Apparatus, systems, and methods for detection and treatment of cancer

Abstract

The present disclosure relates to apparatus, systems, and methods for detecting an altered physiological state in a subject. More particularly, the present disclosure relates to apparatus, systems, and methods for detecting cancer in a subject using volatile-organic- chemical biomarkers emitted from a sample of the subject's biofluids or in a sample of the subject's exhaled breath. The present disclosure provides apparatus, systems, and methods for detecting prostate cancer, and for distinguishing aggressive prostate cancer from indolent prostate cancer, using VOC biomarkers emitted from one or more samples of the subject's urine.

Description

Att’y Docket No.: IUIC-168 APPARATUS, SYSTEMS, AND METHODS FOR DETECTION AND TREATMENT OF CANCER CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of prior-filed United States provisional application number 63 / 555,214 filed on February 19, 2024, which is incorporated by reference in its entirety herein. STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH

[0002] This invention was made with government support by the United States Department of Veteran Affairs. The government has certain rights in the invention. FIELD

[0003] The present disclosure relates to apparatus, systems, and methods for detecting an altered physiological state in a subject. More particularly, the present disclosure relates to apparatus, systems, and methods for detecting prostate cancer using volatile-organic-chemical biomarkers. BACKGROUND

[0004] Volatile-organic-compound (VOC) biomarkers are byproducts of metabolic pathways, which may be altered by disease initiation and progression. VOCs originate from a subject’s body and may carry information that can be expressed in lung alveolar breath, or in biofluids such as blood, urine, sweat, saliva, and others. VOCs evaporate at room temperature and standard pressure. Detection of VOCs offers a potential avenue for achieving a non-invasive readout of in a subject’s physiological state.

[0005] Current screening methods for cancer include assessing risk factors such as family history, screening for genetic risk factors and / or mutations, measuring serum antigen levels, biopsy and histology, among other things. However, these screening methods can be invasive and uncomfortable, have limited sensitivity and specificity, and frequently cannot determine tumor aggressiveness. Therefore, it is of clinical interest to develop a noninvasive and accurate screening apparatus and method to detect patients with cancer, and differentiate indolent tumors from aggressive tumors. There is a significant unmet need for reliable non-invasiveAtt’y Docket No.: IUIC-168 systems and techniques for detecting the presence or absence of cancer in a subject, and features of any cancer present, using VOC from alveolar breath samples or biofluid samples such as urine. The present disclosure addresses this need. SUMMARY

[0006] The present disclosure relates to apparatus, systems, and methods for detecting cancer state in a subject using volatile-organic-chemical biomarkers present in a sample of the subject’s biofluids such as urine or a sample of the subject’s exhaled breath.

[0007] In an aspect, the present disclosure provides an apparatus detecting cancer in a subject from volatile-organic-compound biomarkers present in the subject’s biofluid, the apparatus comprising: a sensor array of two or more sensor subunits, each sensor subunit operable to detect at least one volatile-organic-compound biomarker in a sample of VOCs emitted from the subject’s biofluid wherein each sensor subunit is selectively chemiresistive in response to exposure to a chemical of a functional group selected from: noncyclic hydrocarbon, unconjugated cyclic hydrocarbon, alcohol, aldehyde, amide, aromatic, carbonyl, carboxylic acid, ester, ether, ketone, or terpene; and wherein contacting the sensor subunit with a volatile- organic-compound biomarker having a functional group for which the sensor is selectively chemiresistive causes the sensor subunit to generate an electronic signal; wherein each sensor subunit corresponds independently to a signal output channel, each signal output channel operably connected to a computation module for receiving each of the electronic signals; and wherein the computation module is programmed to interpret an aggregate of electronic signals as corresponding to a subject cancer state. In any embodiment, the computation module may be further programmed to interpret an aggregate of electronic signals as corresponding to a cancer aggressiveness state. In any embodiment, the cancer may be prostate cancer. In any embodiment, the biofluid may be urine.

[0008] In any embodiment of the apparatus, the apparatus may comprise a sensor subunit selectively chemiresistive for an alcohol. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for an unconjugated cyclic organic compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for an aromatic compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for a carbonyl-bearing compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for an ketone compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for a terpene.Att’y Docket No.: IUIC-168

[0009] In any embodiment of the apparatus, the apparatus may comprise a first sensor subunit selectively chemiresistive for unconjugated cyclic compounds, a second sensor subunit selectively chemiresistive for carbonyl-bearing compounds, and a third sensor subunit selectively chemiresistive for terpenes.

[0010] In any embodiment, each sensor subunit may be operable to detect a change in concentration above a baseline threshold one or more of: β-thujone, dimethyl trisulfide, α- curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, 2,3,4,5-tetramethyl-2-cyclopenten- 1-one. In any embodiment, each sensor subunit may be operable to detect a changein concentration above a baseline threshold one or more of: thymol, 6-amyl-α-pyrone, 2,3,4,5- tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, or allyl-2-furoate.

[0011] In any embodiment, an electrical signal corresponding to a change in concentration from baseline of any of: unconjugated cyclic compounds, aromatic compounds, carbonyl- containing compounds, ketones, alcohols, terpenes, esters, and amines, or any combination thereof, correlates with the presence or absence of prostate cancer in the subject. In any embodiment, an electrical signal corresponding to a change in concentration from baseline of any of: β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5- dimethylbenzaldehyde, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, or any combination thereof, correlates with the presence of prostate cancer. In any embodiment, an electrical signal corresponding to a change in concentration from baseline of any of: thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2-furoate or any combination thereof, may correlate with the presence of aggressive prostate cancer in the subject.

[0012] In any embodiment, the apparatus may comprise an inlet piece operable to receive aa sample of volatile-organic-compounds emitted from a sample of the subject’s biofluid or receive a sample of the subject’s exhaled breath, the inlet piece operably coupled to a housing, the sensor array disposed inside the housing. In any embodiment, the apparatus may comprise an inlet piece operable to receive a sample of volatile-organic-compounds emitted from a sample of the subject’s urine, the inlet piece operably coupled to a housing, the sensor array disposed inside the housing.

[0013] In any embodiment, the subject may be a human. In any embodiment, the subject may be biologically male. In any embodiment, the subject may be a human who is diagnosed as having a high risk for prostate cancer. In any embodiment, the subject may be a human aged 40 to 85 years.Att’y Docket No.: IUIC-168

[0014] In any embodiment, the sensor subunits may comprise polyethylenimine-ether functionalized gold nanoparticles. In any embodiment, the sensor subunits may comprise tin (IV) oxide, tungsten oxide, or a combination of tin (IV) oxide and tungsten oxide. In any embodiment, the sensor subunits may comprise polyetherimide / carbon black composite. In any embodiment, the sensor subunits may comprise solid phase microextraction fibers coated with polyvinylidene fluoride – carbon black.

[0015] In another aspect, the present disclosure provides a system for determining whether a subject has prostate cancer from a sample of the subject’s exhaled breath or a sample of volatile-organic-compounds emitted from a sample of the subject’s biofluid, the system comprising: a sample collection means for receiving the sample of the subject’s exhaled breath or the sample of volatile-organic-compounds emitted from a sample of the subject’s biofluid; inside the housing, a sensor array comprising at least four sensor subunits, each sensor subunit comprising a chemiresistive substrate, and each sensor subunit independently corresponding to a signal output channel; each of the at least four sensor subunits independently tuned to: selectively output a resolvable electronic signal in response to a change in concentration in the subject’s exhaled breath or biofluid from baseline of one any of certain functional groups and / or species which correlate with the presence or absence of prostate cancer; and wherein each of signal output channels is communicably coupled to a general purpose computer processor unit programmed to receive the resolvable electronic signals and determine whether the resolvable electronic signals correlate with the presence of prostate cancer in the subject. In any embodiment, the biofluid may be urine.

[0016] In any embodiment, the general purpose computer processor unit may be further programmed to receive the resolvable electronic signals and, if it is determined that the electronic signals correlate with the presence of prostate cancer in the subject, determine whether the resolvable electronic signals correlate with indolent prostate cancer or aggressive prostate cancer. In any embodiment, each of the at least four sensor subunits may be independently tuned to: selectively output a resolvable electronic signal in response to a change in concentration in the subject’s exhaled breath or biofluid from baseline of one any of: β- thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, or 2,3,4,5-tetramethyl-2-cyclopenten-1-one.In any embodiment of the system, the four sensor subunits may be, independently, selective for detecting four of the chemical species concentration changes selected from: change in concentration from baseline of β-thujone, change in concentration from baseline of dimethyl trisulfide, change in concentration from baseline of α-curcumene, change in concentration from baseline of diethyl phthalate, changeAtt’y Docket No.: IUIC-168 in concentration from baseline of 2,5-dimethylbenzaldehyde, and change in concentration from baseline of 2,3,4,5-tetramethyl-2-cyclopenten-1-one. In any embodiment of the system, the four sensor subunits may be, independently, selective for detecting four of the chemical species concentration changes selected from: change in concentration from baseline of thymol, change in concentration from baseline of 6-amyl-α-pyrone, change in concentration from baseline of 2,3,4,5-tetramethyl-2-cyclopenten-1-one, change in concentration from baseline of p-cresol, change in concentration from baseline of 2-methylpropanoic propanoic anhydride, change in concentration from baseline of methyl salicylate, and change in concentration from baseline of allyl-2-furoate. In any embodiment, the change in concentration may be an increase in concentration.

[0017] In any embodiment of the system, the chemiresistive substrate may comprise polyethylenimine-ether functionalized gold nanoparticles. In any embodiment, the chemiresistive substrate may comprise tin (IV) oxide, tungsten oxide, or a combination of tin (IV) oxide and tungsten oxide. In any embodiment, the chemiresistive substrate may comprise polyetherimide / carbon black composite. In any embodiment, the chemiresistive substrate may comprise solid phase microextraction fibers coated with polyvinylidene fluoride – carbon black.

[0018] In still another aspect, the present disclosure provides a sensor array comprising: at least a first sensor subunit, a second sensor subunit, and a third sensor subunit, each sensor subunit comprising a heat-treated chemiresistive polyetherimide / carbon black composite substrate; the first sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 291.1 eV; the second sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 531.4 eV; and the third sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 284.7 eV.

[0019] In still another aspect, the present disclosure provides a method of detecting a prostate cancer status of a subject in need thereof, the method comprising: contacting volatile-organic- compounds emitted from a sample of the subject’s urine with a sensor array; the sensor array comprising two or more sensor subunits, each sensor subunit operable to detect at least one volatile-organic-compound biomarker in the VOCs emitted from the urine sample such that each sensor subunit is selectively chemiresistive to a chemical functional group corresponding to a volatile-organic-compound biomarker correlated with a presence of prostate cancer in the subject; wherein contacting the sensor subunit with a volatile-organic-compound biomarkerAtt’y Docket No.: IUIC-168 having a functional group for which the sensor is selectively chemiresistive causes the sensor subunit to generate an output signal; relaying output signals generated by the sensor subunits of the sensor array to a computation module; the computation module correlating the electronic signals to the presence or absence of prostate cancer in the subject.

[0020] In any embodiment of the method, the volatile-organic-compound biomarkers may be selected from any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5- dimethylbenzaldehyde, and 2,3,4,5-tetramethyl-2-cyclopenten-1-one. In any embodiment of the method, the volatile-organic-compound biomarkers may be selected from any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2-furoate.

[0021] In any embodiment of the method, the subject may be human. In any embodiment of the method, the subject may be biologically male. In any embodiment of the method, the subject may be age 40–85.

[0022] In another aspect, the present disclosure provides a method of detecting a prostate cancer status of a subject in need thereof, the method comprising: introducing volatile- organic-compound biomarkers emitted from a biofluid sample into a GC-MS device; relaying output signals generated by the GC-MS device to a computation module; the computation module correlating the output signals to the presence or absence of prostate cancer in the subject. In any embodiment, the biofluid may be urine.

[0023] In embodiments, the volatile-organic-compound biomarkers are selected from any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, and 2,3,4,5-tetramethyl-2-cyclopenten-1-one. In embodiments, the volatile-organic-compound biomarkers are selected from any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2- cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2-furoate.

[0024] In any embodiment, the subject may be human. In any embodiment, the subject may be biologically male. In any embodiment, the subject may be age 40–85.

[0025] In an embodiment, the GC-MS device is a tabletop device. In another embodiment, the GC-MS device is a portable device.

[0026] In another aspect, the present disclosure provides a device for detecting cancer in a subject cancer in a subject from volatile-organic-compound biomarkers present in the subject’s biofluid, the apparatus comprising: a gas chromatography (GC) system, mass spectrometry (MS) system, or gas chromatography – mass spectrometry (GC-MS) system configured to receive volatile organic compounds emitted from a sample of the subject’s biofluid; and aAtt’y Docket No.: IUIC-168 computation module programmed to interpret a signal output as corresponding with a presence or absence of cancer in the subject. In any embodiment, the computation module may be further programmed to interpret the signal output as corresponding with whether any cancer present in the subject is aggressive cancer or indolent cancer. In an embodiment, the device comprises a GC-MS system.

[0027] In any embodiment of the device, the biofluid may be urine. In any embodiment, the subject is biologically male.

[0028] In an embodiment, the GC-MS system is a tabletop system. In another embodiment, the GC-MS system is a portable system.

[0029] In any embodiment, the cancer may be prostate cancer.

[0030] In any embodiment, the computation module may be programmed to interpret a signal corresponding with a change in a baseline concentration of any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, or 2,3,4,5-tetramethyl- 2-cyclopenten-1-one, or any combination thereof, as corresponding with the presence of cancer. In any embodiment, the computation module may be programmed to interpret a signal corresponding with a change in a baseline concentration of any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, allyl-2-furoate, or any combination thereof, as corresponding with aggressive cancer.

[0031] In a further aspect, the present disclosure provides an electronic communication system comprising: any embodiment of the apparatus, or system, or sensor array, or device described herein communicatively coupled with a computing device having a display and installed with software programmed to receive data from the apparatus, system, sensor, or device and displaying data on the display. In an embodiment, the software may be further programmed to compute and display diagnostic information which assists an end user in monitoring, preventing, and / or treating cancer.

[0032] In another aspect, the present disclosure provides a method of monitoring a prostate cancer status in a subject in need thereof, the method comprising: performing any embodiment described herein of the method of detecting a prostate cancer status of a subject in need thereof at a first timepoint to record a first reading; and performing any embodiment described herein of the method of detecting a prostate cancer status of a subject in need thereof at a second timepoint to record a second reading.Att’y Docket No.: IUIC-168 BRIEF DESCRIPTION OF THE FIGURES

[0033] Some embodiments of the invention are herein described, by way of example only, with reference to the accompanying drawings. With specific reference now to the drawings in detail, it is stressed that the particulars shown are by way of example and for purposes of illustrative discussion of embodiments of the invention. In this regard, the description taken with the drawings makes apparent to those skilled in the art how embodiments of the invention may be practiced.

[0034] FIGs. 1A–1D are microscopy photographs of exemplary interdigitated electrodes shown at four different progressively more magnified scales. Scale bar in FIG. 1A is 10 millimeters; scale bar in FIG. 1B is 500 µm; scale bar in FIG. 1C is 150 µm; and scale bar in FIG. 1D is 50 µm. The gaps between the pictured electrodes is 8 µm. Each sensor array has four individual sensor subunits.

[0035] FIGs. 2A–2C are magnified photographs of exemplary sensor arrays of the present disclosure. FIG.2A shows an exemplary interdigitated electrode. FIG.2B shows an exemplary interdigitated electrode having ether functionalized gold nanoparticle chemiresistive substrate drop cast onto the electrodes. FIG. 2C shows an exemplary interdigitated electrode having polyethylenimine-ether electrospun onto the ether functionalized gold nanoparticles.

[0036] FIG.3 is a schematic of an exemplary experimental setup for testing of volatile-organic- compound biomarker sensors of the present disclosure.

[0037] FIG.4 shows exemplary workup parameters for an apparatus of the present disclosure.

[0038] FIG. 5 shows an exemplary perspective view of a schematic of an apparatus of the present disclosure.

[0039] FIGs. 6A–6B show exemplary photographs of a sensor array of the present disclosure communicatively coupled to a microprocessor, FIG.6A, and a housing which may house such sensor array, FIG.6B.

[0040] FIG. 7 is another exemplary perspective view of a schematic of an apparatus of the present disclosure.

[0041] FIGs.8A–8B are hierarchical clustergrams for training, FIG.8A, and testing, FIG.8B, data sets with 40 volatile-organic-compound biomarkers indicative of hypoglycemic state in a subject.

[0042] FIGs.9A–9C are graphs showing univariate and multivariate chemometric analysis of mouse urine for prostate cancer VOC biomarker identitication. FIG. 9A is a volcano plot interpolating statistical significance as a function of log2-fold change shows more VOCs were upregulated by prostate cancer. FIG. 9B is a hierarchical heatmap of the VOCs differentiallyAtt’y Docket No.: IUIC-168 expressed shows that VOCs have higher interclass variation. FIG.9C is a principal component analysis (PCA) of the top five VOCs completely distinguishes healthy controls from tumor- bearing mice with 100% accuracy.

[0043] FIGs. 10A–10B are graphs showing univariate chemometric analysis of human urine for VOC biomarker identification regarding any grade or stage of prostate cancer. FIG.10A is a volcano plot showing that more VOCs are upregulated by prostate cancer in human urine relative to those that are downregulated, and FIG.10B shows a hierarchical heatmap of VOCs identified with a p-value of < 0.05.

[0044] FIGs.11A–11D are graphs showing supervised multivariate of human urine to identify a biosignature of VOCs that could classify prostate cancer through LDA. (a) LD 1 scores for a model of six urinary volatiles that separates any grade of prostate cancer with (b) ROC AUC equal to 0.76 in the training data set and 0.75 in the cross-validated data set. (c) LD 1 scores for an alternative model of seven urinary VOCs that could distinguish prostate cancer with (d) ROC AUC equal to 0.80 in the training data and 0.75 in the cross-validated data set.

[0045] FIGs. 12A–12B are graphs showing univariate chemometric analysis of VOCs in human urine to distinguish aggressive grades of prostate cancer. (a) The volcano plot shows that more VOCs are downregulated by aggressive prostate cancer in human urine, and (b) the hierarchical heatmap of VOCs identified with a p-value < 0.05 shows that the analytes have high variation regardless of sample class.

[0046] FIGs.13A–13E are graphs showing supervised multivariate analysis of human urine to identify a biosignature of VOCs that could classify aggressive prostate cancer through LDA. (a) LD 1 scores for a model of six urinary volatiles that could classify aggressive tumors with (b) ROC AUC equal to 0.83 in the training data set and 0.82 in the cross-validated data set. (c) one-dimensional LDA plot for training and cross-validation to stratify indolent and aggressive prostate cancers using seven VOCs. (d) LD 1 scores for an alternative model of seven urinary VOCs that could stratify aggressive prostate cancer with (e) ROC AUC equal to 0.89 in the training data and 0.87 in the cross-validated data set.

[0047] FIGs.14A–14B are graphs showing PCA of VOCs in human urine that were identified with p‐value < 0.05 when comparing men with and without prostate cancer. (a) Two‐ dimensional PCA plot explains 27.2% of the variation in the data but shows a high degree of overlap between the sample classes of interest. (b) One‐dimensional PCA plot accounts for 15.3% of the variation in the data and shows greater statistical significance relative to any individual VOC.Att’y Docket No.: IUIC-168

[0048] FIGs.15A–15B are graphs showing PCA of VOCs in human urine that were identified with p‐value < 0.05 when comparing men with aggressive prostate cancer to those with indolent grades and negative biopsy results. (a) Two‐dimensional PCA plot explains 37.7% of the variation in the data but shows overlap and limited separation of aggressive prostate cancers from indolent and no cancer. (b) One‐dimensional PCA plot accounts for 20.2% of the variation in the data and shows statistical significance when stratifying aggressive prostate cancer.

[0049] FIGs. 16A–16B are stacked bar graphs showing Functional group frequency and (b) functional group frequency ratio of urinary VOCs identified to be differentially expressed (p- value < 0.05) by prostate cancer in the mouse model and in humans. The same is also displayed for urinary VOCs in humans that were dysregulated by aggressive prostate cancer.

[0050] FIGs. 17A–17B are principal component analyses (PCAs) for an experiment using principles of the apparatus, systems, and methods of the present disclosure, demonstrating separation between hypoglycemic vs. normal subjects based on VOCs in their exhaled breath, for experimental Cohort 1, FIG. 17A, and Cohort 2, FIG. 17B, with 40 volatile-organic- compound biomarkers indicative of hypoglycemic state in a subject having a p-value < 0.2.

[0051] FIGs.18A–18B are PCAs using a subset of six volatile-organic-compound biomarkers, which can distinguish between classes (hypoglycemic vs. normal) for experimental Cohort 1, FIG.18A, and Cohort 2, FIG.18B, using linear discriminant analysis (LDA).

[0052] FIG. 19 is a LDA result showing the same subset of six volatile-organic-compound biomarkers could distinguish between hypoglycemic state and normal state with an area under the curve (AUC) > 0.9 for both Cohort 1 and Cohort 2.

[0053] FIGs. 20A–20B show selectivity optimization results for candidate volatile-organic- compound biomarkers. Each sensor test setup was tested with heating element at 1400 mV, 2000 mV, and 2500 mV. FIG. 20A shows sensitivity of test chip 1 (SnO2, left column) and chip 2 (WO3, right column) on volatile-organic-compound biomarker VOC 1 and VOC 2. FIG. 20B shows sensitivity of test chip 1 (SnO2, left column) and chip 2 (WO3, right column) on VOC 3 and VOC 4.

[0054] FIG.21 shows sensor selectivity for each of four test VOCs isopropanol, octyl acetate, 2,6-dimethylnonane, and dimethyl phthalate on metal oxide sensors are 1400 mV and 85% relative humidity (upper left), 2000 mV and 85% relative humidity (upper right), and 2500 mV and 85% relative humidity.

[0055] FIG. 22 shows calibration curves relating current change to concentration in parts per million (ppm) for a WO3substrate sensor. Sensors were calibrated at 85% relative humidity at each of 1400 mV heater, 2000 mV heater, and 2500 mV heater.Att’y Docket No.: IUIC-168

[0056] FIGs. 23A–23B show selectivity toward ten VOCs related to hypoglycemia and other conditions. C1 refers to SnO2substrate sensors. C2 refers to WO3substrate sensors. FIG.23A shows selectivity for each of VOC 1 through VOC 10. FIG.23B shows the same results with VOC 4 removed and with more expanded Y-axis, for readability.

