Patent Application
20200217847
Previous studies have reported volatile organic compounds (VOCs) in breath as apparent biomarkers of lung cancer. Since seeking breath biomarkers of lung cancer, researchers have employed a wide range of different tools including VOC separation methods using gas chromatography mass spectrometry (GC MS), non-separative detectors, such as electronic noses and chemosensors, analysis of expired breath condensate, measurement of breath temperature, and sniffer dogs. Analysis of breath VOCs with analytical instruments employing 2-dimensional GC has revealed a complex matrix of 2,000 different VOCs in a single sample. Data management tools for metabolomic analysis that were originally developed for genomics and proteomics have been used to manage the information. An increased risk of false discovery of biomarkers can arise when a multivariate model over-fits large number of candidate breath VOCs to a small number of test subjects, these VOCs could have been non-specific biomarkers of malignant as well as non-malignant lung diseases, these VOCs could have been non-specific biomarkers of malignant as well as non-malignant lung diseases.
Despite these concerns, breath biomarkers of lung cancer have been proposed as safe and cost-effective tools to help determine a person's risk of lung cancer. There is a clinical need for such a test because more people in the United States die from lung cancer than from any other type of cancer and early detection can save lives. The National Lung Screening Trial found that screening with low-dose chest CT reduced mortality from lung cancer by 20%. However, the comparatively low positive predictive value (PPV) of chest CT (2.4% to 5.2%) has raised concerns that screening for lung cancer might yield an overwhelming number of false-positive test results.
It has also been found that chest imaging can cause harm. The National Lung Screening Trial 1,2 also revealed two major deficiencies of low-dose computed tomography of chest (LDCT). The LDCT showed low yield only 7.3% of subjects were positive for lung cancer on biopsy, and poor specificity since the false-positive rate was 20.1%. The resulting harms were: over-investigation of false-positive results with potentially harmful tests, such as bronchoscopy, biopsy. Needless exposure to potentially harmful radiation: 92.7% of the irradiated population was cancer-free. LDCT has high costs plus higher costs of needless tests to over-investigate false-positive results. Pulmonary nodules seen on LDCT frequently elicit false-positive reports of malignancy. False-positive reports can lead to needless procedures such as bronchoscopy and lung biopsy that are invasive, costly, and potentially hazardous.
Radiologists have attempted to minimize false-positive LDCT results by stratifying cancer risk according to the radiographic appearance of a nodule. Pulmonary nodule features seen on LDCT that are most suggestive of malignancy include lesion size >11 mm and ground-glass appearance, while polygonal lesions are usually benign.
The problem with predictions of malignancy based on the appearance of a pulmonary nodule is that no single feature is both highly sensitive and highly specific for disease. Researchers have attempted to improve the sensitivity and specificity of LDCT by employing various combinations of the nodule features shown using naked eye assessment as well as computer-assisted algorithms and machine learning with artificial neural networks. LDCT has also been combined with ancillary imaging modalities such as magnetic resonance imaging (MRI) and positron emission tomography (PET) scanning. The combination of LDCT with MRI and positron emission tomography (PET) scanning has the shortcoming of entailing additional costs and radiation exposure.
It is desirable to provide new and improved methods of predicting nodules on a chest CT with improved accuracy and reduce the number of false-positive and false-negative test findings.
The present invention provides a method and system for identifying a non-invasive biomarker in breath for detecting lung cancer. The disclosed system and method far exceeds the sensitivity and reliability of conventional LDCT and can be used for reducing the number of false-positive reports of malignant pulmonary nodules. In the present invention, the method determines a single breath biomarker that can be used to predict nodules on a chest CT that are read as consistent with lung cancer. The biomarker is referred to as mass abnormalities in gaseous ions with imaging correlates (MAGIIC).
The method for identifying a biomarker to predict nodules on a chest CT as indicating lung cancer comprises the steps of:
collecting a breath sample from subjects known to have nodules on a chest CT and subjects known to be free of nodules on a chest CT;
analyzing the collected breath samples to determine all mass ions in each of the collected breath samples using at least one time-resolved separation technique and at least one mass-resolved separation technique;
identifying a subset of the determined mass ions in a processor as the biomarkers for detecting the disease; and
combining the subset of the determined mass ions in a multivariate algorithm in the processor to generate a value of a discriminant function indicating the likelihood that nodules on a chest LDCT of the subject are consistent with lung cancer.
In one embodiment, the biomarker is determined in breath from a single volatile organic compound (VOC) after bombardment of the breath VOC with high energy electrons using a gas chromatography mass spectrometry (GC MS). Alternatively, the VOC can be analyzed with a surface acoustic wave (SAW) gas chromatography sensor (GC SAW) or flame ion detection (GC FID).
