Profiles of Volatile Biomarkers Detect Tuberculosis from Skin

Abstract Tuberculosis (TB) is an infectious disease that threatens >10 million people annually. Despite advances in TB diagnostics, patients continue to receive an insufficient diagnosis as TB symptoms are not specific. Many existing biodiagnostic tests are slow, have low clinical performance, and can be unsuitable for resource‐limited settings. According to the World Health Organization (WHO), a rapid, sputum‐free, and cost‐effective triage test for real‐time detection of TB is urgently needed. This article reports on a new diagnostic pathway enabling a noninvasive, fast, and highly accurate way of detecting TB. The approach relies on TB‐specific volatile organic compounds (VOCs) that are detected and quantified from the skin headspace. A specifically designed nanomaterial‐based sensors array translates these findings into a point‐of‐care diagnosis by discriminating between active pulmonary TB patients and controls with sensitivity above 90%. This fulfills the WHO's triage test requirements and poses the potential to become a TB triage test.


Off-line Tools for Collecting VOCs from Skin
In the literature review on the collection of skin VOCs, there are many different methods, usually involving uncomfortable sampling procedures, e.g., wrapping the desired area with a plastic bag. During the experiments, comfortable sampling methods that will increase the volunteer's compliance were chosen. Two different absorbing materials were investigated and characterized. Protocols for fabrication, sampling, and storage for both materials were established.

Characterization of PDMS as a Sampling Tool
Several important parameters related to the Polydimethylsiloxane (PDMS) (Specialty Silicone Products Inc., USA) were investigated and optimized in order to fit as a sampling tool to detection of TB VOCs from the skin: 1. PDMS Dimensions optimized to 2.5 cm X 0.5 cm with 0.017'' thickness.
2. PDMS cleaning process was evaluated in a serial experiment using: 1) Decon 90 with distilled water; 2) acetone and methanol washing; and 3) Plasma process. The optimal cleaning process was Decon 90 with distilled water.
3. Thermal conditioning process was examined in a range of temperatures from 180°C to 270°C, under a constant pure gas flow (Nitrogen or Helium) for up to 90 minutes. The optimal conditioning temperature was determined as 270°C, under a constant flow of Nitrogen for 60 minutes. An example of PDMS before and after thermal conditioning is showed in Figure S1. The abundance of the materials was significantly reduced after the conditioning process. The mutual materials between the two graphs are mostly silicon produces from the PDMS and Gas chromatography-mass spectrometry (GC-MS) column. Figure S1. GC MS chromatography for PDMS before the chosen conditioning process (black) and after conditioning process (red).
4. Shelf life of the conditioned PDMS sheets was also examined at 4°C storage conditions. It was found that the samples in the glass vials, sealed with parafilm be stored up to 8 months.
5. Cover material/attachment procedure of PDMS to skin was tested to reduce the background noise during the sampling from both the cover itself and the environment.
In the literature the cover material was a gauze pad [14] ; however, during our evaluation very noisy results were received. A series of experiments was held in which several different cover materials were tested including gauze pad, parafilm, and aluminum foil (see Figure S2). A room sample was collected by hanging the PDMS in the room for the same duration of the experiment; the results are shown as a blue chromatogram in a case of aluminum foil ( Figure S2c). The marked areas in cases of gauze pad and Parafilm ( Figure S2b) highlight the high levels of the noise caused by the cover materials as the black chromatogram had the same abundance (or even higher) as the materials from the skin test. In the case of aluminum foil as a cover material, the noise levels are substantially negligible compared to the skin test. This indicates that the VOCs found in the skin sample are indeed VOCs emitted from the skin. As another control test, the room's PDMS sample had a negligible VOC profile in comparison to the skin sample, indicating that an efficient cover that protects the skin sampling from air pollutants. 7. As the distribution of secretion glands on the skin is heterogeneous, and therefore distinctive VOC profiles can be emitted from different parts of the body. As a result, we have sampled the skin in two regions including the bilateral inner arm (close to the armpit) and chest areas, as different VOC profiles are obtained. The results showed a significant difference in the VOC profile between the inner arm and chest ( Figure S3).
The skin cleaning method prior the sampling increased significantly the repetitiveness of the results and significantly reduced the differences between the different lateral sampling positions. Moreover, the percentage of the unique VOCs from the skin increased following the cleaning process. Skin cleaning process was tested in a serial experiment using distilled water, commercial alcohol preps and combination of both with different waiting time before sampling. It was found out that the optimal skin cleaning process was with alcohol prep 5 min before sampling. The average percentage of relative standard deviations was lower for the series of experiments with the cleaning process (23%) compared to the experiment without cleaning process (59%).