[0057] FIG.24 is an exemplary Wheatstone bridge circuit schematic for chemiresistive sensors of the present disclosure.

[0058] FIG.25 is another schematic of an exemplary experimental setup for testing of volatile- organic-compound biomarker sensors of the present disclosure.

[0059] FIG. 26 shows response of polyetherimide / carbon black composite sensors to 50 ppm of nonanal (also sometimes called nonaldehyde, or pelargonaldehyde) for three cycles. The sensors average resistance of 19 kΩ and thickness of 15 µm.

[0060] FIG.27 shows results of a polyetherimide / carbon black composite sensor degradation study for nonanal.

[0061] FIG. 28 shows results of impact of relative humidity on polyetherimide / carbon black composite sensor response to nonanal.

[0062] FIG. 29 is a bar graph showing polyetherimide / carbon black composite sensor resistance response per part-per-million of selected VOCs (X-axis), calculated from calibration curve.

[0063] FIG. 30 is a bar graph showing polyetherimide / carbon black composite sensor resistance response per part-per-million of selected aldehyde VOCs (X-axis), calculated from calibration curve.

[0064] FIG. 31 is a principal component analysis of polyetherimide / carbon black composite sensor response to nonanal and other VOCs.

[0065] FIG. 32 is shows average response peaks for exposure of VOC 10 to a polyethylenimine-ether functionalized gold nanoparticle sensor at relative humidity levels 0%, 45%, and 85%.

[0066] FIG. 33 shows performance of a polyethylenimine-ether functionalized gold nanoparticle sensor at different VOC concentrations and relative humidity levels.

[0067] FIG. 34 shows a polyetherimide / carbon black composite sensor calibration curve for nonanal from concentration of 1 ppm to concentration of 80 ppm. The sensors average resistance of 19 kΩ and thickness of 15 µm.

[0068] FIG. 35 shows a schematic of dip coating and electrospinning composites of polyvinylidene fluoride-carbon black (PVDF-CB) onto solid phase microextraction fibers.Att’y Docket No.: IUIC-168

[0069] FIG.36 shows an exemplary GC-MS chromatogram of urinary VOC biomarkers from the standard after being extracted with a PVDF-CB SPME fiber, which was fabricated through electrospinning for a total of five minutes.

[0070] FIGs.37A–37D are bar charts showing integrated signal sensitivity for four VOCs after exposing SPME fibers electrospun with PVDF-CB for 5-, 10- and 15 minutes to the human urine standard demonstrates that 10 minutes is the optimal electrospinning time for increased VOC sensitivity. FIG. 37A shows data for 2,4-dimethylbenzaldehyde; FIG. 37B shows data for carvone; FIG. 37C shows data for p-menth-1-en-3-one; and FIG.37D shows data for 2,4- di-tert-butylphenol.

[0071] FIGs.38A–38B show bar charts comparing total number of VOCs detected (FIG.38A) and total integrated GC-MS signal strength (FIG. 38B) when comparing dip-coated and electrospun PVDF-CB SPME fibers. Electrospinning PVDF-CB onto SPME fibers significantly increases the total integrated signal.

[0072] FIG. 39A–39D shows bar charts for 2,4-dimethylbenzaldehyde (FIG. 39A), carvone (FIG.39B), p-menth-1-en-3-one (FIG.39C), and 2,4- di-tert-butylphenol (FIG.39D) show that electrospinning PVDF-CB onto SPME fibers can increase extraction efficiency by factors ranging from 1.5x to 3.4x.

[0073] FIG.40 shows overlaid interval plots for other VOCs that had lower abundance or were identified in previous studies show that electrospinning PVDF-CB onto SPME fibers increases VOC sensitivity.

[0074] FIG. 41 shows a hierarchical heatmap of urinary VOCs detected using dip coated and electrospun PVDF-CB SPME fibers indicate that electrospinning increases the signal of many VOCs expressed in the standard urine sample.

[0075] FIG. 42 shows overlaid box and whisker plots illustrating relative standard deviation values for urinary VOCs when extracted and analyzed using the dip coated and electrospun SPME fibers suggest the electrospun SPME fiber has higher reproducibility.

[0076] FIG.43 shows overlaid Fourier-transform infrared spectroscopy FTIR spectra of PVDF and PVDF composited with CB. Both PVDF and PVDF-CB show similar intensities of FTIR peaks that correspond to the different isomeric phases of the polymer (β phase peaks: 842 cm-1, 511 cm-1; γ phase peak: 1231 cm-1; α phase peak: 880 cm-1).

[0077] FIG.44 shows field emission scanning electron microscopy (FESEM) images of SPME fibers electrospun with the optimized conditions for different durations shows that increasing electrospinning time directly increases fiber coverage, but when SPME fibers are electrospun for 15 minutes the fibers themselves appear to be damaged.Att’y Docket No.: IUIC-168

[0078] FIG.45 shows FE-SEM images of SPME fibers dip coated and electrospun with PVDF- CB shows that electrospun fibers had a diameter of 1.5 µm and increased surface area relative to dip coated SPME fibers.

[0079] FIG.46 is a graph showing the static contact angle measurements for PVDF and PVDF- CB fabricated through dip coating and electrospinning. The addition of CB to PVDF and the electrospinning process itself increases the hydrophobicity of the sensing layer. Static contact angle provides a measure of hydrophobicity.

[0080] FIGs.47A–47B are bar charts GC-MS signals for VOCs with a relatively low molecular weight demonstrate that SPME fibers have higher extraction efficiency toward small analytes (FIG. 47A) and GC-MS signals of VOCs with a slightly higher molecular weight shows that the SPME arrow has slightly higher sensitivity for other analytes (FIG.47B).

[0081] FIGs.48A–48B show a comparison of SPME arrows and fibers, showing no significant differences in performance regarding the number of VOCs detected (FIG. 48A) or the total integrated signal when utilized to extract a standard urine solution of VOCs (FIG.48B).

[0082] FIGs. 49A–49C show certain properties of polyetherimide (PEI) / carbon black (CB) film. FIG. 49A is an FE-SEM image of a surface of PEI / CB film. FIG.49B is a cross-section image of PEI / CB, with scale bar = 10µm. FIG. 49C shows a profilometer analysis of PEI / CB film. A method of heat treating the spin-case PEI / CB at 200 ºC was developed to reduce the composite in the presence of 1-methyl-2-pyrrolidone (NMP) and engineer nanomaterials with high sensitivity and selectivity. The PEI / CB composite was synthesizes in NMP and stirred at 80ºC for 72 hours resulting in the formation of a uniform slurry paste. The slurry paste was spun over the interdigitated electrodes (IDEs) at 3500rpm for 90 seconds via a spin coater. Next, the film was heated on a hotplate at 200ºC for two hours.

[0083] FIG. 50 is a total ion chromatogram of urinary VOCs, analyzed using a portable gas chromatography – mass spectrometry (GC-MS) system.

[0084] FIG.51 is a photograph of an exemplary nanosensor apparatus with an inset illustration of two chips containing tin oxide and / or tungsten oxide sensors.

[0085] FIG. 52 is an exemplary diagram of a portable nanosensor array device, with a digital display, battery, microcontroller chip, and biofluid sample reservoir.

[0086] FIG.53 is an exemplary image of a Teledyne FLIR Griffin™ portable GC-MS device, which may be used for carrying out methods of the present disclosure.

[0087] FIG. 54 is an exemplary flow diagram showing a photolithography method of fabricating interdigitated electrodes (IDEs) for nanosensor arrays for devices of the present disclosure.Att’y Docket No.: IUIC-168

[0088] FIG.55 is an exemplary diagram of SPME fiber manufacture and properties.

[0089] FIG.56 is a graph showing response of a tin oxide nanosensor when tested using a urine standard for various exposure durations (1 minute, 5 minutes, and 10 minutes).

[0090] FIG. 57 depicts a series of exemplary screen displays for a smart device application (SDA) user interface, which may display relevant results and diagnostic information, from sensors according to systems and methods of the present disclosure, to a medical professional and / or end user.

[0091] FIG. 58 is an exemplary illustration of a sensor chip used in nanosensor arrays of the present disclosure.

[0092] FIG.59 is an exemplary diagram of nanosensor mounts used in nanosensor array chips. DETAILED DESCRIPTION

[0093] The present disclosure may be further understood by reference to the following detailed description.

[0094] The terms “comprises”, “comprising”, “includes”, “including”, “having” and their conjugates mean “including but not limited to”.

[0095] The term “consisting of” means “including and limited to”.

[0096] The term “consisting essentially of” means that the composition, method or structure may include additional ingredients, steps and / or parts, but only if the additional ingredients, steps and / or parts do not materially alter the basic and novel characteristics of the claimed composition, method or structure.

[0097] As used herein, the singular form “a”, “an” and “the” include plural references unless the context clearly dictates otherwise. For example, the term “a compound” or “at least one compound” may include a plurality of compounds, including mixtures thereof.

[0098] Throughout this application, various embodiments of this invention may be presented in a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the invention. Accordingly, the description of a range should be considered to have specifically disclosed all the possible subranges as well as individual numerical values within that range. For example, description of a range such as from 1 to 6 should be considered to have specifically disclosed subranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6 etc., as well as individual numbers within that range, for example, 1, 2, 3, 4, 5, and 6. This applies regardless of the breadth of the range.Att’y Docket No.: IUIC-168

[0099] Whenever a numerical range is indicated herein, it is meant to include any cited numeral (fractional or integral) within the indicated range. The phrases “ranging / ranges between” a first indicate number and a second indicate number and “ranging / ranges from” a first indicate number “to” a second indicate number are used herein interchangeably and are meant to include the first and second indicated numbers and all the fractional and integral numerals therebetween.

[0100] As used herein the term “method” refers to manners, means, techniques and procedures for accomplishing a given task including, but not limited to, those manners, means, techniques and procedures either known to, or readily developed from known manners, means, techniques and procedures by practitioners of the chemical, pharmacological, biological, biochemical and medical arts.

[0101] As used herein, the term “treating” includes abrogating, substantially inhibiting, slowing or reversing the progression of a condition, substantially ameliorating clinical or aesthetical symptoms of a condition or substantially preventing the appearance of clinical or aesthetical symptoms of a condition.

[0102] It is appreciated that certain features of the invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable subcombination or as suitable in any other described embodiment of the invention. Certain features described in the context of various embodiments are not to be considered essential features of those embodiments, unless the embodiment is inoperative without those elements.

[0103] Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and broad scope of the appended claims.

[0104] All publications, patents and patent applications mentioned in this specification are herein incorporated in their entirety by reference into the specification, to the same extent as if each individual publication, patent or patent application was specifically and individually indicated to be incorporated herein by reference. In addition, citation or identification of any reference in this application shall not be construed as an admission that such reference is available as prior art to the present invention. To the extent that section headings are used, they should not be construed as necessarily limiting. In addition, any priority document(s) of this application is / are hereby incorporated herein by reference in its / their entirety.Att’y Docket No.: IUIC-168

[0105] Unless otherwise defined, all technical and / or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the invention pertains. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of embodiments of the invention, exemplary methods and / or materials are described below. In case of conflict, the patent specification, including definitions, will control. In addition, the materials, methods, and examples are illustrative only and are not intended to be necessarily limiting.

[0106] Implementation of the method and / or system of embodiments of the invention can involve performing or completing selected tasks manually, automatically, or a combination thereof. Moreover, according to actual instrumentation and equipment of embodiments of the method and / or system of the invention, several selected tasks could be implemented by hardware, by software or by firmware or by a combination thereof using an operating system.

[0107] For example, hardware for performing selected tasks according to embodiments of the invention could be implemented as a chip or a circuit. As software, selected tasks according to embodiments of the invention could be implemented as a plurality of software instructions being executed by a computer using any suitable operating system. In an exemplary embodiment of the invention, one or more tasks according to exemplary embodiments of method and / or system as described herein are performed by a data processor, such as a computing platform for executing a plurality of instructions. Optionally, the data processor includes a volatile memory for storing instructions and / or data and / or a non-volatile storage, for example, a magnetic hard-disk and / or removable media, for storing instructions and / or data. Optionally, a network connection is provided as well. A display and / or a user input device such as a keyboard or mouse are optionally provided as well.

[0108] Researchers have investigated an innovative approach to detect, treat, and alert patients of their current health state in a non-invasive manner. Volatile-organic-compound biomarkers (VOCs) are byproducts of metabolic pathways, which may be altered by disease initiation and progression. VOCs originate from a subject’s body and may carry information that can be expressed in lung alveolar breath, or in biofluids such as blood, urine, sweat, saliva, and others. VOCs evaporate at room temperature and standard pressure.

[0109] Research has shown links between concentration of various volatile-organic-compound biomarkers (VOCs), from example from a sample of a subject’s urine, and disease states in a subject.Att’y Docket No.: IUIC-168

[0110] Urine analysis through gas chromatography / mass spectroscopy (GC / MS) has found correlations between VOCs and cancer. Trained canines can detect VOCs from a subject and can differentiate a patient suffering from cancer from a healthy negative-control patient.

[0111] Additionally, the present inventors studied the relationship between blood glucose levels and the presence of acetone in their breath. It was discovered that the presence of acetone in type 1 diabetes (T1D) patients is 40× higher than in a normal patient. This means that having a device that could detect the high levels of acetone in the breath of patients could help monitor the health of T1D patients. Other researchers studied the relationship between isoprene and hypoglycemia. Similar to acetone, isoprene levels increase when the patient experience hypoglycemic events. The physiological mechanism behind the increase of isoprene remains unknown at the time of the present disclosure, but isoprene has been reported as a VOC biomarker for other diseases. However, while acetone and isoprene have been found to be VOC biomarkers, they lack disease specificity. Therefore, the combination of acetone and isoprene along with other, more specific, VOC biomarkers is needed to create a panel or signature to detect a specific diseases with higher accuracy and specificity.

[0112] For certain examples, some alcohol compounds have been shown to be correlated with disease state. Nutrition, husbandry and environmental factors influences VOC composition due to growth and metabolism effects in our body that result in accumulation of metabolites that are exhaled in the breath. Alcohols are easily dissolved in water ,therefore itis found in body tissues and fluids and are rapidly absorbed in the blood. A significant decrement in concentration of 1- octanol was noticed between colorectal cancer patients and the healthy person. Elevated levels of 1-octanol was reported in women suffering from breast cancer. Proton transfer reaction mass spectroscopy (PTRMS) and solid phase microextraction with gas chromatography and mass spectroscopy (SPM-GCMS) was used to analyze the breath samples from lung cancer patients and methanol was reported as one of the targeting biomarker having p=0.01. A gold- nanoparticle-based flexible sensor was fabricated to detect ethanol as one of the target biomarkers from women suffering from ovarian cancer.

[0113] For another set of examples, formaldehyde was also reported as one of the target VOCs from breath in lung cancer patients and from urine headspace in prostate cancer patients. Smith et al. reported higher levels of acetaldehyde in the range of 100ppb if moderately size lung cancer tumors are present. Researchers worked specifically on aldehydes and mentioned propanal, butanal, pentanal, hexanal, heptanal, octanal, and nonanal and made the comparison between the exhaled level of VOCs between non-small-cell lung cancer (NSCLC) and the control group. Formaldehyde and pentanal were reported as one of the potential biomarkersAtt’y Docket No.: IUIC-168 among 21 VOCs for lung cancer patients. Higher incidence of acetaldehyde and pentanal were reported for patients suffering from prostate cancer. Heptanal, hexanal, octanal and nonanal have been reported with significant concentration difference between breast cancer patients and healthy people. When compared with healthy patients nonanal and decanal levels in urinary VOCs is elevated in renal cell carcinoma patients.

[0114] Following these principles, the investigators of the present disclosure found that change of concentration of certain VOCs in a subject’s urine is correlated (positively or negatively) with high specificity with the subject having prostate cancer. It was further found that properties of any prostate cancer that is present, such as whether the cancer is indolent or aggressive, could be elucidated from VOCs measured in the subject’s urine.

[0115] Accordingly, the present disclosure relates to apparatus, systems, and methods for detecting cancer in a subject using volatile-organic-chemical biomarkers present in a sample of the subject’s biofluids or a sample of the subject’s exhaled breath.

[0116] In an aspect, the present disclosure provides an apparatus detecting cancer in a subject from volatile-organic-compound biomarkers present in the subject’s biofluid, the apparatus comprising: a sensor array of two or more sensor subunits, each sensor subunit operable to detect at least one volatile-organic-compound biomarker in a sample of VOCs emitted from the subject’s biofluid wherein each sensor subunit is selectively chemiresistive in response to exposure to a chemical of a functional group selected from: noncyclic hydrocarbon, unconjugated cyclic hydrocarbon, alcohol, aldehyde, amide, aromatic, carbonyl, carboxylic acid, ester, ether, ketone, or terpene; and wherein contacting the sensor subunit with a volatile- organic-compound biomarker having a functional group for which the sensor is selectively chemiresistive causes the sensor subunit to generate an electronic signal; wherein each sensor subunit corresponds independently to a signal output channel, each signal output channel operably connected to a computation module for receiving each of the electronic signals; and wherein the computation module is programmed to interpret an aggregate of electronic signals as corresponding to a subject cancer state. In any embodiment, the computation module may be further programmed to interpret an aggregate of electronic signals as corresponding to a cancer aggressiveness state. In any embodiment, the cancer may be prostate cancer. In any embodiment, the biofluid may be urine.

[0117] In any embodiment of the apparatus, the apparatus may comprise a sensor subunit selectively chemiresistive for an alcohol. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for an unconjugated cyclic organic compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for anAtt’y Docket No.: IUIC-168 aromatic compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for a carbonyl-bearing compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for an ketone compound. In any embodiment, the apparatus may comprise a sensor subunit selectively chemiresistive for a terpene.

[0118] In any embodiment of the apparatus, the apparatus may comprise a first sensor subunit selectively chemiresistive for unconjugated cyclic compounds, a second sensor subunit selectively chemiresistive for carbonyl-bearing compounds, and a third sensor subunit selectively chemiresistive for terpenes.

[0119] In any embodiment, each sensor subunit may be operable to detect a change in concentration above a baseline threshold one or more of: β-thujone, dimethyl trisulfide, α- curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, 2,3,4,5-tetramethyl-2-cyclopenten- 1-one. In any embodiment, each sensor subunit may be operable to detect a changein concentration above a baseline threshold one or more of: thymol, 6-amyl-α-pyrone, 2,3,4,5- tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, or allyl-2-furoate.

[0120] In any embodiment, an electrical signal corresponding to a change in concentration from baseline of any of: unconjugated cyclic compounds, aromatic compounds, carbonyl- containing compounds, ketones, alcohols, terpenes, esters, and amines, or any combination thereof, correlates with the presence or absence of prostate cancer in the subject. In any embodiment, an electrical signal corresponding to a change in concentration from baseline of any of: β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5- dimethylbenzaldehyde, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, or any combination thereof, correlates with the presence of prostate cancer. In any embodiment, an electrical signal corresponding to a change in concentration from baseline of any of: thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2-furoate or any combination thereof, may correlate with the presence of aggressive prostate cancer in the subject.

[0121] In any embodiment, the apparatus may comprise an inlet piece operable to receive aa sample of volatile-organic-compounds emitted from a sample of the subject’s biofluid or receive a sample of the subject’s exhaled breath, the inlet piece operably coupled to a housing, the sensor array disposed inside the housing. In any embodiment, the apparatus may comprise an inlet piece operable to receive a sample of volatile-organic-compounds emitted from aAtt’y Docket No.: IUIC-168 sample of the subject’s urine, the inlet piece operably coupled to a housing, the sensor array disposed inside the housing.

[0122] In any embodiment, the subject may be a human. In any embodiment, the subject may be biologically male. In any embodiment, the subject may be a human who is diagnosed as having a high risk for prostate cancer. In any embodiment, the subject may be a human aged 40 to 85 years.

[0123] In any embodiment, the sensor subunits may comprise polyethylenimine-ether functionalized gold nanoparticles. In any embodiment, the sensor subunits may comprise tin (IV) oxide, tungsten oxide, or a combination of tin (IV) oxide and tungsten oxide. In any embodiment, the sensor subunits may comprise polyetherimide / carbon black composite. In any embodiment, the sensor subunits may comprise solid phase microextraction fibers coated with polyvinylidene fluoride – carbon black.

[0124] In another aspect, the present disclosure provides a system for determining whether a subject has prostate cancer from a sample of the subject’s exhaled breath or a sample of volatile-organic-compounds emitted from a sample of the subject’s biofluid, the system comprising: a sample collection means for receiving the sample of the subject’s exhaled breath or the sample of volatile-organic-compounds emitted from a sample of the subject’s biofluid; inside the housing, a sensor array comprising at least four sensor subunits, each sensor subunit comprising a chemiresistive substrate, and each sensor subunit independently corresponding to a signal output channel; each of the at least four sensor subunits independently tuned to: selectively output a resolvable electronic signal in response to a change in concentration in the subject’s exhaled breath or biofluid from baseline of one any of certain functional groups and / or species which correlate with the presence or absence of prostate cancer; and wherein each of signal output channels is communicably coupled to a general purpose computer processor unit programmed to receive the resolvable electronic signals and determine whether the resolvable electronic signals correlate with the presence of prostate cancer in the subject. In any embodiment, the biofluid may be urine. In any embodiment, the sample collection means may be an opening, e.g., an inlet, valve, or the like, for receiving emitted volatile organic compounds. In any embodiment, the sample collection means may be a fluid reservoir. In certain embodiments, the fluid reservoir may detachably coupled to a housing. (See, e.g., FIG. 52.)

