Patent Application
20170283883
The present disclosure relates generally to the fields of molecular biology and cancer diagnosis and therapy.
Lung cancer is the leading cause of cancer death worldwide and its incidence continues to grow in women and in developing countries. As lung cancer is asymptomatic in its early stages, the majority of patients are diagnosed with advanced disease, for example, when the tumor is unresectable. Consequently, the overall survival rate is very low: 16% at 5 years. It is important, therefore, that screening programs and novel diagnostic tools are developed to improve the ability to detect the disease in its early stages (stage I-II) when it is still curable.
Low-dose computed tomography (LDCT) is an effective tool for the diagnosis of lung cancer, as proved by several single-arm and randomized studies. However, concerns have been raised about the feasibility and cost of a nationally implemented LDCT screening program. Although LDCT is targeted to high-risk individuals, identified by age and smoking history, a refinement of the pre-selection criteria based on additional risk factors, such as tumor markers, can prove to be of great benefit in enabling the widespread and cost-effective implementation of LDCT screening.
There is a long-felt yet unmet need for a minimally invasive and relatively cheap blood test for use as a first-line screening procedure to pre-select patients who require further diagnostic investigation by LDCT. Such a test would reduce the size of the target screening population, and would undoubtedly be advantageous in terms of costs, screening uptake rates and reduced medicalization of participants.
The disclosure provides a minimally invasive and relatively cheap blood test for use as a first-line screening procedure to pre-select patients who require further diagnostic investigation by LDCT.
The disclosure describes a method for diagnosing lung cancer in a subject by detecting in a biological sample obtained from that patient a miRNA signature, the presence of which provides an earlier indication of cancer than alternative art-recognized methods, including, but not limited to, low-dose computed tomography (LDCT).
MicroRNAs (miRNAs), short non-coding RNAs involved in the regulation of cellular differentiation, proliferation and apoptosis, are emerging as one of the most promising classes of blood-borne tumor markers. The expression of miRNAs is often deregulated in human tumors, leading to alterations in miRNA profiles in bodily fluids, including serum and plasma. Moreover, cell-free miRNAs display remarkable stability in the blood, which is attributed to the protection from harsh conditions and blood RNases provided either by microvesicles, such as exosomes, or by protein complexes, such as high-density lipoprotein complexes or argonaute proteins.
The disclosure provides evidence that a miRNA signature of detected circulating miRNAs provides an early indicator of cancer, including, lung cancer.
The disclosure provides data validating the efficacy of the methods described herein in a large-scale study aiming at detecting lung cancer, including asymptomatic and/or early-stage lung cancers in a cohort of high-risk individuals enrolled in the COSMOS (Continuous Observation of SMOking Subjects) lung cancer screening program. The data demonstrate that the methods of the disclosure provide a cost-effective, easily implementable, diagnostic screening tool for lung cancer.
The disclosure provides a method of diagnosing lung cancer in a subject in need thereof comprising (a) detecting in a biological sample from the subject a decrease in the abundance of each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p compared to a control abundance value corresponding to each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p from a control sample; (b) detecting in the biological sample an increase in the abundance of each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p compared to a control abundance value corresponding to each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p from a control sample; and (c) diagnosing the subject with lung cancer, when a decreased abundance in each of the miRNA in (a) is detected and an increased abundance of each of the miRNA in (b) is detected.
Methods of the disclosure may be used to diagnose asymptomatic lung cancer and/or early stage lung cancer. Methods of the disclosure may be used to diagnose any subtype of lung cancer.
Biological samples and/or control biological samples of the methods of the disclosure may comprise, consist essentially of or consist of a biological fluid. Exemplary biological fluids include, but are not limited to, saliva, urine, blood, or lymph fluid. Biological samples and/or control biological samples of the methods of the disclosure may comprise, consist essentially of or consist of blood, whole blood, blood plasma and/or blood serum. Biological samples of the methods of the disclosure may comprise, consist essentially of or consist of blood serum.
Control biological samples of the methods of the disclosure may comprise, consist essentially of or consist of blood, whole blood, blood plasma and/or blood serum obtained from at least two normal subjects. Control biological samples of the methods of the disclosure may comprise, consist essentially of or consist of blood serum obtained from at least two normal subjects.
According to the methods of the disclosure, a decrease in the abundance of each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p compared to a control abundance value may be a statistically significant difference from the control value. Alternatively, or in addition, an increase in the abundance of each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p compared to a control abundance value may be a statistically significant difference from the control value.
Statistically significant differences of the disclosure may be characterized by a p-value of less than 0.05. Statistically significant differences of the disclosure may be characterized by a p-value of less than 0.001. Statistically significant differences of the disclosure may be characterized by a p-value of less than 0.0001.
According to the methods of the disclosure, the decrease in the abundance of each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p compared to a control abundance value may be expressed as a fold-difference. Alternatively, or in addition, an increase in the abundance of each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p compared to a control abundance value may be expressed as a fold-difference.