The biomarkers in breath can be oxidative stress biomarkers. In one embodiment, the biomarkers in breath are C4 and C5 alkanes or alkane derivatives.
In one embodiment, the method of the present invention is used for predicting the probable presence of lung cancer in a test subject using the method for identifying biomarkers of the present invention.
Another embodiment of the invention features a system for identifying a plurality of biomarkers for predicting lung cancer in a subject including an apparatus for collecting a breath sample from subjects known to have nodules on a chest CT and subjects known to be free of nodules on a chest CT. A mass spectrometer (MS) associated with a gas chromatograph (GC) apparatus analyzes the collected breath samples to determine all mass ions in each of the collected breath samples. A computer identifies a subset of the determined mass ions as the biomarkers for detecting lung cancer as the disease, the subset of the determined mass ions indicate the likelihood that nodules on a chest LDCT of the subject are consistent with lung cancer, and combines the subset of the determined mass ions in a multivariate algorithm to generate a discriminant function. The discriminant function indicates a value of the likelihood that the subject has lung cancer.
It was found that biomarkers determined with the method of the present invention accurately predicted lung cancer in a blinded replicated study
The invention will be more fully described by reference to the following drawings.
The invention will be more fully described by reference to the following drawings.
Reference will now be made in greater detail to a preferred embodiment of the invention, an example of which is illustrated in the accompanying drawings. Wherever possible, the same reference numerals will be used throughout the drawings and the description to refer to the same or like parts.
In block 16, a subset of the determined mass ions are determined that correlated with the presence of structural abnormalities in LDCT that were read as consistent with nodules. In block 18, the subset of the determined mass ions is combined in a multivariate predictive algorithm to generate a value of a discriminant function (DF) indicating the likelihood that the subject has nodules on a LDCT.
Blocks 34, 35 and 36 describe steps using multiple Monte Carlo simulations to identify a set of mass ion biomarkers predicting the likelihood that the subject has nodules on a LDCT with greater than random accuracy. In block 34, a correct assignment curve is constructed with data of the AUC of the ROC curves for all candidate biomarker mass ions. In one embodiment, block 34 can be performed by assigning all data of the AUC of the ROC curves to a series of bins with incremental values. For example, the bins can be assigned values of 0.50 to 0.51, 0.51 to 0.52 and so forth up to 0.99 to 1.0. The correct assignment curve is generated as a plot of the number of mass ions in a bin on the y-axis versus the AUC value of a bin on the x-axis
In block 36, the subset of candidate biomarker mass ions with greater than random ability to identify nodules on a LDCT for identifying lung cancer are identified using a correct assignment curve and a random assignment curve
In block 37, the multi-variate predictive algorithm is constructed using the candidate biomarker mass ions from the correct assignment curve that were identified as having greater than random ability to identify predicting nodules on a LDCT. A list is generated of all candidate biomarker mass ions in the correct assignment curve that were identified as having greater than random ability to identify the disease. Each of the listed candidate mass ions are ranked by the AUC of the ROC curve. The ranking can be from highest to lowest. A predetermined number of candidate biomarker mass ions having the highest ranking are used to generate the multivariate predictive algorithm.
Method 10 for identifying biomarkers and generating an output indicative of lung cancer of the present invention can be used to detect the probable presence of lung cancer in a human subject. A breath sample from a test subject is collected, chemically analyzed and the data is analyzed with the multivariate algorithm to generate a value of the discriminant function for the test subject. The value of the discriminant function for the test subject is compared to the value of the discriminant function determined in block 18.
VOCs are thermally desorbed from the sorbent trap 62, separated by gas chromatography apparatus 70, and injected into mass spectrometry detector 72. In mass spectrometry detector 72 the VOCs are bombarded with energetic electrons in a vacuum and degraded into a set of ionic fragments, each with its own mass/charge (m/z) ratio. Data from gas chromatography apparatus 70 and mass spectrometry detector 72 is received at processor 74.
The unique diagnostic value of the MAGIIC biomarker in this dataset was determined when testing the hypothesis that a single biomarker should predict two conditions simultaneously of pulmonary nodules on a LDCT and biopsy-proven lung cancer. This reduced the universe of 70,000 candidate mass ions biomarkers to a small number of mass ions in which the MAGIIC biomarker delivered the best combination of accuracy, sensitivity and specificity.