[0125] In any embodiment, the general purpose computer processor unit may be further programmed to receive the resolvable electronic signals and, if it is determined that the electronic signals correlate with the presence of prostate cancer in the subject, determineAtt’y Docket No.: IUIC-168 whether the resolvable electronic signals correlate with indolent prostate cancer or aggressive prostate cancer. In any embodiment, each of the at least four sensor subunits may be independently tuned to: selectively output a resolvable electronic signal in response to a change in concentration in the subject’s exhaled breath or biofluid from baseline of one any of: β- thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, or 2,3,4,5-tetramethyl-2-cyclopenten-1-one.In any embodiment of the system, the four sensor subunits may be, independently, selective for detecting four of the chemical species concentration changes selected from: change in concentration from baseline of β-thujone, change in concentration from baseline of dimethyl trisulfide, change in concentration from baseline of α-curcumene, change in concentration from baseline of diethyl phthalate, change in concentration from baseline of 2,5-dimethylbenzaldehyde, and change in concentration from baseline of 2,3,4,5-tetramethyl-2-cyclopenten-1-one. In any embodiment of the system, the four sensor subunits may be, independently, selective for detecting four of the chemical species concentration changes selected from: change in concentration from baseline of thymol, change in concentration from baseline of 6-amyl-α-pyrone, change in concentration from baseline of 2,3,4,5-tetramethyl-2-cyclopenten-1-one, change in concentration from baseline of p-cresol, change in concentration from baseline of 2-methylpropanoic propanoic anhydride, change in concentration from baseline of methyl salicylate, and change in concentration from baseline of allyl-2-furoate. In any embodiment, the change in concentration may be an increase in concentration.

[0126] In any embodiment of the system, the chemiresistive substrate may comprise polyethylenimine-ether functionalized gold nanoparticles. In any embodiment, the chemiresistive substrate may comprise tin (IV) oxide, tungsten oxide, or a combination of tin (IV) oxide and tungsten oxide. In any embodiment, the chemiresistive substrate may comprise polyetherimide / carbon black composite. In any embodiment, the chemiresistive substrate may comprise solid phase microextraction fibers coated with polyvinylidene fluoride – carbon black.

[0127] In still another aspect, the present disclosure provides a sensor array comprising: at least a first sensor subunit, a second sensor subunit, and a third sensor subunit, each sensor subunit comprising a heat-treated chemiresistive polyetherimide / carbon black composite substrate; the first sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 291.1 eV; the second sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 531.4 eV; and the third sensor subunit configured to selectively output anAtt’y Docket No.: IUIC-168 electrical signal when contacted with a chemical compound having a binding energy of 284.7 eV.

[0128] In still another aspect, the present disclosure provides a method of detecting a prostate cancer status of a subject in need thereof, the method comprising: contacting volatile-organic- compounds emitted from a sample of the subject’s urine with a sensor array; the sensor array comprising two or more sensor subunits, each sensor subunit operable to detect at least one volatile-organic-compound biomarker in the VOCs emitted from the urine sample such that each sensor subunit is selectively chemiresistive to a chemical functional group corresponding to a volatile-organic-compound biomarker correlated with a presence of prostate cancer in the subject; wherein contacting the sensor subunit with a volatile-organic-compound biomarker having a functional group for which the sensor is selectively chemiresistive causes the sensor subunit to generate an output signal; relaying output signals generated by the sensor subunits of the sensor array to a computation module; the computation module correlating the electronic signals to the presence or absence of prostate cancer in the subject.

[0129] In any embodiment of the method, the volatile-organic-compound biomarkers may be selected from any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5- dimethylbenzaldehyde, and 2,3,4,5-tetramethyl-2-cyclopenten-1-one. In any embodiment of the method, the volatile-organic-compound biomarkers may be selected from any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2-furoate.

[0130] In any embodiment of the method, the subject may be human. In any embodiment of the method, the subject may be biologically male. In any embodiment of the method, the subject may be age 40–85. For example, the subject may be age 40 years, 45 years, 50 years, 55 years, 60 years, 65 years, 70 years, 75 years, 80 years, or 85 years.

[0131] In another aspect, the present disclosure provides a method of monitoring a prostate cancer status in a subject in need thereof, the method comprising: performing any embodiment described herein of the method of detecting a prostate cancer status of a subject in need thereof at a first timepoint to record a first reading; and performing any embodiment described herein of the method of detecting a prostate cancer status of a subject in need thereof at a second timepoint to record a second reading. In any embodiment, the method of monitoring may further comprise repeating the steps. In any embodiment, the method of monitoring may further comprise performing any embodiment described herein of the method of detecting a prostate cancer status of a subject in need thereof at a third timepoint to record a third reading. In any embodiment, the method may comprise, e.g., a third, fourth, fifth, sixth, seventh, eighth, ninth,Att’y Docket No.: IUIC-168 tenth, or greater number of readings at a third, fourth, fifth, sixth, seventh, eighth, ninth, tenth, or greater number of timepoints.

[0132] In another aspect, the present disclosure provides a method of detecting a prostate cancer status of a subject in need thereof, the method comprising: introducing volatile-organic- compound biomarkers emitted from a biofluid sample into a GC-MS device; relaying output signals generated by the GC-MS device to a computation module; the computation module correlating the output signals to the presence or absence of prostate cancer in the subject. In any embodiment, the biofluid may be urine.

[0133] It should be understood that the method of detecting a prostate cancer status may be carried out using any suitable e-nose, GC-MS-based system, GC-based system, MS-based system, or combination thereof capable of identifying and distinguishing small quantities of VOCs in a sample. For non-limiting example, the method may be carried out using benchtop GC-MS or portable GC-MS; using hand-held gas sensor arrays (e.g., electronic noses, i.e., “e- noses”); and other technologies. For additional non-limiting examples, the method may be carried out using gas chromatography (GC) with a flame ionization detector (FID), thermal conductivity detector (TCD), photoionization detector (PID), or combination thereof. For further non-limiting examples, the method may be carried out using mass spectrometry (MS), such as direct injection MS, selected ion flow tube (SIFT), ion mobility, proton transfer reaction (PTR), or any combination thereof. Recent developments have permitted development of 2- dimensional GC systems (“GCxGC”), which it is readily understood may be employed with MS, FID, TCD, PID, or any combination thereof for carrying out the method of the present disclosure.

[0134] In embodiments, the volatile-organic-compound biomarkers are selected from any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, and 2,3,4,5-tetramethyl-2-cyclopenten-1-one. In embodiments, the volatile-organic-compound biomarkers are selected from any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2- cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2-furoate.

[0135] In any embodiment, the subject may be human. In any embodiment, the subject may be biologically male. In any embodiment, the subject may be age 40–85.

[0136] In an embodiment, the GC-MS device is a tabletop device. In another embodiment, the GC-MS device is a portable device. The portable device may be, e.g., a Teledyne FLIR Griffin™ portable GC-MS device.Att’y Docket No.: IUIC-168

[0137] In another aspect, the present disclosure provides a device for detecting cancer in a subject cancer in a subject from volatile-organic-compound biomarkers present in the subject’s biofluid, the apparatus comprising: a gas chromatography (GC) system, mass spectrometry (MS) system, or gas chromatography – mass spectrometry (GC-MS) system configured to receive volatile organic compounds emitted from a sample of the subject’s biofluid; and a computation module programmed to interpret a signal output as corresponding with a presence or absence of cancer in the subject. In any embodiment, the computation module may be further programmed to interpret the signal output as corresponding with whether any cancer present in the subject is aggressive cancer or indolent cancer. In an embodiment, the system is a GC-MS system.

[0138] In any embodiment of the device, the biofluid may be urine. In any embodiment, the subject is biologically male.

[0139] In an embodiment, the GC-MS system is a tabletop system. In another embodiment, the GC-MS system is a portable system. The portable system may be, e.g., a Teledyne FLIR Griffin™ portable GC-MS. It should be understood that the device may comprise any suitable e-nose, GC-MS-based system, GC-based system, MS-based system, or combination thereof capable of identifying and distinguishing small quantities of VOCs in a sample. For non- limiting example, the device may comprise hand-held gas sensor arrays (e.g., electronic noses); and other technologies. For additional non-limiting examples, the device may comprise gas chromatography (GC) with a flame ionization detector (FID), thermal conductivity detector (TCD), photoionization detector (PID), or combination thereof. For further non-limiting examples, the device may comprise mass spectrometry (MS), such as direct injection MS, selected ion flow tube (SIFT), ion mobility, proton transfer reaction (PTR), or any combination thereof. Recent developments have permitted development of 2-dimensional GC systems (“GCxGC”), which it is readily understood may be employed in the device with MS, FID, TCD, PID, or any combination thereof.

[0140] In any embodiment, the cancer may be prostate cancer.

[0141] In any embodiment, the computation module may be programmed to interpret a signal corresponding with a change in a baseline concentration of any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, or 2,3,4,5-tetramethyl- 2-cyclopenten-1-one, or any combination thereof, as corresponding with the presence of cancer. In any embodiment, the computation module may be programmed to interpret a signal corresponding with a change in a baseline concentration of any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride,Att’y Docket No.: IUIC-168 methyl salicylate, allyl-2-furoate, or any combination thereof, as corresponding with aggressive cancer.

[0142] In a further aspect, the present disclosure provides an electronic communication system comprising: any embodiment of the apparatus, or system, or sensor array, or device described herein communicatively coupled with a computing device having a display and installed with software programmed to receive data from the apparatus, system, sensor, or device and displaying data on the display. In an embodiment, the software may be further programmed to compute and display diagnostic information which assists an end user in monitoring, preventing, and / or treating cancer. EXAMPLES

[0143] The present disclosure may be further understood by reference to the following examples. Example 1. Detection of Prostate Cancer and Characteristics of Such Cancer in a Subject By Volatile Organic Compound Biomarkers Emitted from a From a Sample of the Subject’s Urine.

[0144] Current screening methods for prostate cancer include digital rectal exams (DREs), assessing risk factors including family history, and measuring prostate-specific antigen (PSA) levels in blood samples. However, these screening methods have limited sensitivity / specificity and cannot determine tumor aggressiveness. With these deficiencies, the results from PSA tests often lead to overtreatment and unnecessary biopsies. Confirmatory diagnostics through biopsy are utilized after prostate cancer screening for pathological grading. Many patients are diagnosed with low grades of prostate cancer and undergo active surveillance where they are routinely biopsied every 6–12 months. Therefore, it is of significant clinical interest to develop a noninvasive and accurate screening method to detect patients with prostate cancer and differentiate indolent from aggressive tumors. Researchers are seeking alternative methods to screen for prostate cancer through identifying biomarkers using various “omic” techniques. These include genomics, transcriptomics, proteomics, metabolomics, and lipidomics. One emerging urine-based assay is called SelectMDx, which monitors the expression of two strands of RNA to identify men who would benefit from a biopsy. The benefits of this assay include high diagnostic specificity for patients with advanced grades of prostate cancer (Gleason score (GS) ≥ 7 or International Society of Urological Pathology (ISUP) grade ≥ 3). However, additional biomarkers could complement and improve the ability of prostate cancer screening.Att’y Docket No.: IUIC-168

[0145] Among some of the most promising and important lines of research include implementing liquid biopsy, which has the ability to measure circulating tumor cells (CTCs), cell-free DNA, exosomal miRNA, and circulating DNA. These methods have been extensively studied for prostate cancer prognosis and diagnosis. One example of this is androgen-receptor splice variant 7 (AR-V7), phosphatase and tensin homolog (PTEN), and total CTC counts have been previously shown to be significantly altered by advanced prostate cancer and therefore may be able to provide useful prognostic information

[0020] . Another example is several strands of exosomal miRNA (miR-141-3p, miR-375, miR-21, miR-141, and miR-375) have been correlated with prostate cancer metastasis to the bone. Previous studies have also shown the outstanding capability of canines to detect prostate cancer by smelling volatile organic compounds (VOCs) in urine headspace. The results have demonstrated that canines can detect prostate cancer more accurately than any other current biological assay. For example, Taverna et al. showed that canines could identify prostate cancer with over 97% sensitivity and specificity. Taverna, G.; Tidu, L.; Grizzi, F.; Torri, V.; Mandressi, A.; Sardella, P.; La Torre, G.; Cocciolone, G.; Seveso, M.; Giusti, G.; et al. Olfactory system of highly trained dogs detects prostate cancer in urine samples. J. Urol.2015, 193, 1382–1387.

[0146] The results from canines have inspired researchers to identify VOC biomarkers emanating from urine samples. Headspace analysis coupled to gas chromatography-mass spectrometry (GC-MS) was established as the state of the art for VOC biomarker identification. In 2015, Khalid et al. identified a panel of four VOCs in urine headspace using solid phase micro-extraction (SPME) and GC-MS that could identify prostate cancer with 65% accuracy. Khalid, T.; Aggio, R.; White, P.; De Lacy Costello, B.; Persad, R.; Alkateb, H.; Jones, P.; Probert, C.S.; Ratcliffe, N. Urinary Volatile Organic Compounds for the Detection of Prostate Cancer. PLoS ONE 2015, 10, e0143283. More recently in 2019, Lima et al. utilized headspace SPME GC-MS to identify VOC biomarkers for prostate cancer. Lima, A.R.; Pinto, J.; Azevedo, A.I.; Barros-Silva, D.; Jerónimo, C.; Henrique, R.; de Lourdes Bastos, M.; Guedes de Pinho, P.; Carvalho, M. Identification of a biomarker panel for improvement of prostate cancer diagnosis by volatile metabolic profiling of urine. Br. J. Cancer 2019, 121, 857–868. This study pretreated urine with NaCl and utilized divinyl-benzene / carboxen / polydimethylsiloxane (DVB / CAR / PDMS) SPME fibers. This method detected six VOCs that had higher accuracy than previous analyses (sensitivity = 89% and specificity = 83%). A follow-up study identified that some of the same VOCs, along with other analytes, could distinguish prostate cancer from bladder and kidney cancer with 92% accuracy.Att’y Docket No.: IUIC-168

[0147] Previous studies have utilized murine models to identify VOC biomarkers of cancer but have not demonstrated if the results are translatable to humans. Herein, urine specimens were collected from patients prior to undergoing prostate cancer biopsy or radical prostatectomy. In parallel, urine was also collected from mice with induced prostate cancer. The samples were analyzed through headspace analysis SPME coupled to GC-MS to identify VOCs in urine samples. Chemometric analyses were implemented to identify VOCs differentially expressed due to any form of prostate cancer and between indolent and aggressive prostate cancer samples. Materials and Methods Materials and Instrumentation

[0148] UTAK Laboraties drug-free normal urine (88121-CDF), Parafilm, pH paper, and polypropylene cups for urine collection were purchased from ThermoFisher Scientific (Waltham, MA, USA). Sodium hydroxide (50% wt. in solution, nitrogen flushed extra pure) and sodium chloride (99.85% pure) were also purchased from ThermoFisher Scientific. Guanidine hydrochloride (GHCl; pH = 8.5) was purchased from Sigma Aldrich (St. Louis, MO, USA). SPME fibers coated with DVB / CAR / PDMS (two centimeters in length) were purchased from Supelco (Bellefonte, PA, USA). SPME arrows of a similar chemical composition (DVB / Carbon Wide Range / PDMS) were obtained from Restek (Bellefonte, PA, USA). Headspace vials (10 mL) with screw-on caps were purchased from Restek or Agilent (Santa Clara, CA, USA). An Agilent 7890A GC system coupled to an Agilent 7200 MS quadrupole time-of-flight (QTOF) equipped with a PAL autosampling system (CTC Analytics, Zwingen, Switzerland) was used to incubate, extract, and analyze the VOCs. The GC column utilized for VOC separation was a Restek Rxi-5ms column of 30 m in length, a 0.25 mm internal diameter, and a 0.25 µm film thickness. Patient Recruitment

[0149] The inclusion criteria for sample collection from human subjects included the follow- ing: veterans at the Richard L. Roudebush Veteran’s Affairs Medical Center (VAMC) in Indianapolis recommended for prostate cancer biopsy, aged 40–85 years old, and a small set of men who were scheduled for radical prostatectomy. The exclusion criteria included serious additional medical problems (including another form of cancer within the last five years) or previous treatment for prostate cancer. The patients provided consent and urine samples were collected on the same day. Institutional Review Board (1506963798) and R&D permissionAtt’y Docket No.: IUIC-168 from Indiana University and the VAMC were obtained. The samples were deidentified and classified by International Society of Urological Pathology (ISUP) grades ranging from 0 to 5, including samples scored 0 (no cancer identified), 1 (the cancer cells found were predominantly indolent), 2 (indolent and aggressive cancer cells found), and 3–5 (cancer cells found predominantly aggressive). For the purpose of developing a screening test (no cancer vs. any form of cancer), samples with any ISUP grade > 0 were classified as cancer. For the purpose of developing a surveillance test, samples with ISUP grades ≤ 2 were classified as indolent and samples with ISUP grade ≥ 3 were classified as aggressive. Murine Model for Prostate Cancer

[0150] All of the procedures conducted were approved by the Indiana University Animal Care and Use Committee (protocol #SC311R) and complied with the Guiding Principles in the Care and Use of Animals, supported by the American Physiological Society (APS). Ten male C57BL / six mice (~8 weeks old) received transgenic adenocarcinoma mouse prostate (TRAMP-C2) cancer cells (approximately 0.1 million cells in 20 mL PBS) to the proximal tibia. Urine was collected and all of the tumor-bearing mice were sacrificed on day 21 after tumor injection. Prior to tumor injection, urine from the mice was obtained to serve as the healthy control group. The mice were caged at room temperature and fed the same diet (mouse chow ad libitum) during the experiment. The urine was collected over dry ice using Pasteur pipettes into glass centrifuge tubes. Furthermore, 75 µL aliquots of urine were transferred to a 10 mL headspace vial and stored in a -80 °C freezer before SPME GC-MS QTOF analysis was implemented. Human and Mouse Urine Sample Processing

[0151] The samples from the patients were collected in polypropylene cups, sealed with parafilm, refrigerated, deidentified, and transported on ice to Indiana University–Purdue University Indianapolis (IUPUI) where the samples were aliquoted into headspace vials (2.5 mL urine) and stored at -80 °C until SPME GC-MS QTOF analysis. The GC-MS assays were randomized but stratified so that the samples with the same ISUP grade were not run consecutively. The samples were defrosted 30 minutes prior to SPME coupled to GC-MS and saturated with salt (0.9 g NaCl in 2.5 mL urine). Additionally, all of the urine was pH corrected to 6.5–7 with small amounts of 1M NaOH as saturating with salt lowers the pH of samples differentially, which undesirably varies the VOC profile sampled through SPME. Regarding the analysis of urinary VOCs in the murine model of prostate cancer, the urine was defrostedAtt’y Docket No.: IUIC-168 30 minutes prior to SPME GC-MS analysis and treated with GHCl in a 1:1 volumetric ratio. This is significant as GHCl denatures the major urinary proteins (MUPs) in mouse urine that bind VOCs in hydrophobic pockets. SPME GC-MS QTOF Protocols

[0152] Prior to daily runs, the SPME fiber was preconditioned at 250 °C for 10 minutes. Vials with urine were incubated at 60 °C and agitated at 250 RPM for a total of 60 minutes, with extraction by SPME during the last 30 minutes. After extraction, the SPME fiber was injected into the inlet of the GC-MS QTOF which was held at 250 °C to thermally desorb the VOCs into the system. The oven temperature program utilized an initial temperature of 40 °C held for 2 minutes, followed by a ramp of 8 °C / min to 100 °C, 15 °C / min to 120 °C, 8 °C / min to 180 °C, 15 °C / min to 200 °C, and lastly 8 °C / min to 280 °C. The MS transfer line temperature was held at 250 °C during the chromatographic runs. Identical SPME GC-MS procedures (other than sample pretreatment) were implemented for the analysis of VOCs in mouse and human urine. External analytical reference standards in the form of high-density polyethylene (HDPE) were analyzed daily and demonstrated a high degree of instrumental and method reproducibility. Before analyzing mouse and human urine samples for prostate cancer biomarkers, UTAK urine standards were ran using a DVB / CAR / PDMS SPME fiber and a DVB / Carbon Wide Range / PDMS SPME arrow to determine which SPME type would be more effective in preconcentrating urinary VOCs. Data Screening and Chemometric Analysis

[0153] Using a previously described method, (Woollam, M.; Teli, M.; Liu, S.; Daneshkhah, A.; Siegel, A.P.; Yokota, H.; Agarwal, M. Urinary Volatile Terpenes Analyzed by Gas Chromatography–Mass Spectrometry to Monitor Breast Cancer Treatment Efficacy in Mice. J. Proteome Res. 2020, 19, 1913–1922; Woollam, M.; Teli, M.; Angarita-Rivera, P.; Liu, S.; Siegel, A.P.; Yokota, H.; Agarwal, M. Detection of volatile organic compounds (VOCs) in urine via gas chromatography-mass spectrometry QTOF to differentiate between localized and metastatic models of breast cancer. Sci. Rep. 2019, 9, 1–12) chromatographic deconvolution and spectral alignment were implemented through MassHunter Profinder for the mouse and human urine samples independently. Alignment was verified by observing the retention time spread for each molecular feature, and compounds with high retention time variation were eliminated from the matrix.Att’y Docket No.: IUIC-168

[0154] Spectrally aligned features that clearly contained two distinct molecular subpopulations were removed. VOCs present in fewer than 50% of samples in at least one class were also discarded. Features that were silica-based due to interaction with the SPME fiber and / or column were rejected, as were VOCs that were known exogenous pollutants and plasticizers. Preprocessing of the data matrices continued by normalizing each integrated VOC signal by relative abundance to account for variations in urine flow rate and concentration due to water intake. Two-tailed Student’s t-tests were implemented on the normalized data to screen for VOC differences between cancer and no cancer in both human and murine models. Univariate significance testing was also undertaken to identify VOCs differentially expressed in aggressive prostate cancer for human urine analysis.