The disclosure provides a method of diagnosing lung cancer in a subject in need thereof comprising (a) detecting in a biological sample from the subject a decrease in the abundance of each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p compared to a control abundance value corresponding to each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p from a control sample; (b) detecting in the biological sample an increase in the abundance of each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p compared to a control abundance value corresponding to each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p from a control sample; (c) calculating a risk score and (d) diagnosing the subject with lung cancer, when a decreased abundance in each of the miRNA in (a) is detected and an increased abundance of each of the miRNA in (b) is detected. In certain embodiments of this method, the risk score is a product of (a) a fold decrease and a weight coefficient, or (b) a fold increase and a weight coefficient, wherein the weight coefficient is determined by, for example, a diagonal linear discriminant analysis (DLDA).
Control values for each miRNA of the methods of the disclosure may be determined by a method comprising detecting in a control biological sample from a normal subject an abundance of each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, hsa-mir-331-3p, hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p. In certain embodiments of these methods, the at least one normal subject of the disclosure does not have cancer. Alternatively, or in addition, the at least one normal subject does not have lung cancer.
Subjects of the methods of the disclosure may be male or female. Subjects of the methods of the disclosure may be any age; however, subjects are preferably adults.
Subjects of the methods of the disclosure may be asymptomatic.
Subjects of the methods of the disclosure may have stage I or stage II lung cancer.
Subjects of the methods of the disclosure may present one or more risk factors for developing lung cancer. Exemplary risk factors for developing lung cancer include, but are not limited to, a personal or family history of cancer, a history of smoking and/or exposure to second-hand smoke, and/or having limited access to preventative or curative medical care. Alternatively, or in addition, a subject who is exposed to fine particulates from his or her environment is at an increased risk of developing lung cancer. For example, an individual who is exposed to smoke (including secondhand smoke), radon gas, asbestos and other chemicals (e.g. arsenic, chromium, and/or nickel), and/or nano-size particulates (e.g. dust and particulates from a manufacturing facility and/or motor vehicle exhaust) are at an increased risk of developing lung cancer. The combination of a family history of cancer and any one or more of the risk factors described herein may further increase an individual's risk of developing lung cancer.
Methods of the disclosure may further comprise the step of performing low-dose computed tomography (LDCT) or referring the subject for LDCT if the subject is diagnosed with lung cancer.
Methods of the disclosure may further comprise the step of providing a treatment or referring the subject for treatment if the subject is diagnosed with lung cancer.
According to the certain embodiments of the methods of the disclosure, detecting step of (a) and/or (b) may comprise transcribing each of the miRNA in steps (a) and/or (b) using at least one human miRNA stem-loop primer specific for at least one miRNA of (a) or (b) to generate a complementary DNA (cDNA) corresponding to each of the miRNA, amplifying each of the cDNA, and determining a relative abundance of each of the cDNA compared to at least one housekeeping miRNA. In certain embodiments of these methods, the at least one human miRNA stem-loop primer specific for at least one miRNA and the at least one miRNA hybridize to form at least one duplex. In certain embodiments of these methods, the amplifying step comprises performing quantitative real-time polymerase chain reaction (qRT-PCR). In certain embodiments of these methods, the at least two housekeeping miRNA comprise miR-197, miR-19b, miR-24, miR-146, miR-15b, or miR-19a. Alternatively, the at least two housekeeping miRNA may comprise miR-197, miR-19b, miR-24, miR-146, miR-15b and miR-19a.
In certain embodiments of these methods, the relative abundance of each of the cDNA is determined by adding a scaling factor to the raw cycle threshold (CT) of each of the cDNA to generate a normalized (CT) value. Scaling factors of the disclosure may be determined by determining a mean cycle threshold of the at least two housekeeping miRNA selected from the group consisting of miR-197, miR-19b, miR-24, miR-146, miR-15b, and miR-19a and subtracting the mean CT from a constant value (K). In certain embodiments of the methods of the disclosure, the constant value (K) equals 21.646. In certain embodiments of these methods, the normalized CT value is used to determine a weight coefficient.
Methods of the disclosure may further comprise the step of determining a risk score for the subject. In certain embodiments of these methods, risk score is calculated according to: RS=−(Σiwi xi−THRES), wherein i corresponds to each of the miRNAs of (a) and (b), respectively, and wherein THRES=−261.779, wi is the weight coefficient of the ith miRNA, and xi is the expression value of the ith miRNA. The expression value of the ith miRNA may also be referred to as the raw CT of the ith miRNA. In certain embodiments of these methods, the risk score is greater than or equal to 5, indicating that the subject has a high risk of developing cancer. In certain embodiments of these methods, the risk score is less than 5 and greater than or equal to −5, indicating that the subject has an intermediate risk of developing cancer. In certain embodiments of these methods, the risk score is less than −5, indicating that the subject has a low high risk of developing cancer.