Breath tests were performed in a group of 301 subjects undergoing screening for lung cancer. All subjects donated a sample of alveolar breath. Collection of breath VOC samples was performed in accordance with method 10 for identifying biomarkers and generating an output indicative of lung cancer and system 60. A subject wears a nose clip and breathes normally through a disposable valved mouthpiece and bacterial filter into the BCA for 2.0 min. Alveolar breath VOCs are captured on to a sorbent trap that is immediately sealed in a hermetic container. Since there is low resistance to expiration (˜6 cm water), breath samples could be collected without discomfort from elderly patients and those with respiratory disease. In order to minimize the risk of potential site-dependent confounding factors such as environmental contamination of room air, subjects in all four groups donated breath samples in the same room at each clinical site. All subjects donated two samples for replicate assay at two independent laboratories (Menssana Research, Inc referred to as laboratory A and American Westech, Inc., Harrisburg, Pa. referred to as laboratory B). Samples were stored at −15° C. prior to analysis.
Analysis of breath VOC samples: Analysis of breath VOC sample was performed with method 10 for identifying biomarkers and generating an output indicative of lung cancer and system 60. Statistical analysis identified a breath mass ion biomarker mass abnormalities in gaseous ions with imaging correlates (MAGIIC) that correlated with the presence of structural abnormalities in LDCT that were read as consistent with lung cancer. Using automated instrumentation, VOCs were thermally desorbed from the sorbent trap 62, cryogenically concentrated, and assayed by gas chromatography mass spectrometry (GC MS). A known quantity of an internal standard (bromofluorobenzene) was automatically loaded on to all samples in order to normalize the abundance of VOCs and to facilitate alignment of chromatograms. GC MS data from both laboratories was pooled for analysis and development of a single predictive algorithm.
Alignment of single ion masses in chromatograms: Chromatograms were processed with metabolomic analysis software (XCMS in R) in order to generate a table listing retention times with their associated ion masses and intensities. Retention times and ion mass intensities were normalized to the bromofluorobenzene (ion mass 95) internal standard in each chromatogram. The aligned data was then binned into a series of 5 sec retention time segments. Identification of biomarker single ions: The statistical methods have been previously described. Mass ions as candidate biomarkers of lung cancer were ranked by comparing their intensity values in subjects with lung cancer (Group 3 lung cancer confirmed by tissue diagnosis shown in table 3) to cancer-free controls (Group 1 with negative chest CT). In each 5 sec time segment, the diagnostic accuracy of each mass ion was ranked according to its C-statistic value [(area under curve (AUC) of the receiver operating characteristic (ROC) curve]. Multiple Monte Carlo simulations were employed in order to minimize the risk of including random identifiers of disease by selecting the mass ions in each time segment that identified active lung cancer with greater than random accuracy. The average random behavior of mass ions in each time segment was determined by randomly assigning subjects to the “lung cancer” or the “cancer-free” group and performing 40 estimates of the C-statistic. For any given value of the C-statistic, it was then possible to identify the ionic biomarkers that exhibited greater diagnostic accuracy with correct assignment than with multiple random assignments. Development of predictive algorithm: Biomarker ions that identified lung cancer with greater than random accuracy were employed to construct a predictive algorithm using multivariate weighted digital analysis (WDA).
A receiver operating characteristic (ROC) curve of MAGIIC is shown in
1. List every subject in a row that contains their MAGIIC score and the presence of nodules (yes/no)
2. Use the Excel function to rank all rows according to MAGIIC score, ranging from lowest MAGIIC score in the top row down to highest score in bottom row.
3. Insert two new columns in the spreadsheet labeled sensitivity and specificity
4. For each subject, calculate sensitivity and specificity row by row:
sensitivity=TP/(TP+FN)
specificity=TN/(TN+FP)
where TP=true positives
The VOC compound names were identified based on mass spectrum of the MAGIIC breath VOC biomarker with known mass spectra of other compounds. In one embodiment, the MAGIIC biomarkers in breath are C4 and C5 alkanes or alkane derivatives. In alternate embodiments, the MAGIIC biomarkers in breath are selected from 1,4-butanediol, 2-pentanamine,4-methyl-, 2-propanamine, 3-butenamide, acetamide, 2-cyano-, alanine, N-methylglycine or octodrine.
The abundance of the MAGIIC biomarker was determined in a different group of 158 subjects undergoing LDCT. The study was blinded and monitored with Good Clinical Practice (GCP). MAGIIC was assayed in duplicate breath VOC samples analyzed at two independent laboratories, and predicted the outcome of LDCT with 80% at laboratory A as shown in
The sensitivity and specificity of MAGIIC for nodules observed on LDCT varied with its abundance in breath is shown in
It is to be understood that the above-described embodiments are illustrative of only a few of the many possible specific embodiments, which can represent applications of the principles of the invention. Numerous and varied other arrangements can be readily devised in accordance with these principles by those skilled in the art without departing from the spirit and scope of the invention.
| Number | Date | Country | |
|---|---|---|---|
| 62174256 | Jun 2015 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 15177695 | Jun 2016 | US |
| Child | 16735118 | US |