[0155] Regarding multivariate analysis, subsets of VOCs were subject to principal component analysis (PCA) to observe global patterns in the data. PCA was implemented for the mouse and human urine samples independently. PCA reduces data dimensionality and identifies outliers, but this multivariate statistical technique is unsupervised and therefore may not differentiate sample classes well if the distinguishing features for the disease are not the ones with maximum variance in all samples. In such cases, supervised methods such as linear discriminant analysis (LDA) are used to separate samples when reliable class labels are available. Supervised iterative LDA builds the model using a forward feature selection method and has been previously utilized for identifying VOC biosignatures of cancer. Starting with the set of screened VOCs, LDA was performed on all permutations of three compounds, with it finding the three VOCs with the highest ability to distinguish two classes. Next, by reserving one of these three and re-analyzing, the best four compounds are found, and the process continues. The final set of compounds is obtained when the addition of more VOCs does not improve the cross-validated classification accuracy. The stability of the LDA models was tested by data perturbation. To perturb the data set, a hold-out scheme was employed using the Matlab classifier application. In this system, a fraction (one fifth for five-fold cross-validation) of the samples was excluded to train the data set, and the held-out samples were used to test the model. The hold-out method was repeated using different randomly held out samples (1000 times) and an average receiver operating characteristic (ROC) curve was generated. Only the human urine sample data set was analyzed using supervised multivariate statistical methods. Results SPME OptimizationAtt’y Docket No.: IUIC-168

[0156] Prior to analyzing the samples, UTAK urine standards were analyzed using SPME fibers and arrows to identify the optimal extraction method. Here, DVB / CAR / PDMS SPME fibers were quantitatively compared to DVB / Carbon Wide Range / PDMS SPME arrows. Both SPME devices were analyzed using the same extraction protocol, and the results showed that after chromatogram deconvolution, there were no significant differences in the number of VOCs detected or the total integrated GC-MS signal (FIGs. 49A–49B). Upon integrating individual VOC signals, it was realized that the SPME fiber was much more sensitive toward VOCs with a relatively low molecular weight. For example, the fiber could successfully extract analytes including acetone, ethyl acetate, 2-butanone, and chloroform. These VOCs could not be detected when extracting the urine standards with the SPME arrows. The SPME fiber also extracted 2-pentanone, 3-hexanone, 3-methyl-2-pentanone, and toluene with higher sensitivity relative to the arrow (FIG. 47A). Regarding VOCs with a higher molecular weight, varying results were identified in the optimization experiments. For example, the SPME arrow was slightly more sensitive to 2,5-dimethylbenzaldehyde, carvone, 2-ethyl-1-hexanol, and bornyl acetate. The SPME fiber on the other hand was slightly more sensitive to 2,4-di-tert- butylphenol and no significant differences were identified for 4-heptanone and p-menth-1-en- 3-one (FIG.47B). Therefore, it was decided for the SPME fiber to be used for prostate cancer VOC biomarker discovery. Patient Recruitment and Urine Collection

[0157] Human subjects (n = 162) meeting the inclusion and exclusion criteria provided consent and donated urine specimens prior to prostate cancer biopsies or prior to radical prostatectomy at the Richard L. Roudebush VAMC. Biopsy results from the subjects’ medical records were used to classify samples by ISUP grade and GS. The patient group is not random as all the men presented were scheduled for and underwent biopsies mainly due to elevated PSA levels. In total, 67 patients had negative biopsy results and 95 had positive prostate cancer biopsy results. Of those diagnosed with prostate cancer, there were 38 men diagnosed with ISUP grade 1, 30 with ISUP grade 2, 8 with ISUP grade 3, 5 with ISUP grade 4, and another 14 patients diagnosed with ISUP grade 5. The men were subsequently stratified into the following sample classes: (1) no cancer (those with negative biopsies, ISUP grade 0), (2) prostate cancer (positive prostate cancer biopsy with ISUP 1), (3) indolent prostate cancer (ISUP grades 1 and 2), andAtt’y Docket No.: IUIC-168 (4) aggressive prostate cancer (ISUP grades ranging from 3 to 5). Regarding the murine model of prostate cancer, 8 urine samples were collected prior to tumor injection to serve as healthy controls and 9 samples were collected after tumor injection. Mouse Urine Volatile-Organic-Compound Analysis

[0158] After SPME GC-MS analysis of the mouse urine samples, spectral alignment and data screening procedures yielded a total of 161 VOCs for chemometric analysis. Of these VOCs, 16 molecular features were identified by the Student’s t-test to have a p-value < 0.05 and 25 had a p-value < 0.10. A volcano plot interpolating statistical significance as a function of log2- fold change (FC) can be observed in FIG. 9A and shows that there were slightly more VOCs upregulated by prostate cancer relative to those downregulated. This was observed in the data set containing all VOCs, and the analytes which were identified to be differentially expressed due to induced prostate cancer. To visualize the VOC signals in each of the samples collected, a hierarchical heatmap was generated for all of the VOCs identified with a p-value < 0.10 (FIG. 9B). The heatmap also demonstrates there were more upregulated VOCs compared to downregulated features. In general, the VOCs in the heatmap display low intraclass variation and high interclass variation. Interestingly, the VOCs upregulated by prostate cancer in the mouse model had lower variations within the healthy control samples relative to the downregulated compounds. Lastly, unsupervised PCA was implemented on the five VOCs with the lowest p-values and the first two principal components accounted for 62.4% of the variation present in the sample data. Prostate cancer mouse urine samples were distinguished from healthy controls with 100% accuracy in the two-dimensional PCA plot (FIG. 9C), showing the discriminatory power of urinary VOCs. Other relevant species identified from univariate analysis include: Distinguishing Prostate Cancer in Humans

[0159] Human urine samples were also analyzed by SPME GC-MS, and similar data processing and analyses were implemented to distinguish men diagnosed with prostate cancer from those with negative biopsy results. Spectral alignment of qualified samples produced a matrix of 367 VOCs present in at least 50% of one of the four sample classes of interest (no cancer, prostate cancer, indolent prostate cancer, or aggressive prostate cancer). The Student’s t-test was used to probe differences in human urine due to the presence of any grade of prostate cancer and it identified 17 VOCs with a p-value < 0.05 and 29 features with a p-value < 0.10. Volcano plots were generated using these results and showed that most VOCs were upregulatedAtt’y Docket No.: IUIC-168 in the urine samples collected from men with prostate cancer (FIG.10A). This especially holds true for VOCs identified with relatively low p-values and high absolute log2FC values. Hierarchical heatmaps were made for VOCs that had a p-value < 0.05, and they show that VOCs had relatively high variation within both the no cancer and prostate cancer sample classes (FIG.10B).

[0160] To identify patterns in the data, unsupervised multivariate statistical analysis was implemented in the form of PCA. Two-dimensional PCA was tested with different permutations and combinations of VOCs, and limited separation of prostate cancer from the no cancer sample class was observed. For example, when PCA was implemented on features identified with a p-value < 0.05, the first two principal components accounted for 27.2% of the variation in the sample data and there was a large degree of overlap between the prostate cancer and no cancer samples (FIG. 15A). Even though PCA could not separate the sample classes with high accuracy, the first principal component showed higher statistical significance relative to any individual VOC (FIG. 15B). Regardless, PCA demonstrates poor sensitivity and specificity values, indicating that unsupervised multivariate analysis was not sufficient for accurate prostate cancer classification. Therefore, forward feature selection coupled to LDA was implemented on qualified VOCs to build a predictive classification model and identify a biosignature of VOCs that classifies prostate cancer. This method identified six VOCs that discriminated between sample classes with an area under curve (AUC) equal to 0.76, sensitivity equal to 76%, and specificity equal to 67% in the training data set. This set included: β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, and 2,3,4,5- tetramethyl-2-cyclopenten-1-one.

[0161] The average five-fold cross-validation AUC was equal to 0.75 (sensitivity = 75% and specificity = 69%), which demonstrated that the model was not overfit. Linear discriminant 1 (LD 1) scores and ROC curves for this classification model can be observed in FIGs.11A, 11B. Because the model was not overfit, the team continued with forward feature selection to add more VOCs to the model and increase the cross-validated classification accuracy. The algorithm selected a slightly different panel of seven VOCs that had higher classification accuracy in the training data (AUC = 0.80, sensitivity = 82% and specificity = 66%). Even though higher accuracy was observed in the training data, there were no increases in AUC values when the model was perturbed by five-fold cross-validation (AUC = 0.75, sensitivity = 82%, and specificity = 61%), indicating that this model is more overfit relative to the previously presented biosignature. LD 1 scores and ROC curves for this model can be observed in Figure 11C, 11D.Att’y Docket No.: IUIC-168

[0162] Chemical species identified as having p-value < 0.10 for presence of prostate cancer are shown in Table 1A, below: Table 1A. Cancer vs. No-Cancer Univariate VOC ID (NIST Systemic Name) RT Base m / z p-value 6 65 84 05 79 49 97 66 7 09 47 62e ca spec es e e as ca g p ese ce o p os a e ca ce a e s o in Table 1B, below: Table 1B. Cancer vs. No-Cancer BiosignatureStratifying Indolent and Aggressive Prostate Cancers

[0164] VOCs in human urine headspace were also analyzed by similar chemometric approaches to distinguish aggressive prostate cancer from indolent prostate cancer and men with negative prostate cancer biopsies (no cancer). Here, differences in VOC expression were probed between patients with ISUP grades 0–2 and men with ISUP grades 3–5. UnivariateAtt’y Docket No.: IUIC-168 analysis was implemented on the same data matrix of VOCs and identified 16 VOCs with a p- value < 0.05 and 40 volatiles with a p-value < 0.10.

[0165] As opposed to the previous analysis distinguishing any grade of prostate cancer from the healthy controls, volcano plots for this statistical comparison demonstrated that there were more differentially expressed VOCs downregulated in aggressive prostate cancer samples relative to those that were upregulated (FIG. 12A). Hierarchical heatmaps were also constructed for the features identified by the Student’s t-test to visualize their normalized signals in all of the samples (FIG.12B). Similar to the prostate cancer results, these VOCs also displayed high intraclass variation, indicating that these molecules alone may not have a high ability to stratify aggressive prostate cancer.

[0166] Next, multivariate analyses were implemented to assess the ability of multiple VOCs to separate aggressive prostate cancer. Unsupervised approaches through PCA were utilized in a similar fashion as previously shown for prostate cancer samples. The first two principal components explained 37.7% of the variation in the samples when PCA was used to reduce data dimensionality for the VOCs with a p-value < 0.05 (FIG.15A). Even though the first two principal components accounted for more variation in the data relative to the PCA results for separating prostate cancer, the aggressive cancer samples were not accurately stratified from indolent cancer and no cancer. The first principal component alone accounted for 20.2% of the variance (FIG. 15B) in the data and aggressive cancer displayed significant differences when compared to the other samples (p-value = 0.0001). Unlike the PCA for prostate cancer, PCA of aggressive cancer did not display greater significant differences relative to individual VOCs alone, as the VOC with the lowest p-value was an aromatic compound and had a similar p- value when compared to the first principal component.

[0167] Because PCA did not provide adequate stratification of aggressive prostate cancer, forward feature selection coupled to LDA was implemented on the screened data. This approach initially produced a predictive classification model of six VOCs that could distinguish prostate cancer with an AUC equal to 0.83 in the training data set (sensitivity and specificity = 81%). These included: Five-fold cross-validation was performed 1,000x and the average AUC was equal to 0.82 (sensitivity = 81% and specificity = 76%) which demonstrated that the model was not overfit. FIGs 13A, 13B show the LD 1 scores and ROC curves for this classification model. Analysis continued by identifying an alternative biosignature of seven VOCs that could distinguish aggressive prostate cancer with even higher accuracy. This new model demonstrated an AUC equal to 0.89 with relatively higher specificity values in the training data set (sensitivity = 81% and specificity = 89%). The model of VOCs also showed a high degreeAtt’y Docket No.: IUIC-168 of stability, as the average five-fold cross-validated ROC curve displayed an AUC equal to 0.87 with a sensitivity equal to 78% and a specificity of 85% (LD 1 scores and ROC curves shown in FIGs.13C, 13D).

[0168] Chemical species identified as having p-value < 0.10 for aggressive cancer are shown in Table 2A, below: Table 2A. Aggressive vs. Indolent Univariate VOC ID (NIST S stemic Name) RT (av ) Base m / z Aggressive vs.Att’y Docket No.: IUIC-168 Hex-3-en-1-yl propyl carbonate 14 67.0533 0.087263621 Cyclohexanone, 2-(1-methylethylidene)- 10.82 67.0535 0.09000646below: Table 2B. Aggressive vs. Indolent Biosignature VOC ID (NIST Systemic Name) RT (avg) Base m / zo pa g o a e s ouse e a u a e

[0170] Once urinary VOCs were analyzed for prostate cancer biomarkers in mice and humans independently, the results were compared. First, the specific functional groups were evaluated to assess the structural similarity of the candidate biomarkers in both models. VOCs with a p- value < 0.05 in the mouse model and human samples (including prostate cancer vs. no cancer and aggressive cancer vs. indolent / no cancer) were assessed for functional group frequency and functional group frequency ratio (FIGs 16A, 16B).

[0171] Overall, there was a high degree of similarities between the mouse and human urinalysis. The most common functional groups detected among both models consisted of nonaromatic (unconjugated) cyclics, aromatics, ketones, terpenes, and esters. Aromatic VOCs showed the highest abundance for aggressive prostate cancer comparison in humans, while ketones were the most prominent functional group detected in the mouse model. Nonaromatic cyclics were the most abundant VOC for prostate cancer stratification in humans but were also highly implicated in the other comparisons. Lastly, alcohols were more abundant in the mouse model and esters were more prominent in humans for the prostate cancer vs. no cancer univariate test.Att’y Docket No.: IUIC-168 Discussion

[0172] The VOCs emanating from urine in this study were assessed for their ability to stratify prostate cancer and aggressive prostate cancer, but there has been significant research on various cancer types and other medical conditions. Some of the most well studied cancer types for VOC biomarker detection include prostate cancer, breast cancer, lung cancer, and bladder cancer. For example, some of the current authors have extensively published and identified VOC biomarkers in urine using mammary-tumor-induced murine models for breast cancer biomarker discovery. These studies show VOCs in urine (particularly aromatics, terpenes, and carbonyls) have the potential of determining tumor location, identifying the effect of antitumor therapies, and tracking mammary tumor progression over the course of time. Breast cancer biomarkers have also been identified in human urine with varying diagnostic sensitivity / specificity values, and some of the most cited VOCs as potential biomarkers for breast cancer in these studies also consists of terpenes / terpenoids and ketones. More recently, Kure et al. built a classification model based on two urinary VOCs, 2-butanone and isopropanol (ketone and alcohol), which identified breast cancer with sensitivity and specificity equal to 93% and 83%, respectively. Kure, S.; Satoi, S.; Kitayama, T.; Nagase, Y.; Nakano, N.; Yamada, M.; Uchiyama, N.; Miyashita, S.; Iida, S.; Takei, H.; et al. A prediction model using 2-propanol and 2-butanone in urine distinguishes breast cancer. Sci. Rep.2021, 11, 19801.

[0173] The current analysis, however, focuses on prostate cancer detection through urinary VOCs and showed that method sensitivity is a significant factor to consider prior to biomarker discovery. The initial experiments in this study optimized SPME type by comparing fibers and arrows with similar chemical composition (DVB / CAR / PDMS fiber and DVB / Carbon Wide Range / PDMS arrow). The results revealed that even though SPME arrows have a larger surface area relative to fibers, there were no major differences in the total number of VOCs detected between the SPME types (FIGs. 49A–49B), and arrows are not efficient for extracting low molecular weight VOCs (FIGs.9A–9C). The rationale is that CAR has high adsorption affinity for low molecular weight VOCs and Carbon Wide Range does not (SPME Fiber Coating Selection Guide, provided by Sigma Aldrich). Even though the arrows did have higher sensitivity for a handful of larger VOCs, the SPME fiber was still able to preconcentrate these analytes with sufficient sensitivity. Therefore, for prostate cancer VOC biomarker discovery, it was decided for SPME fibers to be utilized. After quantitative comparison of SPME fibers and arrows, mouse and human urine samples were analyzed for VOC biomarkers. The volcano plots showed that urinary VOCs were upregulated by prostate cancer in both mice and humans (FIGs. 9A, 10A). When comparing aggressive cancers to the indolent and healthy controls,Att’y Docket No.: IUIC-168 more VOCs were significantly downregulated relative to those that were upregulated (FIGs. 13A, 13B).

[0174] There is not yet a robust physiological explanation as to why VOCs were more abundantly downregulated in aggressive cancer, but previous reports have shown that metabolites can be downregulated by cancer. This may potentially be due to cancer cells using these VOCs to meet the increased energetic requirements for tumor growth. Next, hierarchical heatmaps were constructed to visualize the expression of VOCs in the mouse and human urine data sets. VOCs identified in mouse urine (FIG. 9B) had low intraclass variation and high interclass variation, but when analyzing human urine (FIG.11B, 13b), VOCs had much higher variations within sample classes. This is expected as simplified models of disease including murine and in vitro studies offer experimental control of parameters that may undesirably vary urinary VOC profiles (diet, external environment, etc.). Nonetheless, simplified models of cancer do not take into account VOC differences that are due to the interactions between the tumor and the local microenvironment.

[0175] Multivariate analysis was implemented as a panel of VOCs may better reflect multiple pathways affected by a medical condition and can be cross-validated for accuracy better than a single compound. Regarding mouse urine analysis, only unsupervised analysis was utilized due to the limited number of samples. Supervised multivariate analyses are more prone to overfitting when employed on small data sets. PCA was capable of perfectly distinguishing prostate cancer in the mouse model (FIG.9C), but not in humans (FIGs.14A–14B, 15A–15B). This is most likely because the highest amount of variation in the mouse data was due to prostate cancer, as there was a high degree of control over other variables. To identify a biosignature of VOCs to separate cancer / no cancer and aggressive / indolent prostate cancer in humans, iterative LDA was implemented. A panel of six VOCs was identified to distinguish prostate cancer with a cross-validated AUC equal to 0.75, and a separate set of seven VOCs had a cross-validated AUC of 0.75 (FIGs.11A–11B).

[0176] The panel of six VOCs was highly stable, while the panel of seven VOCs showed signs of overfitting. Iterative LDA also identified a panel of six VOCs that could distinguish aggressive cancer with a cross-validated AUC of 0.82 and a separate model of seven VOCs with a cross-validated AUC equal to 0.87 (FIG. 13A–13D). Higher accuracies for classifying aggressive cancer were observed, which may be because there are metabolic similarities between indolent prostate cancer grades and healthy controls.

[0177] Mouse and human urine were initially analyzed independently, and then the results were compared. Functional group frequency analysis (FIGs. 16A, 16B) indicated that mouseAtt’y Docket No.: IUIC-168 and human urine showed similar VOC functionality. The results from these analyses align with metabolic pathways dysregulated by prostate cancer. Carbonyls including volatile ketone bodies are produced when cytochrome P450 (CYP450) reduces hydroperoxides through fatty acid oxidation. Increases in fatty acid oxidation are correlated with prostate cancer and have been studied as a dominant process that facilitates tumor growth by providing the necessary bioenergetics. Moreover, P450 enzymes enable tumor progression as they activate carcinogens and reactive oxygen species. Volatile terpenes / terpenoids have also been implicated as potential biomarkers and they are hypothesized to be biosynthesized via the mevalonate (MVA) pathway. Cancer requires increases in MVA products, including cholesterol and prenylated proteins, which have been used to develop inhibitors of farnesyl transferases, a common protein prenylator. Increased MVA activity increases invasiveness in prostate cancer cells and may play a role in cancer progression.

[0178] The results from this study show urinary VOCs can accurately differentiate any form of prostate cancer and distinguish aggressive cancer with even higher sensitivity and specificity. The VOCs implicated as biomarkers for prostate cancer in the mouse and human urine samples were also shown to be structurally similar and are associated with expected metabolic pathways or have been previously identified as VOCs of inflammation or cancer. The use of a urine test to differentiate aggressive prostate cancer could reduce procedures on men identified with indolent prostate cancer, at a savings to the healthcare industry and the patients.

[0179] Rapid and point-of-care technologies for prostate cancer screening and distinguishing indolent / aggressive tumors may be able to improve medical decision-making, increase widespread testing, and decrease morbidity associated with prostate cancer in the future. Example 2. Detection by Portable Gas Chromatography – Mass Spectrometry of Volatile Organic Compound Biomarkers.

[0180] It is of significant clinical interest to develop a noninvasive and accurate complementary screening method to detect patients with prostate cancer and differentiate indolent from aggressive tumors. Innovative and novel technologies to improve current prostate cancer diagnosis have the potential to increase widespread screening, increase early detection, decrease unnecessary biopsies and treatments, and identify patients who will benefit most from radical prostatectomies (RPs) and / or other therapies.Att’y Docket No.: IUIC-168

[0181] Researchers are seeking alternative techniques to screen for and detect prostate cancer through identifying biomarkers in different sample types using various “omic” techniques. These methodologies include but are not limited to genomics, transcriptomics, proteomics, metabolomics and even lipidomics. For instance, one emerging urine-based assay currently used in the clinic is called SelectMDx, which monitors the expression of two different strands of RNA (HOXC6 and DLX1) to identify men who would benefit from a prostate cancer biopsy16. The benefits of this assay include high diagnostic specificity, especially for patients with advanced grades of prostate cancer (Gleason Score (GS) ≥ 4+3 or International Society of Urological Pathology (ISUP) grade ≥ 3). However, this method is still shrouded with uncertainty as there are varying reports of the test’s overall diagnostic accuracy.

[0182] GC-MS is regarded as the gold standard for separation, structure elucidation, and quantification of volatile carbonyls, terpenes / terpenoids and other VOCs in complex sample matrices including urine. Portable GC-MS systems can be adapted to detect urinary VOC biomarkers for prostate cancer in a similar fashion as the FDA approved VOC assay for COVID-19, thereby permitting widespread, rapid, and accurate screening for prostate cancer. (It should be understood that GC-MS approach to could alternatively be carried out using gas chromatography (GC) with a flame ionization detector (FID), thermal conductivity detector (TCD), photoionization detector (PID), or combination thereof; or using mass spectrometry (MS), such as direct injection MS, selected ion flow tube (SIFT), ion mobility, proton transfer reaction (PTR), or any combination thereof. Recent developments have permitted development of 2-dimensional GC systems (“GCxGC”), which it is readily understood may be employed with MS, FID, TCD, PID, or any combination thereof for carrying out the method of the present example.)