Figure S3. VOC profiles by PDMS absorbing material from the chest area (blue), inner arm area (red)
and a room sample as a control.

Characterization of Tenax as a Sampling Tool
In order to increase the possibility to identify TB VOC profile via skin headspace, Tenax porous polymer (2,6-diphenyl-p-phenyleneoxide) (Buchem B.V.) as additional absorbing material was used. As PDMS and Tenax have different VOC absorbing materials, the use of both had the potential to cover a chemically wider range of emitted VOCs from the skin.
Tenax is known as an excellent absorbing material for detection of VOCs via exhaled breath [71] , once the Tenax is trapped inside a glass tube. In order to adjust the use of such a powdery polymer for skin sampling, several alternatives were investigated.
Initially, a commercial tag, 'ULTRA Passive Sampler for ppb-Level Organic Vapors' (www.skcinc.com) was used to trap the Tenax and allow it to be near the skin headspace without a direct contact with the skin that may cause irritation. As expected, the results showed VOC profiles from both inner arm and chest area, which were significantly different from the profiles obtained with the PDMS (see Figure S4). Since the solution of such tags were extremely expensive and required purchase of hundreds of tags for the clinical study, more cost affordable approaches were tested. The Tenax polymer was trapped in home-made envelopes sealed with heat presser which were made from polymeric membranes in order to allow vapor transfer, while ensuring no direct contact between the Tenax and the skin. The tested membranes included polyester with different meshes (30 and 47 µ) as well as 5 µ Polyethersulfone membrane (PES) that can control the preferred polarity of the transferred gases for elimination of humidity effect. As can be seen from Figure S5 and Figure S6, the skin VOC profiles obtained from Tenax with different envelops and Ultra tag are similar and there were no significant differences. As the PES membrane was hard to be sealed by heat, it was eliminated. 47µ polyester membrane was chosen due to high availability and low price. Figure S4. Differences in the skin VOC profile at the same body location, chest area, between the two absorbing materials, PDSM (black) and Tenax (red).

Collection of Skin Headspace
Each volunteer was sampled as follows:  Two Tenax patches on the inner arm area  Two Tenax patches on the chest area  Two PDMS patches on the inner arm area  Two PDMS patches on the chest area  One Tenax patch for room sampling, placed on a table during the skin measurement as a reference.
 One PDMS patch for room sampling, placed on a table during the skin measurement as a reference.
The duplicates of the different absorbing materials are used for two lab instruments:  GC-MS for detection and characterization the skin TB-VOCs; and  Laboratory nanomaterial-based sensors array chamber for sensor performance assessment. Figure S7. Developed DFA accuracies as a function of feature number for both training (70%) and test (30%) datasets.

Wearable device
The internet of medical things IoMT device consists of a Data Acquisition System coupled with the electrodes-Microchip to pick up vital signals from the human body. The device is well equipped with an analog front end which facilitates the signal extraction and conditioning. The system also has an analog to digital device (ADC), for converting the extracted analog signal to digital form, and a microcontroller which sends digital signals to the Bluetooth transceiver, which enables communication with external devices, such as:  Programming and Debug Interfaces PDI (Program and Debug Interface).

(iv) I/O and Packages
 34 Programmable I/O Pins.