, with and without flushing the instrumental dead space via 1–2 demo exhalations. Results are presented in Fig. 4.
Pair-wise multiple comparisons of observed differences (in VOC concentrations) between direct breath vs. syringe samples are tested via repeated measurements ANOVA on ranks (Dunn’s post-hoc method, p value ≤ 0.05). Results are presented in Supplementary Fig. S1 online.
Advancement of experimental setup for manual offline sampling at the COVID-19/similar test centre and in the COVID-19 intensive care unit (ICU)
As per Fig. 6b, investigator/operator wore the recommended personal protective equipment (PPE) at all time. Test subjects sat straight on a disinfected chair and did not touch any instrumental parts but only encircled the breathing-end of the sterile mouthpiece by his/her lips for oral breathing. In accordance to our standard sampling manoeuvre2, after a minute of metronome controlled paced breathing (with respiratory rate of 12/min), volunteers continued spontaneous breathing (i.e. without the metronome) for next minutes. The syringe samples were collected from the second minute onward - of the spontaneous breathing. A transparent Teflon-sheet made protective separation was used to separate the mouthpiece parts at test subject’s end from the sampling parts (i.e. Capnometer, syringe-filter, glass syringe and end-point HEPA filter) at the operator’s side. Syringes were flushed-in and -out with two exhaled breaths before taking the actual sample in order to overcome any possible dilution effects from the instrumental (syringe + filter) dead space. Immediately after sampling, the syringe (with filters attached in front) was closed via the Discofix→ 3C adapter (Braun, Melsungen, Germany) adapter-system (i.e. placed between the syringe-filter and syringe) and then was detached from the syringe-filter for storage/transport. Three or more syringes were sampled from each subject. Corresponding room air was sampled separately with syringe-filter attached (via luer-lock adapter-system) for measurements and comparison of inspiratory concentrations of VOCs.
Analysis protocol for syringe samples via PTR-ToF-MS and NTME coupled with GC-MS
Injection of syringe sample with a uniform manual push over a min into PTR transfer-line:
Our preoptimized instrumental parameters for continuous VOC analysis are described earlier. Sampled syringes were analysed via direct injection to PTR transfer line. The constant transfer-line flow of 20ml/min was not enough to create adequate vacuum within the syringe to sample those uniformly. Therefore, we pushed the plunger to inject 50ml volume over one minute manually. In order to do so, an automated volume/time-controlled injector device can be used. This will overcome the problem of unavoidable vaping/condensation (storage/transport) driven mechanical resistance at the contact surface of the barrel and plunger of a glass syringe. PTR data of syringe samples was analysed as an average over one minute.
NTME of VOCs from collected syringe samples via automated flow-volume controlled sampling box:
NTDs were connected to an automatic alveolar sampling box (PAS Technology Deutschland GmbH, Magdala, Germany). An IN-stopper coupled with Discofix→ 3C adapter (Braun, Melsungen, Germany) was connected and NTDs were pierced through the IN-stopper membrane into the 50 mL glass syringes. 20 mL of samples (breath gas and room air) was sampled unidirectionally.
Comparisons of data for inter-syringe variations, effects of storage conditions on VOC stability
Transport of sampled syringes from bedside to bench
In order to prevent temperature change or condensation effects we have stored glass-syringes in tightly-sealed Styrofoam boxes. Pre-heated gel-bags were kept inside these boxes to maintain the temperature at around 37 οC. All empty/unused and sampled syringes were kept in the same temperature during storage and transport. Syringe samples were also tested by keeping in an incubator to maintain constant temperature at around 37 οC. Results are presented in Fig. 5, where VOC data are normalised onto the corresponding values from the syringe sample i.e. analysed immediately after sampling.
Effects of sample storage time (15–120 min):
In order to evaluate the effects of sample storage time, we have sampled multiple syringes from same individuals and stored them for 15, 30, 60 and 120 min respectively prior to analysis. Any loss, degradation, clustering or condensation of VOCs were analysed via online analysis. Results are presented in Fig. 5, where VOC data are normalised onto the corresponding values from the syringe sample i.e. analysed immediately after sampling. Pair-wise multiple comparisons of observed differences (in VOC concentrations) between instantly analysed syringe sample vs. syringe samples stored for different time and under different conditions are tested via repeated measurements ANOVA on ranks (Dunn’s post-hoc method, p value ≤ 0.05). Results are presented in Supplementary Fig. S1 online.
Inter-sample variations and repeatability:
Differences and variations in VOC concentrations between samples (at the steady state of breathing) from same individuals were examined in order to check the robustness and repeatability of our above-described method. Statistically significant differences were tested via repeated measurements ANOVA on ranks (Dunn’s post-hoc method, p value ≤ 0.05). Results are presented in Supplementary Table S1 online.