[0183] Portable GC-MS instruments can be utilized for point-of-care detection in the clinic or other locations. Urinary VOC analysis will also be performed in parallel utilizing a miniaturized, cost-effective, and sensitive integrated nanosensor array. Implementing gas sensors for VOC biomarker detection represents an emerging frontier in diagnostic tools for improving medical decision-making, and one that can be adapted for home use. VOC detection using gas sensors (also termed electronic noses, e-Noses) relies on a change in optical, electrical (conductivity or resistance), or physical properties when biomarkers are exposed to the sensing layer. Current e-Nose systems are fabricated using different polymers and nanomaterials including metal oxides, gold or other metallic nanoparticles, carbon black, graphene, carbon nanotubes, nanocrystals, and metallic organic frameworks. These devices have been utilized to detect VOC biomarkers in urine or other biological samples for differentAtt’y Docket No.: IUIC-168 types of cancer, including but not limited to lung, breast, bladder, gastric, colorectal, and even prostate cancer. Recently, Taverna et al. developed an integrated gas sensor array of metal oxides (TiO2, ZnO and SnO2) to analyze urinary VOC biomarkers of prostate cancer, with results showing relatively high accuracy (sensitivity equal to 85% and specificity equal to 79%). Taverna, G.; Tidu, L.; Grizzi, F.; Torri, V.; Mandressi, A.; Sardella, P.; La Torre, G.; Cocciolone, G.; Seveso, M.; Giusti, G.; et al. Olfactory system of highly trained dogs detects prostate cancer in urine samples. J. Urol.2015, 193, 1382–1387; Taverna, G.; Grizzi, F.; Tidu, L.; Bax, C.; Zanoni, M.; Vota, P.; Lotesoriere, B. J.; Prudenza, S.; Magagnin, L.; Langfelder, G.; Buffi, N.; Casale, P.; Capelli, L., Accuracy of a new electronic nose for prostate cancer diagnosis in urine samples. International Journal of Urology. 2022. A significant downfall of these and other studies that utilize sensor arrays to diagnose cancer is they rely on a black box approach. Individual sensors within the array adsorb a wide range of VOCs present in biological samples, and therefore rely on machine learning algorithms for accurate classification. Therefore, characteristic sensor responses cannot be attributed to specific VOC biomarkers for prostate cancer. A coordinated and integrated approach will be utilized to validate the VOC biomarkers of prostate cancer and identify their ability to classify indolent and aggressive cases.

[0184] In an ongoing study, 162 urine samples were collected from patients undergoing prostate cancer biopsy at the Roudebush VAMC.67 samples were from patients with no sign of prostate cancer and 95 were from patients with a positive prostate cancer biopsy result. Urine samples were analyzed using a GC-MS quadrupole time-of-flight (QTOF) system with VOCs preconcentrated by a DVB / CAR / PDMS SPME fiber. Chromatograms were spectrally aligned to identify conserved VOCs and data screening procedures were implemented to remove method artifacts. Machine learning performed through forward feature selection coupled to linear discriminant analysis (LDA) identified a panel of six VOCs that could distinguish prostate cancer with a receiver operator characteristic (ROC) area under the curve (AUC) equal to 0.76 (sensitivity = 75% and specificity = 70%) in the training data set. Five-fold cross validation was undertaken and showed that the chemometric model of VOCs is robust, as these results showed similarities to the training data. The one-dimensional LDA plot and ROC curve for training and cross validation are shown in Figures 11A and 11B.

[0185] In addition to performing chemometric analysis to distinguish any form of prostate cancer, the team has also assessed the ability of VOCs to identify aggressive cases (ISUP grade ≥ 3). Similar methodologies were implemented, and the statistical analyses implicated a panel of seven VOCs that had high discriminatory power. These VOCs when analyzed by LDA hadAtt’y Docket No.: IUIC-168 a ROC AUC equal to 0.89 (sensitivity = 81% and specificity = 89%) in the training data set, and similar results were obtained upon cross validation (ROC AUC = 0.87). The one- dimensional LDA plot and the ROC curves are shown in Figures 13C and 13E. These results indicate that the model is not overfit and can be used to make clinically relevant predictions.

[0186] Our team has also preliminarily assessed the ability of the Teledyne FLIR Griffin™ G510 portable GC-MS to profile VOCs emanating from a urine standard. The preliminary analyses used GC protocols, MS parameters and SPME procedures which were unoptimized. The results show that the portable system can detect an abundant number of urinary VOCs (total ion chromatogram presented in Figure 50). The team has identified a handful of analytes within the chromatogram, and these include but are not limited to heptanone, nonanal, carvone, acetone (volatile carbonyls) and volatile terpenes / terpenoids. The method could even detect a previously implicated biomarker for prostate cancer (2,5-dimethylbenzaldehyde) with high sensitivity.

[0187] The team has also tested the MOx nanosensor array using human urine standards. These experiments sought to identify the minimum exposure time that produced adequate sensitivity. Purified air at a fixed flow rate was bubbled through the urine standard, and the resulting gas was exposed to the surface of the sensor array for one minute, five minutes and ten minutes (n = 3). The response curve for one of the sensors can be seen in Figure 56, which shows that exposure times as low as one minute yielded sufficient sensitivity. 5-minute and 10-minute exposures showed a saturated response as a clear plateau can be observed. Results also showed high reproducibility among the replicates (relative standard deviation < 5%). These preliminary studies show that the nanosensor array produces a rapid and reproducible response toward VOCs in urine, which will be further explored and optimized in future experiments. These results detecting VOCs in urine to distinguish any form of prostate cancer and aggressive grades had higher accuracy when compared to the results of the PSA test. PSA levels in blood could only distinguish prostate cancer with a ROC AUC equal to 0.58 (sensitivity and specificity = 56%). The PSA test did have increased ability to distinguish aggressive cancers, but only reached a ROC AUC equal to 0.67. Optimization of Portable Gas Chromatography – Mass Spectrometry Analysis of Volatile Organic Compounds.

[0188] The team will use a state-of-the-art Teledyne FLIR Griffin™ G510 portable GC-MS (Figure 53) with a linear quadrupole mass analyzer to analyze VOCs in urine headspace. ThisAtt’y Docket No.: IUIC-168 off-the-shelf system will be capable of analyzing a wide range of VOCs and can be utilized via direct injection, thermal desorption and SPME techniques. The instrumentation demonstrates excellent sensitivity, with limits of detection (LOD) that can reach down to low part per trillion levels, especially when utilizing preconcentration techniques such as SPME. The GC column lengths are typically 15-20 meters, which results in an approximate 15-minute run time typically. Data from the system can be transmitted to a tablet or laptop using Bluetooth technologies. The system is equipped with an array of processing software, which includes Griffin System Software (GSS), National Institute of Standards and Technology (NIST) mass spectral libraries, and the Automated Mass Spectral Deconvolution and Identification System (AMDIS), allowing for deconvolution as well as spectral alignment. The operating parameters of the portable GC-MS will be optimized utilizing a robust design of experiments. Rigorous optimization is an objective means of determining the best values for instrumental parameters without performing every possible combination. Standard urine and analytical standards containing previously identified biomarkers for prostate cancer (carbonyls, terpenes / terpenoids and others) will be used for optimization. Chromatographic resolution will be optimized while maintaining a rapid analysis by systematically varying parameters including but not limited to oven ramp temperature programming and chemical composition of the GC column stationary phase. VOCs in the analytical standard will demonstrate baseline resolution (≥ 1.5), but values will likely be lower in urine samples given the complexity of the sample matrix. Maximizing resolution with standards will enhance the potential for deconvoluting the most VOC biomarker candidates in human samples.

[0189] In addition, dynamic linear ranges and sensitivity of the instrument will be quantified. Internal and external reference standards will be prepared and replicates of each calibrant at various concentrations will be analyzed. Linearity of the calibration curves will be assessed with a target of R2> 0.99. The LOD will be determined by calculating the signal-to-noise ratio (S / N) for each of the VOCs. A plot of the S / N as a function of concentration will be created and the calculated concentration at which the S / N is 3 will be defined as the LOD, and S / N equal to 10 for limit of quantification (LOQ). To correct for carryover, the team will evaluate samples with a concentration that is twice the highest calibrator and then analyze a negative control. If VOCs are greater than the LOD in the negative control, sample injection will be adjusted to allow for increased rinses / purges between samples. Accuracy and precision of the portable GC-MS will also be assessed by analyzing positive controls with relatively high and low concentrations (n = 5 for each control). The experiments will be repeated for three more days for a total of four experiments.Att’y Docket No.: IUIC-168 Subject Selection and Recruitment.

[0190] Urine samples will be collected from patients during their scheduled visits to urology clinics for prostate cancer biopsies. Subjects will be recruited from two patient populations: the VAMC Urology Clinic (primary recruitment location) and from the Indiana University Urology Department. The advantages of this multi-site observational study are that it will increase sample size and make results of this study more generalizable. Urine from patients with prostate cancer (positive control) will be collected before biopsy or therapies. Samples from subjects undergoing biopsy where no tumor is found will be collected to serve as prostate cancer negative controls. Urine samples from subjects diagnosed with prostate cancer will be sorted based on the pathological results from biopsy or surgery through radical prostatectomy. Collection of urine will continue until 75 samples have been collected from each of the three sample classes of interest: 1) men with negative biopsy prostate cancer biopsy results, 2) patients diagnosed with indolent prostate cancer (GS ≤ 3+4, ISUP grade ≤ 2), and 3) subjects with an aggressive prostate cancer diagnosis (GS ≥ 4+3, ISUP grade ≥ 3). It is also important to note that for urine specimens collected prior to biopsy, if that subject undergoes radical prostatectomy within 6 months, the results from the surgery will be used to determine if the patient has aggressive or indolent prostate cancer. The team will also collect an additional 75 urine samples from patients diagnosed with urinary tract infections (negative control) who have received negative prostate cancer biopsy results within three months prior to sample collection. The team will utilize exclusion criteria for recruitment, which includes patients that have another type of cancer or blood in their urine. Sample Collection and Tumor Grading.

[0191] Subjects will be asked to provide urine samples (approximately 30–40 mL) at the time of their regularly scheduled visits at the urology clinics. Immediately after sample acquisition, the urine will be aliquoted into headspace vials for SPME GC-MS (2.5 mL of urine) and into a separate polypropylene urine collection cup for nanosensor analysis (10 mL of urine). These samples will be analyzed by both systems in the clinic after sample collection and preparation, which will be described below in further detail. Remaining urine samples in the initial collection container will be double wrapped in parafilm. Samples in all cases will be de-identified on site and will be transported to IUPUI and stored in a -80°C freezer. Excess samples (unused aliquots of urine) from this study will be preserved in a Prostate Cancer Biorepository that will enable de-identified urine samples with relevantAtt’y Docket No.: IUIC-168 medical histories to be stored and shared with the wider prostate cancer community. Tumor grading will occur after prostate cancer biopsy and / or radical prostatectomy and the results will be shared with the team and associated with the de-identified urine samples. SPME and Portable GC-MS Analysis.

[0192] Prior to analysis using the portable GC-MS, VOCs in urine will be sampled and preconcentrated using a SPME fiber. While canines can identify VOC biomarkers for prostate cancer in untreated urine samples at room temperature, samples can be pretreated and heated to enhance the concentration of pre-existing urinary VOCs into the sample headspace, and therefore increase the sensitivity of the method. Previous experiments have shown that salting out urine with high grade NaCl, pH correcting the samples to 7 and extracting VOCs at 60°C using a DVB / CAR / PDMS SPME fiber is an optimal method, as it produced higher VOC sensitivity compared to using NaOH for sample pretreatment and extracting volatile analytes using a CAR / PDMS fiber. Therefore, this methodology will be employed immediately after the sample is collected and aliquoted into a headspace vial. Prior to patient recruitment, the team will also undertake experiments to determine the minimum extraction time that can sensitively sample VOCs, to develop the most rapid protocol for analysis without compromising method fidelity. After VOCs from patient samples have been extracted, the SPME fiber will be injected into the portable GC-MS inlet held at 250°C to thermally desorb the analytes into the system for chromatographic separation, structural elucidation, and quantification. The instrument will be calibrated daily, and an analytical reference standard will be run every day of sample analysis. Urine samples do not contain equivalent concentrations of VOCs, and therefore different strategies will be undertaken to normalize the resulting data. Techniques that will be explored include but are not limited to normalization through creatinine levels, MS Total Useful Signal (MSTUS) and auto scaling (z-scoring). Volatile Organic Compound Identification and Machine Learning.

[0193] Data from the portable GC-MS will be deconvoluted using AMDIS and spectrally aligned using open source software (XCMS, SpectConnect or similar) to identify conserved VOCs in an automated fashion. VOC identification will be undertaken initially utilizing the NIST17 mass spectral library, but pure analytical standards will be purchased and analyzed for verification purposes (retention time and mass spectral fragmentation pattern matching). GC-MS results will be subject to significant analysis for microarray (SAM) to identify VOCs significantly enriched or depleted due to any form of prostate cancer or aggressive tumors.Att’y Docket No.: IUIC-168 Data mining will be performed utilizing unsupervised principal component analysis to identify global patterns in the data and observe how the samples naturally separate into their sample classes of interest. Next, identified VOCs implicated by SAM will be screened to exclude suspected exogenous molecular features including metabolites of foods or drugs, environmental toxins, artifacts of SPME GC-MS or air / water pollutants. Furthermore, the team will also highlight the previously identified VOCs for prostate cancer, focusing on volatile carbonyls, terpenes / terpenoids and other putative biomarkers. Multivariate chemometric analysis through developing machine learning algorithms will be used to classify prostate cancer from the negative controls, and aggressive from indolent prostate cancer. The team will utilize support vector machine (SVM), random forest (RF), linear discriminant analysis (LDA), artificial neural networks (ANNs) and / or other algorithms to identify and validate a biosignature that stratifies the sample classes of interest. This will include a targeted approach by only assessing previously identified biomarkers of prostate cancer and performing an untargeted analysis of all qualified VOCs. The team also plans to compare the positive predictive values of the VOC results to alternative tests including SelectMDx, Sentinel, Prostate Health Index, 4K, and even PSA. In parallel, the team will also use machine learning to couple the VOC data to the results of the other tests which could increase the accuracy for prostate cancer detection. Example 3. Evaluation Of Complementary Metal Oxide (MOx) Nanosensor Array To Detect Urinary VOC Biomarkers For Prostate Cancer, And To Distinguish Aggressive And Indolent Cancers.

[0194] A commercially available nanosensor array (e.g., manufactured and supplied by NanoZ) will also be utilized to analyze VOC biosignatures correlating to prostate cancer diagnosis. An array of non-selective gas sensors may be utilized and machine learning implemented to detect prostate cancer. An exemplary device contains two sensor chips, where one chip contains four tin oxide sensors, and the other chip has four tungsten oxide sensors. Each sensor chip within the device contains two heaters and four detection zones, and therefore offers the benefit of several modes of operation including single or multi-sensor applications. Implementing multi-sensor mode will allow for more sensor features to be extracted from the sample data and therefore will be utilized for prostate cancer classification. One of the unique characteristics of this device is that the sensor heaters are located on the same plane as theAtt’y Docket No.: IUIC-168 sensing element (tin oxide or tungsten oxide). An image of the integrated MOx sensor device is shown in Figure 51. Identifying Nanosensor Sensitivity and Selectivity.

[0195] Even though the sensing layer will not be tuned to detect the specific biomarkers of prostate cancer, sensitivity and selectivity of the MOx nanosensor array will be qualified by testing the device with urinary VOCs including those previously identified as biomarkers for prostate cancer. The sensors will be tested with the VOCs in a background of air (80% nitrogen, 19% oxygen, <1% carbon dioxide). Purified air at a fixed flow rate will be bubbled through an aqueous standard of pure VOCs, and the resulting gas will be diluted to appropriate concentrations through controlling the flow rate. The gas will be exposed to the surface of the sensor, e.g., using a proprietary casing provided by NanoZ, and the VOCs will adsorb to the surface of the sensor. The sensing mechanism relies on ionizing molecular oxygen adsorbed onto the surface of the sensor. When VOCs are exposed to the MOx sensor, they either reduce the ionized oxygen or replace them through competitive adsorption, leading to a measurable change in conductivity, resistance or current. The resulting data is acquired and transmitted to a laptop containing, e.g., NanoZ software, where data can be exported. Pure standards of VOC biomarkers for prostate cancer will be tested at relevant and varying concentrations to develop calibration curves and determine the sensitivity and dynamic linear range of the eight sensors toward each analyte. Additionally, the time required to reach 90% saturation value (response time) and the time required to return to 10% of its value (recovery time) will be recorded. Selectivity of the sensor towards VOCs will be observed by plotting the sensitivity of the prostate cancer biomarkers and comparing them to each other and to other analytes in urine with high concentration (determined through the results of GC-MS analysis). Reproducibility and repeatability of the sensors will be tested by exposing multiple replicates of pure VOC standards (n ≥ 5). For the sensors to be reliably used for biosample analysis, the intraday and interday deviations should be within a margin of error (percent difference between the response in the presence and absence of VOCs) that will be determined in preliminary experiments. Reproducibility of sensors fabricated in different batches will also be qualified. Lastly, sensor degradation will be measured by placing the sensors in an ambient environment and measuring baselines and responses to VOC exposures over the course of one month. Accelerated degradation studies will also be conducted to determine the long-term stability of the nanosensors.Att’y Docket No.: IUIC-168 Quantifying the Nanosensor Array for VOC Analysis of Urine.

[0196] The sensors will be placed in the headspace of healthy urine standards for a predetermined time, which will be identified through optimization experiments. This will determine the shortest amount of time that the sensors need to be exposed to VOCs in urine headspace to obtain sufficient S / N and sensitivity. The sensor array will be purged with an air canister to remove any residual VOCs from the system between exposures. Pure standards and / or mixtures of volatiles that have been previously identified for prostate cancer will be spiked into healthy urine standards at appropriate concentrations, and the sensors will be tested. Machine learning algorithms described above (for analysis of the portable GC-MS data) will be implemented on nanosensor signals to distinguish the VOCs spiked into the healthy urine standard. Necessary adjustments will be made to reduce error and improve sensitivity, specificity, and accuracy of the nanosensor array. Such adjustments will be made to one or more of the following elements: sensors, sensor array, readout levels, analysis, or method of comparison with the standards. The nanosensor system will also be tested to determine the length of time required for the residual VOCs to be cleared from the device at a normal rate of airflow. The refinement process will be iterated until a satisfactory performance against predefined sensing quality requirements on sensitivity and specificity (>90% accuracy) are achieved. A prototype of the portable device is shown in Figure 52. The output of the sensors will be displayed through a readout circuit as instructed by the microcontroller; it will apply appropriate potentials to the sensor electrodes and measure the change in current for each sensor in the array. Data will be stored along with the sensor header in memory for further analysis. Testing Urine Samples and Data Analysis.

[0197] Once the nanosensors have been qualified to distinguish prostate cancer VOC biosignatures spiked in urine standards, they will be tested using the collected patient samples described in the Research Design of Example 2, supra. After sample analysis, sensor features will be extracted using an in-house algorithm developed in Python. Here, Matplotlib and NumPy libraries provide visualizations of the sample data, traverse through complex data sets, and implement calculations to extract sensor attributes. Additionally, the algorithm integrates elements of MATLAB to allow the user to create interactive graphical user interfaces (GUIs). Certain features of interest and their corresponding equations used for extraction can be seen in in Table 3, below.)Att’y Docket No.: IUIC-168 Table 3. Equations for Selected Features of Interest Features Equations, the eight sensors in the array. Univariate analysis through SAM will be implemented to initially screen for features that are significantly different between the sample classes of interest (prostate cancer positive control vs. negative control and indolent vs. aggressive cancer). Features implicated will be analyzed using the machine algorithms described above (SVM, RF, LDA, ANN, etc.) and the results will be cross validated, independently validated and functionally perturbed in a similar fashion to the portable GC-MS data.

[0199] Individual features of the sensors and machine learning algorithms will be correlated through univariate and multivariate regression analysis to the results of the portable GC-MS. For example, strong correlations are expected to be observed between the sensor features that distinguish prostate cancer / aggressive cancer and the individual VOC biomarkers identified by portable GC-MS. These strategies will help elucidate which specific VOC biomarkers are leading to sensor responses of interest, which is a significant downfall in other current studies using e-Nose devices to distinguish different forms of cancer. Example 4. Development of a Smart Device Application (SDA) and User-Friendly Interface Between Portable System and Clinicians, and Between System and End User.

[0200] Software Interface Design / Development and Data Computation for two VOC platforms will be developed for their respective devices: SDA Platform #1: A smart device applicationAtt’y Docket No.: IUIC-168 (with respective interfaces) will be developed to interface between the portable GC-MS and a tablet. The SDA will establish secure communication between the GC-MS (i.e., the Griffin G510) and tablet via Bluetooth. Input from clinicians (via the tablet interface) operating the SDA will provide the necessary patient data (e.g., biometric / vitals) and commands to the microcontroller to perform tasks such as system calibration and running samples prior to testing and transmitting the data to the SDA. Once VOC data is received (on the tablet) from the GC- MS, the SDA will implement the machine learning algorithm to classify the sample (prostate cancer screening). SDA Platform #2: A second (different) smart device application will be developed for the MOx sensor array and smartphone app (previously developed smartphone app for hypoglycemia shown in Figure 57) for single patient monitoring. The SDA will also be designed to be compatible with tablets for use by clinicians. The SDA will establish secure communication between the sensor device via Bluetooth with the user’s smartphone. The sensor platform interface will consist of an off / on button, charger port, battery light, and on / off light. Once the MOx sensor array is charged, turned on, and paired with the smartphone (via Bluetooth), the SDA will provide the necessary patient data. The sensor readout circuit will manage the sequence of data readings and perform any necessary sensor resetting or clearing function after the read-out or when it receives such an instruction from the on-board microcontroller. The microcontroller will process the data using a set of instructions provided to compute a VOC status value and transmit it to the smartphone. Once data is received by the smartphone, an algorithm will generate a data matrix, from which data and data visualization interfaces will be displayed via the smartphone app. The smartphone app interface will be designed and developed with the necessary communication features to display longitudinal data using data visualization and a warning display to alert the user of their current health status. Human-Centered Design and U.S. Food & Drug Administration Human Factors Validation of the SDA.

[0201] Iterative interface design and testing methods for all SDAs will be integrated into the prototyping and development process using human-centered design principles and practices. This includes (early stage) design and usability testing methods to validate all initial interface prototypes for both SDA platforms and associated devices. Initial design / testing processes will ensure that they are adaptable, intuitive, and user-friendly to maximize point-of-care (clinical) application and user (patient) ease of use. In accordance with FDA human factors and usability engineering standards, all (final stage) platform interfaces will include validation testing with all devices. The evaluation will identify both ease of use and user errors that might result inAtt’y Docket No.: IUIC-168 serious harm or misuse. Besides usability testing, human factors validation testing will include an assessment of product effectiveness regarding risk measures. Product Testing Details. SDA Testing on all device interface prototypes will consist of scenario-based usability time- on-task testing, followed by semi-structured interviews. Approximately 8–10 participants will be recruited for early-stage testing. Time-on-task measures time to execute task, success / fail rate of task, and error rate to execute select tasks. Test findings will identify the degree of ease of use and learning, effectiveness (ability to complete a task), and efficiency (amount of effort required to complete a task). Interviews will provide greater qualitative insights into a range of human-computer interaction issues and / or user challenges. Analysis of early-stage testing data will include descriptive statistics (with central tendencies), data trends, and standard deviations, comparing quantitative variables of time, error, etc. Semi-structured interviews will focus on ease of use (acceptability), satisfaction, and potential impact on improving clinical outcomes. Regarding data analysis, content analysis of qualitative data from transcribed interviews using computerized content analytics software (QDA DataMiner) will support organizing, retrieving, identifying, and quantifying word associations, themes, and concepts / meanings, with additional recognition of context, and associated meaning.

[0202] Final-Stage SDA Testing using FDA guided test methods for assessing all device interfaces will consist of a rigorous process that includes three parts. This is considered a simulated-use human factors validation test that is sufficiently realistic so that the results are generalizable to actual use and the intended user population. Part 1: Quantitative usability testing to assess: (1) Task action (i.e., a set of actions and sub-tasks) performed by the user (or test participant) to achieve a specifically defined goal, as part of the broader testing scenario. (2) User error identified by / through user action (or lack of action) that deviate from expected outcomes that: a) have the potential to produce results different from what the user anticipated, b) were not caused solely by device failure, and c) may result in harm to the user. (3) User safety that includes freedom from unacceptable user-related risk. As part of the larger human factors and usability risk management oversight of device development and testing, eliminating, or reducing design-related error will be a priority. This includes any form of a user interaction problem that contributes to or causes unsafe or ineffective use of the device. Part 2: Qualitative user assessments will consist of Contextual Inquiry and Semi-structured interviews. Contextual inquiry involves observing test participants as they use the device, while simultaneously askingAtt’y Docket No.: IUIC-168 them questions regarding their real-time interaction experience. For example, the researcher might ask the participant what they were doing at that moment and why they used the device the way they did. Semi-Structured Interviews will include open-ended questions that focus on user ease of use (acceptability), satisfaction, and potential impact on improving health outcomes— similar to the interview methods used in Early-Stage testing. Part 3: A Post-Test (Self-Reporting) NASA Task Load Index Questionnaire will be used to measure physical and mental workload, demonstrated by low to high cognitive effort and physical demand. Ideal findings should demonstrate the product’s design to minimize memory requirements and load. The questionnaire consists of six measures, using a Very Low / Very High 10-point scale: (1) The Mental Demand, (2) Physical Demand, (3) Temporal Demand, (4) Performance, (5) Effort, and (6) Frustration. Final-Stage Test Participants.

[0203] Twenty clinicians will be recruited for final-stage testing of the SDA Platform #1 devices (GC-MS and tablet). Twenty patient participants will be recruited for final stage testing of the SDA Platform #2 devices (MOx nanosensors and smartphone app). The FDA recommended sample size is 15, per Faulkner, L., Beyond the five-user assumption: Benefits of increased sample sizes in usability testing. Behavior Research Methods, Instruments, & Computers 2003, 35 (3), 379–383, who conducted a study that suggested that a sample size of 15 people was sufficient to find a minimum of 90% and an average of 97% of all problems with a software; a sample of 20 people was able to find a minimum of 95% and an average of 98% of the problems. Example 5. Detection of Hypoglycemic State in a Subject By Volatile Organic Compound Biomarkers Collected From a Sample of the Subject’s Breath. Identification and validation of hypoglycemia volatile-organic-biomarkers.

[0204] Methods and Materials: SPME GC-MS Analyses Tedlar gas sampling bags, 20 mL headspace vials, deactivated glass wool, and poly dimethylsiloxane / carboxen / divinylbenzene (PDMS / CAR / DVB) SPME fibers were purchased from Restek (Bellefonte, PA, USA). Alcohol, Parafilm, and Viromax filters were obtained from Fisher Scientific (Florence, KY, USA). An Agilent 7890A GC system coupled to an Agilent 7200 accurate-mass QTOF MS system with a PAL autosampling system were obtained from Agilent (Santa Clara, CA, USA) and CTC Analytics (Raleigh, NC, USA).3.1.2 Sensor Fabrication Round bottom borosilicate glass tubes were obtained from Fisher Scientific (Florence, KY, USA). Flowmeters withAtt’y Docket No.: IUIC-168 pressure control were purchased from Alicat (Tucson, AZ, USA). Stainless steel needles were obtained from Med-Vet International (Mettawa, IL, USA). Pure Nitrogen and dry medical air were purchased from Praixair (Danbury, CT, USA). Silicon wafers coated including Gold (90nm), Chromium (5nm), and Silicon Dioxide (300 nm) were fabricated by Hionix Inc (San Jose, CA, USA). Gold etchant, chromium etchant, EGNPs, PEI, and 1-Methyl- 2-pyrrolidinone (NMP) were obtained from Sigma Aldrich (St. Louis, MO, USA). S1813 positive photoresist and MCC Primer (80 / 20) were procured from MicroChem Laboratory (Round Rock, TX, USA). MF-321 developer was obtained from Rohm and Haas Electronic Materials LLC (Marlborough, MA, USA). Microcontroller-Arduino Bluno Nano (SKU DFR0296) was purchased from Digikey (Thief River Falls, MN, USA). The mask was printed by CAD / Art Services,Inc (Bandon, OR, USA). IM2NP-NANOZ and Nanoz Evaluation Kit Generation II was purchased by Nanoz (Rousset, FR). 2701 Digital Multimeter / Data Acquisition / Data Logging system was obtained from Keithley (Beaverton, OR, USA). Dual 4-channel analog multiplexer / demultiplexer 74HC4052D was obtained from Nexperia (Santa Clara, CA, USA). A LM324-N low power Quad- operational amplifier was purchased from Texas Instruments (Dallas, TX, USA). SolidWorks licensed purchased from SOLIDWORKS Corp. (Waltham, MA, USA). Form 3 Stereolithography 3D Printing and tough FLTOTL04 resin was used from Formslabs (Somerville, MA, USA). Keithley 4200-SCS was obtained from Tektronix (Portland, OR, USA). Ram´ehart contact angle goniometer model 200-F4 was obtained from Ram´ehart Instrument Co. (Succasunna, NJ, USA).

[0205] Patient Recruitment and Sample Collection: Hypoglycemia Subjects from ages 7–22 years old diagnosed with T1D and T2D were consented at Camp John Warvel (sponsored by the American Diabetes Association (ADA)) in North Webster, IN. Subjects were recruited by the team on the first day of camp. Subjects and parents were provided with writing and verbal consent detailing the process of the study. This research study was approved by St. Vincent Health Institutional Review Board (IRB ID: MOD00000311, Indianapolis, IN). Breath samples were collected in a Tedlar bag (80% full). Subjects were asked to provide breath samples at three different instances; 1) hypoglycemic event which was when the blood sugar (BG) of the subject was < 70 mg / dL, 2) fasting, and 3) resting which was collected when subjects joined the study. Environmental samples from the meal rooms, cabins, and outside were collected as background signals. Samples collected daily were transported to the laboratory at Indiana University-Purdue University Indianapolis (IUPUI).

[0206] Sample Preparation, Storage, and Processing: Breath samples were cryotransferred to headspace vials filled with deactivated glass wool using a previously published procedure

[0070] .Att’y Docket No.: IUIC-168 In summary, the system utilizes a vacuum line through stainless steel needles to transfer the VOCs from the Tedlar bag to the vial

[0015] . VOCs adsorb to the glass-wool due to the low temperatures created using dry ice (-40°C). After the cryotransfer process, breath samples were stored at -80°C until GC-MS analysis. All samples were analyzed using GC-MS quadrupole time-of-flight (QTOF). Samples were individually incubated and agitated (250 rpm) and extracted using the SPME fiber for 45 minutes (min) at 60°C. After extraction, the fiber was injected into the GC-MS QTOF for VOC separation, quantification, and identification. After each sample run, the SPME fiber was conditioned to decrease sample cross-contamination. A daily reference standard was run to monitor the performance of the GC-MS QTOF during the experiments.

[0207] As shown in Table 4, below, a total of n=37 subjects for Cohort 1 and n=51 subjects from Cohort 2 were recruited for the study. A total of 97 samples from Cohort 1 and a total of 133 samples from Cohort 2 were collected. The hypo glucose range was from 45–70 mg / dL and the non-hypo glucose range was 70–517 mg / dL. After cryothermally transferred to headspace vials, these samples were processed with GC-MS QTOF. Agilent suites were utilized to analyze the data and spectro align the chromatograms. A matrix was created and further analysis was applied. Table 4. Detailed information of the samples collected for hypoglycemia study Cohort 1 Cohort 2 Age range 7–22 7–22, ) 51

[0208] Agilent suites were utilized for the pre-processing of GC-MS data. MassHunter Quantitative Profinder (version B.08.00) was utilized to perform the deconvolution and spectral alignment taking into consideration data output such as retention time (RT) and mass-to-charge (m / z) ratios among the samples. The data was normalized using the Z-score method. EquationAtt’y Docket No.: IUIC-168 (3.1) shows the calculation used for the data normalization. The data was normalized to make all the distributions centered to zero. Several filters such as excluding exogenous VOCs and molecular features not detected in 50% of a sample class were applied for reducing data dimensionality.

[0209] Univariate statistical analysis was performed in Cohort 1 and Cohort 2 post-processed data. Cohort 1 (Training Data) had 481 qualified VOCs and Cohort 2 (Testing Data) had 516 qualified VOCs. Mann-Whitney U-test was used to find VOCs that had a p-value < 0.2 from both cohorts. A total of 40 VOCs with a p-value < 0.2 (shown in Table 5A, below) were found to be identical from Cohort 1 and Cohort 2. These VOCs had the same RT, m / z fragmentation patterns, and compound identification name (NIST). However, none of these VOCs alone were able to differentiate between Hypo and Normal with over 70% accuracy. The highest frequency of functional groups from these sets of 40 VOCs were noncyclic hydrocarbons, aromatic, alcohols, and unconjugated cyclic. The hierarchical clustergram shown in FIGs. 8A–8B represents expression for every VOC in every sample for Hypo and Normal data sets (both cohorts). There is a clear difference between the VOCs correlated to Hypo events than Normal events and how they cluster. Intra-class variation of Hypo and Normal was high and inter-class variation between cohorts was low. This means that there were similar patterns between the Hypo class for Cohort 1 and Cohort 2 and there was a distinct pattern for Normal for both Cohorts. This shows that the VOCs that are up-regulated in Hypo are down-regulated in Normal for both cohorts. This information led the way to further analyze the VOCs and understand how they differ and correlate to hypoglycemia. Table 5A. Volatile-organic-compound biomarkers for hypoglycemia, correlated with p-value < 0.2 aAtt’y Docket No.: IUIC-168 Decane, 3,3,6-trimethyl- 10.18 1072.0213 0.052610429 up p-Cresol 10.2 1073.267 0.165581782 down

[0210] Multivariate statistical analysis was performed on the 40 conserved VOCs with a p- value < 0.2. PCA was used as an unsupervised method to understand the patterns of the data. FIGs.17A–17B show that the samples do not perfectly cluster between the two classes Hypo (Red dots) and Normal (Green dots). Cohort 1 in PC1 had a percentage variation of 15.1% and PC2 of 8.7%. Cohort 2 in PC1 had percentage variation of 17.4% and PC2 of 9.8%. Cohort 1 was the training cohort, however it had higher degree overlap than Cohort 2. LDA was performed using forward feature selection to find the strongest predictive outcome that could separate Hypo from Normal with the highest AUC. As shown in FIG.19, a panel of six VOCs (shown in Table 5B, below) was able to distinguish between Hypo and Normal with an AUCAtt’y Docket No.: IUIC-168 > 0.90 for both cohorts. Cohort 1, the training data set, had an AUC of 0.98 and Cohort 2, the testing data set, had an AUC of 0.93. This means that the panel had a better performance at distinguishing between Hypo and Normal. The training AUC was higher than the testing AUC due to the fact that the panel was obtained by using the data from Cohort 1. The functional groups of the set of 6 VOCs were comprised of commonly noncyclic hydrocarbons, aromatic, and alcohols. There was no correlation between age, gender, and environmental background of this set of 6 VOCs. PCA of the 6 VOC biomarkers is shown in FIGs.18A–18B, the separation between classes for both cohorts increased compared to FIGs. 17A–17B. Further analysis of the BG and VOCs relationship needs to be studied. The discovery of these VOC biomarkers led to the other phase of this research, the development of a sensor array for the detection of the identified VOC biomarkers. Table 5B. Volatile-organic-compound biomarkers for hypoglycemia, with an experimental area under the curve (AUC) of > 0.9 H l mi H l miaun ng and es ng o Go d anopar ce Sensor Subs ra e.

[0211] Detection of VOCs occurs when the gas compounds are exposed to a sensing layer and their interaction causing a change in resistance. Here, the exposure of the gas to the sensor causes a change in sensing layer volume and either increases or decreases the distance of the conductive particles, leading to a measurable change in resistance. Interdigitated electrodes (IDEs) fabricated on a silicon substrate are the most common material used for this type of sensor. The IDEs are coated with the sensing material creating a sensing matrix. The change of resistance is measured by collecting the initial resistance (before VOC exposure) and exposed resistance (after VOC exposure). The sensor response is related to the properties and characteristics of the sensing material such as thickness, uniformity, and molecular structure. Different types of chemiresistive gas sensors have been used for the detection of VOCs related to health conditions, and these include MOX, graphene, carbon nanotube, GNP, and polymer- based sensor.

[0212] Advantages of functionalized gold nanoparticles (f-GNPs) include high sensitivity and the rapid detection of VOCs. The most common nanoparticles for biomedical applications areAtt’y Docket No.: IUIC-168 the EGNPs. These GNPs can be functionalized using thiols groups which form a network. These networks on the surface of the IDEs creates an space for the VOCs of interest to be targeted. The length of the thiol chain is a parameter that needs to be optimized in order to maximize the sensitivity and selectivity of the conductive material. As previously mentioned, some of the advantages of these sensors are their rapid response, low fabrication cost and time, low working temperatures, and high sensitivity and selectivity towards VOCs of interest. Size, shape, and length of the nanoparticles can determine the physiochemical properties. These advantages have allowed EGNP sensors to be applied in the medical field. The sensing mechanism is the of EGNPs activate when the VOCs are exposed to the sensor, and the VOC interact with the thiols causing a change in the electron tunneling. This means that the distance between nanoparticles increases because they adsorbed the VOCs and they become a part of the network.

[0213] Different methods have been used previously in the detection of VOCs

[0042] . First, univariate methods such as Mann-Whitney U-Test and multivariate methods such as PCA and LDA were utilized to analyze the data sets. Second, modeling methods such as the Gaussian process (GP) and Monte Carlo (MC) were used for modeling the sensor data obtained from testing VOCs of interest.

[0214] Mann-Whitney U-test is a univariate statistical test to find difference in two groups that do not have a Gaussian distribution. In other words, this test studies if the two samples are coming from the same population or not. Similar to the Student’s T-test, this method analyzes the source where the data is distributed and helps understand if the data has any relationship. In the case of comparing Hypo with Normal, if the VOC show a similar distribution with similar mean, it is possible that there is not difference in the samples that are identified for each of the data sets. However, if the VOC is found to be significant in two different sources then it can be considered as a targeted VOC. Mann-Whitney U-test different to t-test, it does not required or assume a normal distribution

[0044] ,

[0045] . Instead, studies the origin of the sources, first it studies the possibility to have a unique source, if not it means that the distribution was generated from two distinct sources.

[0215] Principal Component Analysis (PCA) is an unsupervised statistical method that implements data reduction to transform the raw data to orthogonal dimensions that can demonstrate where the maximum variation occurs among all the new dimensions

[0022] . For unsupervised methods including PCA, class labels are not included in the algorithm. Therefore, PCA is used as a tool to understand the patterns of the data. It gives an overview of the data behavior. PCA plots the raw data X (p) with the new data Y (k).Att’y Docket No.: IUIC-168

[0216] Equations demonstrate three conditions for the PCA:and 3) the variance is maximized

[0048] . When applying PCA to the data set, PC 1 always represents the axis with the highest variation and continues with the highest variation to the lowest variation. PCA has previously been used to investigate patterns in GC-MS data related to different diseases such as diabetes

[0015] ,

[0049] , and different types of cancer such as breast cancer

[0023] , lung cancer

[0050] , and head and neck cancer

[0051] . Linear Discriminant Analysis (LDA) on the other hand is a supervised statistical method that transforms the data to find the highest distance between the means to maximize variation between two defined sample classes

[0022] . In other words, it is a method that helps to understand the patterns of the data by clustering the samples to the class that corresponds in a new dimension

[0052] . This method has prior knowledge of the classes of the data. Similar to PCA, LDA finds to translate the data from raw data X to the new dimension Y by finding the optimal translation matrix W. There are two conditions for LDA: 1) maximize separation between classes (equation (2.4)), 2) LDA minimizes variation within samples in the same classes (equation (2.5)), and 3) the total variance is the sum of all class variances (equation (2.5)). To prevent overfit models, cross- validation can be implemented.Att’y Docket No.: IUIC-168

[0217] LDA has been used in multiple areas of research involving data collected with GC-MS such as diabetes different types of cancer such as breast cancer

[0023] , and lung cancer .

[0218] Among the wide variety of supervised learning methods, Gaussian process (GP) regression is one of the most popular, especially, in settings where the data is scarce. GP is a stochastic process in which every finite collection of its random variables has a joint multivariate Gaussian distribution. The possibility to write any collection of random variables as a multi-variate Gaussian distribution is helpful to develop powerful models for regression, classification, and optimization problems. GP regression has been used in a wide range of applications including the design of composite parts under dynamic loading the design of lithium-ion batteries, and multi-objective optimization. Monte Carlo (MC) methods are a set of numerical strategies that rely on repeated random sampling to solve problems that can be difficult to solve by other techniques. Popular applications of MC methods include optimization, numerical integration, and estimation of probability distributions.

[0219] Currently, researchers are working on the development of devices that could detect diseases in a non-invasive manner. In order to create a device that is user-friendly and capable of predicting, monitoring, or detecting the presence of a VOC in the breath, the sensor has to work independently. Further, the device must be low-cost and be able to deliver results in a point-of-care manner. Therefore, the integration of different electrical components coded with different software allows the creation of a breathalyzer. Wheatstone bridge (shown in FIG.24) have been reported to be used in the development of sensor interfaces to convert resistance into voltage. Voltage dividers or half-bridge are simple, however, higher temperatures can alter the identification of small changes in the resistance causing a problem when converting. One of the advantages of the Wheatstone bridge is that overcomes the failure of conversion of a small signal at temperatures due to the symmetry in the parameters. For the application of a hand- held device, the Wheatstone bridge allows the interface of the sensor to be minimized to reduce the size of the sensor. The integration of the communication unit permits the sensor interface to communicate with other devices via Bluetooth Low Energy (BLE). BLE advantages are the low consumption of power and the baud rate (115200 bits / seconds (sec)) which is the rate that the information transfers from the device to an app.

[0220] The EGNPs sensing mechanism is based on the quantum tunneling effect. The EGNPs create a conductive network utilizing the thiols functionalized on the material. When VOCs are exposed to the EGNPs, the sensing layer swells and cause the dielectric constant to change between GNPs. Resistivity either increases or decreases due to 1) distance between the GNP being higher or 2) change in dielectric constant. This sensing mechanism has been researchedAtt’y Docket No.: IUIC-168 extensively and a quantitative model shown in equation (3.2) has been created to calculate and estimate the resistance: where Ea= e2 / (8πrε0εr) is the activation energy. The tunneling decay constant is represented by βd. The edge-to-edgeresistivity of the monolayer protected gold nanoparticles film is represented by σ. Boltzmann constant is R, absolute temperature is T, and the radius of the GNP is r. The first part of the equations explains the probability that electrons travel from a nanoparticle to another one. The distance between nanoparticles is taken into consideration in the equation therefore if the thiol has a longer chain, then the film resistivity increases. Second, the equation explains the relationship between electrons and activation energy. When VOCs are exposed to the network of GNPs and thiols, and the dielectric constant for the VOC is higher than the dielectric constant for thiol, the activation energy reduces which causes a decrease in the film resistivity. Vice versa, when the dielectric constant of the VOC is lower than the dielectric constant for the thiol the film resistivity increases. The EGNPs alone as a sensing material for a breathalyzer are not ideal for the application of the sensor because they are water soluble. The sensor application is to detect VOCs from human breath. Therefore, a PEI polymer layer was added to the sensor to protect the EGNPs from humidity which decreases sensor material degradation. This creates a phenomenon called percolation theory where the PEI is the non-conductive network, and the EGNPs are the conductive network. The polymer material is permeable to gases when there is a high amount of conductive material. Therefore, when the VOCs reach the network of GNPs, the VOCs create a larger distance between the GNPs which leads to higher resistance. Metal oxide sensor substrate.

[0221] Metal oxide gas sensors resistance is altered by oxidation and reduction of the semiconductor. MOX along with other sensing materials such as ZnO, WO3, and In2O3 have been widely utilized for the application of resistive gas sensors. Tin oxide is a semiconductor with a wide-gap n-type which means that the material is more likely to be sensitive to VOCs and toxic gases such as CO and H2S. Sensing materials need to satisfy certain requirements in order to be use in MOX sensors. They need to be selective, sensitivity, and thermally stable. The life-cycle of the sensor is connected to the thermal stability because MOX devices operate at high temperatures compared to other chemiresistive sensors and this property causes the increase of signal through the reduction of VOCs on the surface of the MOX sensor. N-typeAtt’y Docket No.: IUIC-168 semiconductors such as SnO2 have high electron mobility (160 cm2 / V^s) which means they provide high stability, low synthesis cost, and high sensing properties.

[0222] Another sensing material that has been reported to be common for MOX sensors is WO3. MOX sensors operate in high temperatures in a range between 250°C to 500°C. WO3 has demonstrated that when the sensor is exposed to high humidity and the heater is working at 300°C there is an increase in resistance compared to other types of sensing materials. This is because humidity increases WO3 surface oxidation. One of the challenges that MOX sensors face is the low selectivity towards VOCs due to interference from environmental gases, however, WO3 is an n-type material that has shown a good selectivity toward the application of detection of VOCs. WO3is able to detect gases at low VOC concentrations when the sensor is working at high temperatures. The MOX sensor adsorbs oxygen, and when the sensor surface is heated at high temperatures it activates molecular oxygen. When the VOC is exposed, the oxygen and VOC interact decreasing the oxidation level which leads to a change in current. The current change is directly proportional to the concentration of the VOC exposed to the sensor. The oxygen desorbed decreases the resistance of the sensor (increases current). WO3 gas sensors have demonstrated that have optimal performance when they operate at 400°C and it is able to obtain readings from VOCs as low as 0.2 ppm.

[0223] For this example, an IM2NP-NANOZ sensor device Nanoz chips were obtained with two different sensing layers: NZGS002 is Tin oxide (SnO2) and NZGS003 is Tungsten oxide 35 (WO3). An innovation of this commercialized sensor is that the sensing layer, membrane, and heaters are in the same plane. Each Nanoz chip includes a total of four sensors coated either with WO3and SnO2and two heaters included as seen in FIGs.1A–1D.

[0224] The process of photolithography was performed in a class 100 Cleanroom. The procedure was simplified in ten different steps starting with 1) substrate preparation: cleaning the wafer rigorously with acetone, ethanol, isopropanol, and deionized (DI) water to remove any dust or impurities. This step was key for the success of the process because it allowed the chemicals to adhere to the surface and eliminated impurities and defects.2) Photoresist coating: the MCC Primer (80 / 20) was applied over the IDEs to improve the adhesion of other chemicals. It was spin-coated for 50 sec at 5500 rpm creating a uniform layer. The S1813 positive photoresist was applied after and was the light-sensitive material that was activated during exposure. Similar to the primer it was spin-coated for 50 seconds at 5500 rpm. 3) Soft bake: sensor was exposed to a hot plate for 10 sec at 80°C allowing the primer to cross link with the photoresist, subsequently enhancing the adhesion of chemicals to the surface. 4) AlignmentAtt’y Docket No.: IUIC-168 and exposure: the prepared substrate and the mask was exposed to high-resolution UV light, transferring the IDEs pattern from the mask to the substrate. The light was exposed for 5.6 seconds (this time was previously optimized). Exposure time was critical in order to obtain the right pattern dimension. 5) Development: the substrate was immersed in MF-321 Developer for 70 seconds (removing the excess photoresist) and cleaned with DI water to remove chemical excess.6) Hard bake: sensor was exposed to a hot plate for 10 min to finalize cross- linking and hardening of all chemicals.7) Inspection: sensor was inspected under the Keithley 4200-SCS microscope to verify the dimensions of the sensor and detect any impurities introduced through the process. 8) Etching: The sensor was immersed and shaken in the gold etchant for 30 sec removing the excess gold. It was cleaned with DI water. Then the sensor was immersed and shaken in chromium etchant for 30 sec to remove the excess chromium. 9) Photoresist strip: substrate was washed with acetone, ethanol, isopropanol, and DI water to remove the photoresist that was covering the pattern.10) Final inspection: this step checks the defects and impurities of the pattern before fabrication of the sensor.

[0225] PEI was utilized to create a hydrophobic layer to protect the conductive material and the sensor from high humidity. A mixture of PEI and NMP using a ratio of 1:3 (w / w) was created. The mixture was stirred for 48 hours at 80°C before use. The material converts in a gel-liked mixture with a clear yellow appearance. The EGNPs are dissolved in ethanol. The EGNPs used for the experiment had a particle size of 3.5 - 5.5 nm and were stored at -30°C. The ratio between the PEI and the EGNPs is important to increase the sensitivity of the material.

[0226] The fabricated IDEs need the incorporation of a sensing layer in order to be applicable. Therefore, the conductive material and polymer material were deposited on the surface of the IDEs. As shown in FIGs.2A–2C, the EGNPs were drop cast over the IDEs creating a layer of conductivity. The sensor is exposed to low pressure air to dry the EGNPs. The sensors were stored in a vacuum dry for 48 hours. This step is to reassure the adhesion of the EGNPs to the surface of the sensor. The PEI was spin coated over the IDEs and EGNPs at 5500 rpm for 50 seconds. These parameters were previously optimized to create a uniform layer of PEI

[0048] . The sensor was heat-treated at 80°C for 3 min to increase adhesion of materials and evaporate the NMP. The sensors were stored again in a vacuum oven at ambient temperature for 48 hours. After this procedure, the sensors were ready to be tested. To evaluate the hydrophobicity of the PEI-ENGP sensing material, a Ram´ehart contact angle goniometer model 200-F4 was used.2 µl of water was casted on the surface of the sensor and the contact angle was analyzed withAtt’y Docket No.: IUIC-168 ImageJ. This software takes the water drop contact angle using an ellipse and finds the best fit for the image.

[0227] Sensors were exposed to pure VOCs and real human breath. The sensor system was designed in a way to mimic human breath that includes relevant concentrations of humidity. The sensors system was built using six different components as seen in FIG.3; 1) Dry medical air (76.5%-80.5% Nitrogen, 19.5%-23.5% Oxygen) was the carrier gas to stabilize the sensor system. 2) Mass flow controllers were used to regulate the flow of the different gases. Three mass flow controllers were used to carry the dry medical air, VOCs, and humidity to the sensor. 3) VOCs and water were placed in round bottom flasks and sealed to prevent VOCs or humidity from escaping.4) The gas mixer allowed the gases to combine and create the simulated breath. 5) Testing chamber encapsulated the sensors to decrease the amount of interaction with the atmosphere. 6) Data was collected for the PEI-EGNP sensor using the Keithley KickStart software with a RS232 interface and for the Nanoz sensor, the Nanoz configuration tool software was implemented. The data was saved in Excel (.csv format). These files were input into a python code that imported the raw data and calculated the change of current for each sensor and the slope. This was by finding the minimum and maximum points for each peak and each point was averaged with the next ± 10 points. This code is being validated by comparing results with the data analyzed manually for future use. These files were processed manually using the same methods.

[0228] In summary, the system carries the dry medical air throughout the different glass tubes at a fixed rate programmed by the mass flow controller. Then the VOCs are bubbled to the gas mixer with the humidity and air. Once the sensors were stabilized only with dry medical air and humidity, the VOC was exposed to the sensor for a fixed time (1 min) and the sensor was brought back to baseline with dry medical air. This procedure was performed for three different RH levels; 0%, 45%, and 85% at room temperature (21°C – 25°C). The RH levels and temperature were validated using a CO2 sensor. The sensors were tested in a range of concentrations (200 ppb – 20 ppm).

[0229] Equation (3.3)

[0100] describes the methodology to find the desired concentration of the VOC with respect to the vapor pressure of each VOC and the total flow of the system. The flow of the system was constant for all experiments at 1.000 standard liters per minute (SLPM). The Nanoz sensor was tested with VOC 1 to VOC 9 which are gases identified as VOC biomarkers for hypoglycemia and COVID-19. The PEI-EGNP sensor was tested using VOC 10. Optimization of the commercialized sensor Nanoz was conducted in order to find theAtt’y Docket No.: IUIC-168 optimal parameters that the sensor work for high humidity environments such as breath. The Nanoz sensor was exposed to VOCs under low and high humidity levels. The heater voltage was optimized to enhance sensitivity and selectivity of the sensor.SPME) fibers coated with polyvinylidene fluoride-carbon black.

[0231] Materials and Instrumentation: Polyimide-coated optical fibers with a fused silica substrate (200 µm diameter) were obtained from Ocean Optics (Orlando, FL, USA). PVDF was purchased from Arkema Group Kynar (Colombes, France), CB (Black Pearl 2000) was acquired from the Cabot Corporation (Boston, MA, USA), and NMethyl-2-pyrrolidone (NMP) was obtained from Sigma Aldrich (St. Louis, MO, USA) for fabricating the polymeric composite. Human urine standards (UTAK Laboratories; Valencia, CA, USA) were aliquoted into 40 mL glass vials with screw cap lids (Sigma Aldrich) for VOC analysis. A Digital Dry Bath / Block Heater (Thermo Fisher Scientific; Waltham, MA, USA) was used to incubate and heat urine samples for SPME. SPME fiber assemblies were used for fabrication and were purchased from Restek (Bellefonte, PA, USA). Urinary VOCs concentrated by the SPME fibers were chromatographically separated and identified using an Agilent (Santa Clara, CA, USA) 7890A GC with a Rxi-5ms column (Restek) coupled to a 7200 Accurate-Mass Quadrupole time-of-flight (QTOF) MS system. For electrospinning SPME fibers, a syringe pump was purchased from KD Scientific (Holliston, MA, USA) and blunt needles were obtained from SAI Infusion Technologies (Lake Villa, IL, USA).

[0232] Polymer Preparation: Polymer suspensions of PVDF composited with CB were prepared in NMP at a concentration that was previously optimized to increase the hydrophobicity of the stationary phase38 . The optimal suspension utilized for coating SPME fibers prepared PVDF and CB at 13% w / w (weight by weight) and 1.875% w / w in the organic solvent respectively. The polymer suspensions of PVDF-CB were stirred and heated to 90 °C for 24 hours to produce homogenized solutions. The optical fused silica fibers were cut to an applicable length for the SPME fiber assembly, and the polyimide coating was removed from 2 cm of the fiber through burning for ~ 45 seconds which exposed the bare silica substrate. After removing the preexisting coating, the fiber was rinsed in ethanol to remove any contaminants or stray polyimide from the surface. Once rinsed, the fiber was dried at 60°C in a vacuum oven for at least two hours to evaporate any residual solvent from the substrate. TheAtt’y Docket No.: IUIC-168 pre-cleaned and dried SPME fiber substrate was used for both dip coating and electrospinning the PVDF-CB solution.

[0233] Dip Coating and Electrospinning SPME Fibers: SPME fibers were fabricated through dip coating and electrospinning. Dip coating was achieved by slowly lowering the SPME fiber into the polymeric suspension, where the stationary phase was physically deposited onto the fiber surface. For electrospinning, the polymer composite was loaded into a 10 ml syringe with a needle size of 23 G and was arranged in a horizontal configuration relative to the conductive target (stainless steel) with a working distance of 15 cm. Electrospinning was carried out in ambient conditions and all the experimental parameters were kept constant during the process. The polymer was pushed through the syringe using a pump, and a positive 17 kV D.C. voltage was applied to the tip of the needle which ejected the composite toward the stainless-steel collector (80 cm x 20 cm x 0.2 cm). The fused silica SPME fiber was placed in an in-house built rotor which rotated the SPME fiber in front of the conductive collector at a rate equal to 45 RPM. As a result, the PVDF-CB electrospun fibers were deposited uniformly onto the conductive collector and the SPME fiber surface. The platform is driven by a Nema 17 stepper motor, characterized by its step angle of 1.8 degrees, current rating of 2 Amperes, and holding torque of 59 Newton-centimeters. The motor operates at 12 V and is controlled by an Arduino microcontroller along with a DM542T stepper motor drive. Next, dip coated and electrospun SPME fibers are dried in a vacuum oven at 60 °C for at least two hours to evaporate excess NMP solvent, and thermally conditioned at 260 °C for a total of 30 minutes to desorb any VOC contaminants on the surface. The SPME fibers were immobilized to commercially available SPME fiber hubs using epoxy. The hub was screwed into a standard SPME fiber assembly and used for urinary VOC extraction and analysis. An illustration of the dip coating and electrospinning processes are shown in FIG.35.

[0234] Electrospinning Parameter Optimization: The optimization of electrospinning parameters is of great significance to produce high quality fibers with the desired characteristics and properties. The quality of electrospun fibers is affected by several factors, including the applied voltage, flow rate, viscosity of the polymer solution, needle size, and distance between the needle tip and the collector (working distance). To optimize these parameters, systematic experimental design was undertaken. The optimization of the applied voltage in these experiments was achieved by varying the voltage applied to the tip of the needle (17 kV to 20 kV in increments of one kV). High voltages with a high electric field strength generally result in electrospun fibers with large diameters, while a low voltage produces fibers with a lower diameter. Flow rate is another significant parameter that was optimized (0.15 mL / hr to 0.25Att’y Docket No.: IUIC-168 mL / hr tested in increments of 0.05 mL / hr) as it can influence fiber deposition, beading and diameter. Other parameters which were explored include the viscosity of the polymer solution, working distance and needle size. For example, a solution with relatively low viscosity will produce electrospun fibers at a high rate but may result in thin and non-uniform fibers. Highly viscous solutions on the other hand are more likely to generate fibers with improved diameter and uniformity, but with reduced flow rate of the polymer through the needle. Needle size also plays a role, as larger needle sizes result in a higher fiber diameter and flow rate, while a small needle size produces fibers with a lower diameter and flow rate. Working distance is also a parameter of interest because large distances produce fibers with a lower diameter and electric field strength, while a small distance yields fibers with a relatively higher diameter. Finally, the electrospinning time onto the solid phase microextraction (SPME) fiber was optimized by varying the duration (5-, 10- and 15-minutes were tested).

[0235] HS-SPME GC-MS and Data Analysis: Electrospun and dip coated PVDF-CB SPME fibers were conditioned daily at 260 °C for 10 minutes to remove any VOCs on the fiber before analysis. The urine standard was aliquoted into 40 mL vials and stored in a -80 °C freezer. Samples were thawed 30 minutes prior to HS-SPME GC-MS analysis and saturated with NaCl to increase the ionic activity and help partition VOCs into the sample headspace. SPME fibers were inserted through the septum of the vial into the sample headspace while the urine sample was heated at 60 °C for 45 minutes. The SPME fiber was manually injected into the inlet of the GC-MS system kept at 250 °C for 4.5 minutes to thermally desorb VOCs. An Ultra High Pure helium mobile phase was utilized with a flow rate equal to 1.2 mL / min, and the mass transfer line was held at 250 °C. The oven temperature was held at 40 °C for the first 2 minutes of the run. After, the temperature was ramped to 100 °C at a rate of 8 °C / min, followed by a 15 °C / min ramp to 120 °C, 8 °C / min to 180 °C, 15 °C / min to 200 °C and finally an 8 °C / min ramp to 260 °C. Urine samples were analyzed in triplicate for each of the experiments. Chromatograms were integrated in Agilent MassHunter Qualitative Navigator, and VOCs with high relative abundance (or previously identified) were quantitatively assessed. SPME fiber sensitivity was explored for optimizing the time SPME fibers were electrospun, and to identify differences in performance between the dip coated and electrospun fibers.

[0236] Characterization of Materials: A Nicolet iS10 Fourier Transform Infrared (FTIR) spectrometer was used for chemical characterization of electrospun fibers deposited on a potassium bromide (KBr) disc. Omnic software (Thermo Fisher Scientific; version 8.2) was used to process, analyze, and measure the FTIR peaks of electrospun PVDF and PVDF-CB to determine the effects of CB addition on the structure of the polymer. A JSM-7800F fieldAtt’y Docket No.: IUIC-168 emission scanning electron microscope (FESEM) manufactured by JEOL (Peabody, MA, USA) was used to observe the production and measure the diameter of electrospun fibers on foil in the electrospinning parameter optimization experiments, and to visualize the differences in the surfaces of dip coated and electrospun SPME fibers. The surface of the fibers was sputtered with evaporated gold in an atmosphere of argon for a total of 45 seconds utilizing a Denton Desk V (Denton Vacuum LLC, Moorestown, NJ, USA). FESEM imaging was undertaken on the fibers at a working distance of 10 mm and an accelerating voltage equal to 10 kV. To characterize the hydrophobicity of the materials, the polymeric suspensions were coated on to a glass slide and a Raméhart goniometer model 200-F4 (Succasunna, NJ, USA) was used to measure the contact angle.

[0237] Optimizing Parameters for Electrospinning: To optimize the electrospinning parameters, including applied voltage and flow rate, the electrospun PVDF-CB fibers fabricated on aluminum foil were imaged using FESEM. The results of the experiments demonstrated that fiber deposition improved as the applied voltage increased from 17 kV to 20 kV. However, it was found that applied voltages above 17 kV resulted in fiber beading, which reduces the surface area and leads to a nonuniform coating. For these reasons, 17 kV was selected as the optimal applied voltage. Additionally, the flow rate of the syringe pump was optimized by testing values ranging from 0.15 mL / hr to 0.25 mL / hr. The results showed that increasing the flow rate of the syringe pump increased the coverage of fiber deposition and decreased fiber beading, leading to the determination of the optimal flow rate being equal to 0.25 mL / hr. The optimal viscosity was produced in a polymeric suspension of PVDF and CB (13% PVDF and 1.875% CB relative to NMP solvent), which was identified in a previously published study. Alternative PVDF and CB concentrations were investigated but did not show improved fiber deposition or decreased fiber diameter. Other parameters that did not have a significant impact on the quality of the electrospun fibers included needle size (23 G needles implemented in the following experiments) and working distance (fixed at 15 cm).

[0238] Optimizing Electrospinning Time Prior to identifying how electrospinning PVDF-CB stationary phases increases urinary VOC sensitivity relative to dip coating, the time that the SPME fiber was electrospun was optimized. SPME fibers were coated with the PVDF-CB composite using the optimal conditions presented in the previous section (applied voltage = 17 kV, flow rate = 0.25 mL / hr) for 5-, 10- and 15-minutes. Urine was analyzed by HS-SPME GC- MS, and the integrated signals of chromatographic peaks corresponding to VOCs with the largest signal (along with previously identified VOCs) were quantified. (See Woollam, M.; Grocki, P.; Schulz, E.; Siegel, A. P.; Deiss, F.; Agarwal, M., Evaluating polyvinylidene fluorideAtt’y Docket No.: IUIC-168 - carbon black composites as solid phase microextraction coatings for the detection of urinary volatile organic compounds by gas chromatography-mass spectrometry. Journal of chromatography. A 2022, 1685, 463606). A sample chromatogram of one of the SPME fibers which was electrospun for 5 minutes is shown in FIG. 36. The first 10 minutes of the chromatographic run did not display a high number of urinary VOCs and mostly consisted of silanes / siloxanes (thermal degradation products of SPME GC-MS) and / or VOCs in the background including carbon dioxide. However, the last 15 minutes of the sample chromatogram is rich with VOCs emanating from the urine standard. 2,4- dimethylbenzaldehyde had the largest signal, and this is consistent with our previous analyses. Other VOCs with large signal were comprised of carvone, 2,5-di-tert-butylbenzoquinone and dibutyl phthalate. Previously analyzed volatiles were also detected and included p-menth-1-en- 3-one and 2,4-di-tert-butylphenol. To identify the optimal electrospinning time for PVDF-CB fiber deposition on to SPME, the integrated signals of four VOCs (2,4-dimethylbenzaldehyde, carvone, p-menth-1-en-3-one, and 2,4-di-tert-butylphenol) were quantitatively compared. The bar charts shown in FIGs.37A–37D demonstrate that increasing the electrospinning time from 5 to 10 minutes increases VOC sensitivity. The signal of 2,4-dimethylbenzaldehyde increased by a factor of 4.9, and other VOCs showed greater increases in sensitivity (carvone by a factor of 8.7, p-menth-1-en-3-one by a factor of 8.6, and 2,4-di-tert-butylphenol by a factor of 9.1). Upon increasing the time to 15 minutes, all the VOC signals in FIGs. 37A–37D significantly decreased. For example, 2,4-dimethylbenzaldehyde sensitivity decreased by a factor of 1.6 (FIG.37A) and the other VOCs were minimized approximately by a factor of 2.5 (FIGs.37B– 37D). Therefore, the team determined that PVDF-CB fiber deposition was optimal when the suspension was electrospun for 10 minutes, which was compared to the previously published dip coated PVDF-CB SPME fiber in the following section.

[0239] Dip Coated and Electrospun SPME Fiber Comparison After the duration of electrospinning was optimized, the SPME fiber electrospun with PVDF-CB for 10 minutes was compared to the fiber fabricated through dip coating. Initially before analyzing individual VOCs, the number of VOCs detected, and the total integrated signal were quantified (FIGs 38A–38B). This preliminary analysis indicated that electrospinning does not increase the number of VOCs detected, but significantly increased the total integrated GC-MS signal by a factor of 2.4. Next, the same VOCs presented in FIGs 37A–37D were analyzed and their integrated signals were quantified in both SPME fiber types. Bar charts for these VOCs can be observed in FIGs. 39A–39D and demonstrated that all four analytes were detected with significantly higher sensitivity when extracted using the electrospun PVDF-CB SPME fiber.Att’y Docket No.: IUIC-168 p-Menth-1-en-3-one showed the lowest increase in extraction efficiency, as the signal for this VOC was only enriched by a factor of 1.5. 2,4-dimethylbenzaldehyde and carvone showed similar enhancements in VOC sensitivity with factors of 2.4 and 2.6 respectively. However, the VOC with the highest improvement in signal when analyzed using the electrospun SPME fiber was 2,4-di-tert-butylphenol, as this compound had a higher signal by a factor of 3.4. Overlayed interval plots for other VOCs with lower abundance (or identified in previous work) are shown in FIG. 40 and include dibutyl phthalate, 2,5-di-tert-butylbenzoquinone, decalactone, p-cresol, benzaldehyde, acetophenone, indole, thymol, and other analytes. These plots show that these compounds also have significantly higher signals when analyzed using the electrospun SPME fiber. Extraction efficiency was enriched by factors ranging from 2.0 (acetophenone) to 5.5 (indole) for this alternative set of urinary VOC analytes. A hierarchical heatmap was generated for a larger set of 45 urinary VOCs detected using both fiber types to visualize their autoscaled GC-MS signals in all urine samples, and this can be observed in FIG. 41. Samples analyzed by the dip coated and electrospun fibers are shown on the x-axis (columns), while the 45 VOCs are illustrated on the y-axis (rows). Within the heatmap itself, red signifies relatively high signals, black denotes average values and green represents relatively low VOC signals. It was observed that there was high interclass and low intraclass variation between the two PVDF-CB fiber types, indicating a significant increase in VOC sensitivity when extracting the analytes using the electrospun fiber. To further probe intraclass variation and reproducibility in a quantitative fashion, relative standard deviation (RSD) values were calculated for all VOCs presented in FIGs. 39A–39D, 40–41. An overlayed box and whisker plot with data distributions for the RSD values can be observed in FIG. 42. Even though many of the RSD values for the dip coated and electrospun SPME fibers overlap, a paired Student’s T-test revealed that dip coated fibers had significantly higher RSD values (p < 0.001), indicating that the electrospun fiber had superior reproducibility.

[0240] Fourier-transform Infrared Spectroscopy Characterization: To chemically characterize the electrospun fibers, FTIR spectroscopy was undertaken on PVDF and PVDF-CB to identify isomeric changes in the polymeric structure due to the addition of CB. It is important to note that PVDF is a mixture of different isomeric phases (α-phase, β-phase, and γ-phase), and FTIR spectroscopy can probe isomeric changes in the PVDF stationary phase. Overlayed FTIR spectra obtained for electrospun PVDF and PVDF-CB are shown in FIG. 43, and there was a lack of fluctuation in peak intensity between PVDF and PVDF-CB. PVDF shows FTIR peaks originating from CF2stretching in the α-phase at 880 cm−1, CH2rocking arising from the β- phase at 842 cm-1, asymmetric CF2 stretching at 1176 cm-1, vibrations arising from the γ-phaseAtt’y Docket No.: IUIC-168 at 1231 cm-1, and CH2 wagging at 1405 cm-1. These FTIR peaks arising from the different isomeric phases do not fluctuate in intensity when CB is composited with PVDF, suggesting no unintentional structural changes in the polymer were observed due to the addition of CB.

[0241] FESEM Imaging of SPME Fibers: Optimizing the time PVDF-CB was electrospun on to SPME fibers demonstrated that 10 minutes is optimal, and FESEM imaging was utilized to provide a scientific rationale for this result. FIG. 44 shows SEM images of SPME fibers electrospun for different durations (5-, 10- and 15 minutes). It can be observed that increasing the electrospinning time is positively correlated with the abundance of electrospun fibers on the surface of the SPME fiber. For an example, when electrospun for 5 minutes, the PVDF-CB fibers do not entirely cover the substrate as the fused silica fiber substrate is still visible and the substrate becomes decreasingly visible as the electrospinning time is increased. The FESEM images also demonstrate that the diameters of the electrospun fibers immobilized onto SPME fibers are relatively homogeneous, even between the different electrospinning times themselves. Fiber diameters between the different electrospinning durations were 1.4 µm for 5 minutes, 1.5 µm for 10 minutes and 1.2 µm for 15 minutes. In the SEM images of the SPME fibers electrospun for 15 minutes, the PVDF-CB fibers appear to be damaged and aggregated, which may be the reasoning for its decreased performance. Lastly, SEM images are presented for the dip coated and electrospun SPME fibers in FIG. 45. These images also show that the electrospun fibers had a diameter of approximately 1.5 µm and had increased surface area relative to the dip coated SPME fibers.

[0242] Contact Angle Measurements: Static contact angle measurements were implemented for dip coated and electrospun fibers to determine the wettability and hydrophobicity of the different sensing layers. Both PVDF and PVDF composited with CB were dip coated and electrospun with the optimized parameters to observe the effects of CB addition and electrospinning on hydrophobicity. The static contact angle measurements between a 2 µL droplet of water and the stationary phases are presented in FIG.46. The contact angle for PVDF without the addition of CB was measured to only be approximately 81.57° when electrospun. The contact angle for PVDF when electrospun was determined to be slightly lower when compared to dip coated PVDF sensing layers. When PVDF is composited with CB, the contact angle for both the electrospun and dip coated composites is significantly increased. Dip coated PVDF-CB displayed a contact angle equal to 96.83° while the electrospun PVDF-CB had an even higher contact angle of 102.67°. This demonstrates that both compositing PVDF with CB and the electrospinning process itself can increase the hydrophobicity of the SPME coatings.Att’y Docket No.: IUIC-168

[0243] Compounds may be differentiated on a sensor platform by use of, e.g., predicted functional-group binding energies, such as shown below in Table 6: Table 6. Predicted Function Groups, Binding Energies (BEs), and Elemental Atom Percentage of Polyetherimide / Carbon-Black Composite With and Without Heat Treatment Untreated Heat Treated. rch which showed that dip coating SPME fibers with in-house stationary phases, utilizing them to extract VOCs in the headspace of a urine standard, and analyzing them by GC-MS is an efficient method to evaluate how sensing layers adsorb an array of analytes in urine with high efficiency. Our previous work is extended to further show that HS-SPME GC-MS is an efficient technique to determine how electrospinning PVDF-CB on to SPME fibers increases the sensitivity for urinary VOC detection by GC-MS compared to dip coating. Illustrations of the two immobilization methods can be observed in FIG.35. Both techniques relied on physical adhesion between PVDF-CB and the fused silica SPME fiber substrate, which can be effectively accomplished because the high surface energy of the substrate overcomes the low interfacial tension of the polymer composites. Before determining how electrospinning enriches SPME efficiency, a series of optimization experiments were undertaken. First, the parameters which were utilized for the electrospinning process were systematically optimized. Varying the applied voltage from 17kv to 20 kV did not change the diameter of electrospun fibers and increasing the applied voltage enhanced fiber deposition but also raised the probability of beading. The flow rate of the syringe pump was varied from 0.15 mL / hr to 0.25 mL / hr and showed optimal electrospun fiber fabrication at higher flow rates.

[0245] Once the electrospinning parameters were optimized, SPME fibers were immobilized with the electrospun PVDF-CB for 5-, 10- and 15 minutes to identify an optimal duration for coating (sample chromatogram for the results shown in FIG.36). The rationale for optimizing the electrospinning time is that increasing the duration will increase the number of fibersAtt’y Docket No.: IUIC-168 deposited onto the surface of the fused silica substrate. GC-MS results for four analytes (2,4- dimethylbenzaldehyde, carvone, p-menth-1-en-3- one, and 2,4-di-tert-butylphenol) in this experiment are shown in FIGs. 37A–37D and indicated that the highest VOC sensitivity was obtained when electrospinning for 10 minutes. Increasing the electrospinning duration from 5- to 10 minutes drastically enhanced VOC signal, but when the time was further increased to 15 minutes, the VOC signal significantly decreased. The reason for this observation is hypothesized to be due to the electrospun SPME stationary phase being too thick when coated for 15 minutes. If the deposited coating is too thick, the stainless-steel syringe (where the SPME fiber is housed) may damage the stationary phase during exposure and retraction. This hypothesis is supported by the FESEM images shown in FIG. 44, which display that the electrospun SPME fibers were damaged after HS-SPME GC-MS analysis. Therefore, it was determined that electrospinning the SPME fibers with PVDF-CB for 10 minutes was optimal, and was utilized in the experiments aiming to compare the performance of electrospun and dip coated fibers. Electrospun and dip coated PVDF-CB SPME fibers were compared by assessing the number of VOCs detected and the total integrated GC-MS signal (FIGs. 38A–38B). The number of VOCs was not significantly different between the two SPME types, and the total integrated signal was significantly increased when isolating urinary VOCs utilizing the electrospun SPME fiber. This preliminary indicated that the electrospun SPME fiber had much higher sensitivity. Furthermore, no major differences in the number of VOCs detected is expected, as both fiber types have the same chemical composition. To further probe differences between electrospinning and dip coating, the same four VOCs shown in FIGs. 37A–37D for the electrospinning duration optimization experiment were quantified and compared (FIGs. 39A–39D). The results showed that these four VOCs had a higher signal when analyzed using the electrospun fibers, with enrichment factors ranging from 1.5 to 3.4. Additional VOCs that either had high signal or were previously identified are shown in FIG. 40 which displayed similar results. Hierarchical heatmaps were also generated for a larger set of 45 urinary VOCs in FIG.41, and further proved that the electrospun fibers had significantly increased signal for an abundant number of volatile chemicals in urine. Analysis of how the VOCs and associated functional groups clustered within the heatmap was not undertaken as the volatile analytes displayed similar patterns in the sample data (all VOC signals enriched when using the electrospun SPME fiber). This is expected as the electrospinning method should not inherently change the selectivity of the SPME stationary phase because there is no change in chemical composition (the same PVDF-CB suspension was utilized for both fiber types). Lastly, the dip coated SPME fiber presented significantly higher RSD values (FIG. 42), indicating theAtt’y Docket No.: IUIC-168 electrospun SPME fiber had superior VOC reproducibility. Chemical and physical techniques were utilized to characterize the PVDF-CB sensing layers and help understand why electrospun SPME fibers had increased sensitivity to extract urinary VOCs. First, FTIR spectroscopy was implemented to identify any changes in the isomeric phase of the polymer upon addition of CB to the suspension (FIG. 43). Previous studies using poly(vinylidene fluoride- hexafluoropropylene) as sensing elements have demonstrated that CB addition changes the FTIR peaks as the polymeric crystal structure undergoes isomeric change. However, FTIR characterization of electrospun PVDF and PVDF-CB in this study and previously published reports showed that there was no change in the FTIR spectra and therefore the polymeric framework did not undergo any changes in the crystal structure. Next, physical characterization was implemented through FESEM imaging to compare the electrospun and dip coated SPME fibers (FIG. 45). These images showed that electrospinning PVDF-CB increases the surface area of the SPME fiber, and therefore this is hypothesized to be the primary reason why the electrospun SPME fibers had higher ability to adsorb urinary VOCs. FIGs. 44–45 also show that the diameter of the electrospun PVDF-CB fibers onto SPME had an approximate value of 1.5 µm. Static contact angle measurements were also used to measure the hydrophobicity of electrospun, and dip coated PVDF-CB stationary phases (FIG. 46). PVDF was also characterized as our previous study indicated that CB addition increased sensing layer hydrophobicity. The same trends were identified in this study, and it was also observed that electrospun PVDF-CB had higher hydrophobicity when compared to dip coated.

[0246] Taken as a whole, the results presented help validate that GC-MS can evaluate how SPME fibers adsorb a wide range of VOCs in urine with high analytical throughput relative to testing sensors which would require analyzing individual analytes at a time. Additionally, the results also determined that electrospinning PVDF-CB polymer composites can increase GC- MS sensitivity and reproducibility relative to dip coating. The presented methodology will facilitate sensor development by quickly evaluating different dip coated or electrospun sensing elements and tuning them with the desired properties to sensitively detect VOCs for an array of applications. .

Claims

Att’y Docket No.: IUIC-168 IN THE CLAIMS:

1. An apparatus for detecting cancer in a subject from volatile-organic-compound biomarkers present in the subject’s biofluid, the apparatus comprising: a sensor array of two or more sensor subunits, each sensor subunit operable to detect at least one volatile-organic-compound biomarker emitted from a sample of the subject’s biofluid, wherein each sensor subunit is selectively chemiresistive in response to exposure to a chemical of a functional group selected from: noncyclic hydrocarbon, unconjugated cyclic hydrocarbon, alcohol, aldehyde, amide, aromatic, carbonyl, carboxylic acid, ester, ether, ketone, or terpene; and wherein contacting the sensor subunit with a volatile-organic-compound biomarker having a functional group for which the sensor is selectively chemiresistive causes the sensor subunit to generate an electronic signal; wherein each sensor subunit corresponds independently to a signal output channel, each signal output channel operably connected to a computation module for receiving each of the electronic signals; and wherein the computation module is programmed to interpret an aggregate of electronic signals as corresponding to a subject cancer state.

2. The apparatus of claim 1, wherein the computation module is further programmed to interpret an aggregate of electronic signals as corresponding to a cancer aggressiveness state.

3. The apparatus of claim 1, wherein the cancer is prostate cancer.

4. The apparatus of claim 1, comprising a sensor subunit selectively chemiresistive for an unconjugated cyclic organic compound.

5. The apparatus of claim 1, comprising a sensor subunit selectively chemiresistive for an aromatic compound.

6. The apparatus of claim 1, comprising a sensor subunit selectively chemiresistive for a carbonyl-bearing compound.Att’y Docket No.: IUIC-168 7. The apparatus of claim 1, comprising a sensor subunit selectively chemiresistive for a ketone.

8. The apparatus of claim 1, comprising a sensor subunit selectively chemiresistive for a terpene.

9. The apparatus of claim 1, comprising a first sensor subunit selectively chemiresistive for unconjugated cyclic compounds, a second sensor subunit selectively chemiresistive for carbonyl-bearing compounds, and a third sensor subunit selectively chemiresistive for terpenes.

10. The apparatus of claim 1, wherein each sensor subunit is operable to detect a change in concentration from a baseline threshold one or more of: β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, 2,3,4,5-tetramethyl-2- cyclopenten-1-one.

11. The apparatus of claim 1, wherein an electrical signal corresponding to an change in concentration from baseline of any of: unconjugated cyclic compounds, aromatic compounds, carbonyl-containing compounds, ketones, alcohols, terpenes, esters, and amines, or any combination thereof, correlates with the presence or absence of prostate cancer in the subject.

12. The apparatus of claim 1, wherein the wherein an electrical signal corresponding to an change in concentration from baseline of any of: β-thujone, dimethyl trisulfide, α- curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, 2,3,4,5-tetramethyl-2- cyclopenten-1-one, or any combination thereof, correlates with the presence of prostate cancer.

13. The apparatus of claim 1, wherein the wherein an electrical signal corresponding to an change in concentration from baseline of any of: thymol, 6-amyl-α-pyrone, 2,3,4,5- tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, allyl-2-furoate, or any combination thereof, correlates with the presence of aggressive prostate cancer.Att’y Docket No.: IUIC-168 14. The apparatus of claim 1, further comprising an inlet piece operable to receive the volatile-organic-compounds emitted from the sample of the subject’s biofluid, the inlet piece operably coupled to a housing, the sensor array disposed inside the housing.

15. The apparatus of claim 1, wherein the biofluid is urine.

16. The apparatus of claim 1, wherein the subject is a human.

17. The apparatus of claim 16, wherein the subject is biologically male.

18. The apparatus of claim 1, wherein the subject is a human who is diagnosed as having a high risk for prostate cancer.

19. The apparatus of claim 1, wherein the subject is a human aged 40 to 85 years.

20. The apparatus of claim 1, wherein the sensor subunits comprise polyethylenimine-ether functionalized gold nanoparticles.

21. The apparatus of claim 1, wherein the sensor subunits comprise tin (IV) oxide, tungsten oxide, or a combination of tin (IV) oxide and tungsten oxide.

22. The apparatus of claim 1, wherein the sensor subunits comprise polyetherimide / carbon black composite.

23. The apparatus of claim 1, wherein the sensor subunits solid phase microextraction fibers coated with polyvinylidene fluoride – carbon black.

24. A system for determining whether a subject has prostate cancer from a sample of volatile-organic-compounds emitted from a sample of the subject’s urine, the system comprising: a sample collection means for receiving the sample of volatile-organic- compounds emitted from the sample of the subject’s urine;Att’y Docket No.: IUIC-168 inside the housing, a sensor array comprising at least four sensor subunits, each sensor subunit comprising a chemiresistive substrate, and each sensor subunit independently corresponding to a signal output channel; each of the at least four sensor subunits independently tuned to: selectively output a resolvable electronic signal in response to a change in concentration emitted from the urine sample from baseline of one any of: β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, or 2,3,4,5-tetramethyl-2-cyclopenten-1-one; and wherein each of signal output channels is communicably coupled to a general purpose computer processor unit programmed to receive the resolvable electronic signals and determine whether the resolvable electronic signals correlate with the presence of prostate cancer in the subject.

25. The system of claim 24, wherein the general purpose computer processor unit is further programmed to receive the resolvable electronic signals and, if it is determined that the electronic signals correlate with the presence of prostate cancer in the subject, determine whether the resolvable electronic signals correlate with indolent prostate cancer or aggressive prostate cancer.

26. The system of claim 24, wherein the four sensor subunits are, independently, selective for detecting four of the chemical species concentration changes selected from: change in concentration from baseline of β-thujone, change in concentration from baseline of dimethyl trisulfide, change in concentration from baseline of α-curcumene, change in concentration from baseline of diethyl phthalate, change in concentration from baseline of 2,5-dimethylbenzaldehyde, and change in concentration from baseline of 2,3,4,5-tetramethyl-2-cyclopenten-1- one.

27. The system of claim 24, wherein the four sensor subunits are, independently, selective for detecting four of the chemical species concentration changes selected from: change in concentration from baseline of thymol, change in concentration from baseline of 6-amyl-α-pyrone,Att’y Docket No.: IUIC-168 change in concentration from baseline of 2,3,4,5-tetramethyl-2-cyclopenten-1- one, change in concentration from baseline of p-cresol, change in concentration from baseline of 2-methylpropanoic propanoic anhydride, change in concentration from baseline of methyl salicylate, and change in concentration from baseline of allyl-2-furoate.

28. The system of claim 26, wherein the change in concentration is an increase in concentration.

29. The system of claim 24, wherein the chemiresistive substrate comprises polyethylenimine-ether functionalized gold nanoparticles.

30. The system of claim 24, wherein the chemiresistive substrate comprises tin (IV) oxide, tungsten oxide, or a combination of tin (IV) oxide and tungsten oxide.

31. The system of claim 24, wherein the chemiresistive substrate comprises polyetherimide / carbon black composite.

32. The system of claim 24, wherein the chemiresistive substrate comprises solid phase microextraction fibers coated with polyvinylidene fluoride – carbon black.

33. A sensor array comprising: at least a first sensor subunit, a second sensor subunit, and a third sensor subuniteach sensor subunit comprising a heat-treated chemiresistive polyetherimide / carbon black composite substrate; the first sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 291.1 eV; the second sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 531.4 eV; andAtt’y Docket No.: IUIC-168 the third sensor subunit configured to selectively output an electrical signal when contacted with a chemical compound having a binding energy of 284.7 eV.

34. A method of detecting a prostate cancer status of a subject in need thereof, the method comprising: contacting volatile-organic-compounds emitted from a sample of the subject’s biofluid with a sensor array; the sensor array comprising two or more sensor subunits, each sensor subunit operable to detect at least one volatile-organic-compound biomarker emitted from the biofluid sample such that each sensor subunit is selectively chemiresistive to a chemical functional group corresponding to a volatile- organic-compound biomarker correlated with a presence of prostate cancer in the subject; wherein contacting the sensor subunit with a volatile-organic-compound biomarker having a functional group for which the sensor is selectively chemiresistive causes the sensor subunit to generate an output signal; relaying output signals generated by the sensor subunits of the sensor array to a computation module; the computation module correlating the electronic signals to the presence or absence of prostate cancer in the subject.

35. The method of claim 34, wherein the biofluid is urine.

36. The method of claim 34, wherein the volatile-organic-compound biomarkers are selected from any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5- dimethylbenzaldehyde, and 2,3,4,5-tetramethyl-2-cyclopenten-1-one.

37. The method of claim 34, wherein the volatile-organic-compound biomarkers are selected from any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1- one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2- furoate.

38. The method of claim 34, wherein the subject is human.Att’y Docket No.: IUIC-168 39. The method of claim 34, wherein the subject is biologically male.

40. The method of claim 34, wherein the subject is age 40–85.

41. A method of detecting a prostate cancer status of a subject in need thereof, the method comprising: introducing volatile-organic-compound biomarkers emitted from a biofluid sample into a GC-MS device; relaying output signals generated by the GC-MS device to a computation module; the computation module correlating the output signals to the presence or absence of prostate cancer in the subject.

42. The method of claim 41, wherein the biofluid is urine.

43. The method of claim 41, wherein the volatile-organic-compound biomarkers are selected from any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5- dimethylbenzaldehyde, and 2,3,4,5-tetramethyl-2-cyclopenten-1-one.

44. The method of claim 41, wherein the volatile-organic-compound biomarkers are selected from any of thymol, 6-amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1- one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, and allyl-2- furoate.

45. The method of claim 41, wherein the subject is human.

46. The method of claim 41, wherein the subject is biologically male.

47. The method of claim 41, wherein the subject is age 40–85.

48. The method of claim 41, wherein the GC-MS device is a tabletop device.

49. The method of claim 41, wherein the GC-MS device is a portable device.

50. A method of monitoring a prostate cancer status in a subject in need thereof, the method comprising: performing the method of claim 34 at a first timepoint to record aAtt’y Docket No.: IUIC-168 first reading; and performing the method of claim 34 at a second timepoint to record a second reading.

51. A method of monitoring a prostate cancer status in a subject in need thereof, the method comprising: performing the method of claim 41 at a first timepoint to record a first reading; and performing the method of claim 41 at a second timepoint to record a second reading.

52. A device for detecting cancer in a subject cancer in a subject from volatile-organic- compound biomarkers present in the subject’s biofluid, the apparatus comprising: a gas chromatography system, mass spectrometry system, or gas chromatography – mass spectrometry system configured to receive volatile organic compounds emitted from a sample of the subject’s biofluid; and a computation module programmed to interpret a signal output as corresponding with a presence or absence of cancer in the subject.

53. The device of claim 52, comprising a gas chromatography – mass spectrometry system.

54. The device of claim 52, wherein the computation module is further programmed to interpret the signal output as corresponding with whether any cancer present in the subject is aggressive cancer or indolent cancer.

55. The device of claim 52, wherein the biofluid is urine.

56. The device of claim 52, wherein the subject is biologically male.

57. The device of claim 52, wherein the system is a tabletop system.

58. The device of claim 52, wherein the system is a portable system.

59. The device of claim 52, wherein the cancer is prostate cancer.Att’y Docket No.: IUIC-168 60. The device of claim 52, wherein the computation module is programmed to interpret a signal corresponding with a change in a baseline concentration of any of β-thujone, dimethyl trisulfide, α-curcumene, diethyl phthalate, 2,5-dimethylbenzaldehyde, or 2,3,4,5-tetramethyl-2-cyclopenten-1-one, or any combination thereof, as corresponding with the presence of cancer.

61. The device of claim 52, wherein the computation module is programmed to interpret a signal corresponding with a change in a baseline concentration of any of thymol, 6- amyl-α-pyrone, 2,3,4,5-tetramethyl-2-cyclopenten-1-one, p-cresol, 2-methylpropanoic propanoic anhydride, methyl salicylate, allyl-2-furoate, or any combination thereof, as corresponding with aggressive cancer.

62. An electronic communications system comprising: the apparatus of claim 1 communicatively coupled with a computing device having a display and installed with software programmed to receive data from the apparatus, system, sensor, or device and displaying data on the display.

63. The electronic communications system of claim 62, wherein the software is further programmed to compute and display diagnostic information which assists an end user in monitoring, preventing, and / or treating cancer.

Images