The drawings are primarily for illustrative purposes and are not intended to limit the scope of the inventive subject matter described herein. The drawings are not necessarily to scale; in some instances, various aspects of the inventive subject matter disclosed herein may be shown exaggerated or enlarged in the drawings to facilitate an understanding of different features. In the drawings, like reference characters generally refer to like features (e.g., functionally similar and/or structurally similar elements).
The disclosure provides a “miR-Test” for the early detection of lung cancer. An exemplary embodiment of the miR-test comprises a method of diagnosing a subject with lung cancer, including asymptomatic and/or early stage lung cancer, comprising (a) obtaining a biological sample from the subject; (b) detecting in the biological sample a decrease in the abundance of each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p compared to a control abundance value corresponding to each of hsa-mir-92a-3p, hsa-mir-30b-5p, hsa-mir-191-5p, hsa-mir-484, hsa-mir-328-3p, hsa-mir-30c-5p, hsa-mir-374a-5p, hsa-mir-7d-5p, and hsa-mir-331-3p; and (c) detecting in the biological sample an increase in the abundance of each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p compared to a control abundance value corresponding to each of hsa-mir-29a-3p, hsa-mir-148a-3p, hsa-mir-223-3p, and hsa-mir-140-5p; wherein the detection of a decreased abundance in each of the miRNA in (b) and an increased abundance of each of the miRNA in (c) indicates the development of lung cancer in the subject, thereby diagnosing the subject with lung cancer. In support of the miR-Test of the disclosure, data are provided for the validation of this method of diagnosing lung cancer, including asymptomatic and/or early stage lung cancer, in a large cohort of high-risk individuals. In this study, the test reached a sensitivity of 86%, when the high and intermediate risk classes were grouped together, while 53% of individuals were positioned in the low-miRNA risk category. The NPV of the miR-Test was >99%, thus, low-risk individuals can safely avoid subsequent LDCT screening. The high sensitivity and NPV obtained with the miR-Test are comparable to those observed with LDCT alone, which indicates that the miR-Test could substitute the LDCT as a first-line screening tool. Conversely, the lower specificity of the test compared to LDCT, would not affect the overall screening result, as cases with a positive miR-Test result would be required to undergo LDCT to confirm diagnosis and localize neoplastic lesions for subsequent surgery.
With respect to the false negative (i.e. miR-Test-negative lung cancer patients) and the false positive (miR-Test-positive individuals, but negative upon LDCT) results, only 2 out of 15 deaths (Validation and Clinical Sets) were false negatives. The death rate per 1000 patients/year was 0, 51, and 71, in the low, intermediate and high-risk categories, respectively. While the relatively low number of deaths in our cohort limits the statistical power of these results, it is plausible that tumors missed by the miR-Test may be rather indolent or even represent overdiagnosis by LDCT. Notably, 5 of 5 patients with benign tumors, who had a positive LDCT result, were miR-Test-negative. Similarly, most individuals with NMD (72 out of 83; 87%) were miR-Test-negative. This latter result is relevant since in LDCT screening trials there is a high rate of false-positive findings, up to 28%. This complicates the interpretation of LDCT results and ensuing decisions about the screening time interval: a difficulty that might be alleviated by a first-line screening test, such as the miR-Test, that significantly reduces unnecessary LDCTs for individuals without lung cancer. With respect to miR-Test false positives, these events may not necessarily represent an intrinsic limitation of the test, because it is possible that the blood test can anticipate the LDCT diagnosis.
The miR-Test possesses the characteristics of accuracy and robustness required to be introduced as a first-line tool in lung cancer screening programs. If implemented, the miR-Test would result in a reduction in the number of LDCTs by more than 50%, while retaining the diagnostic sensitivity of LDCT.
Four independent cohorts of patients and individuals were employed in the study: i) Calibration set, ii) Validation set, iii) Specificity set, iv) Clinical set (see
Calibration Set:
Twenty-four individuals were selected from the COSMOS trial (12 with screen-detected lung cancer and 12 lung cancer free;
Validation Set:
The miR-Test was validated in an independent set of 1008 individuals enrolled in the COSMOS study including 36 patients with LDCT detected lung cancer and 972 individuals without lung cancer, randomly selected from a consecutive cohort from March 2011 to March 2012.
Specificity Set:
A third cohort of 83 patients was used for further validation of the miR-Test. These individuals were selected from COSMOS study participants and were not included in any of the other sets used in this study. This cohort was composed of: i) 38 individuals with CT-detected solitary pulmonary nodules stable in size at 5 years of follow-up; ii) 16 patients with chronic obstructive pulmonary disease; iii) 24 individuals with pneumonia; iv) and 5 with operated benign lung tumor. Importantly, none of these individuals developed lung cancer during a >5-year follow-up period by LDCT.
Clinical Set:
A fourth independent cohort of 74 patients diagnosed with lung cancer outside of the COSMOS trial was used. These patients underwent surgery at the European Institute of Oncology from November 2005 to January 2008.
Information about the clinical and pathological characteristics for all individuals and patients screened by the miR-Test is reported in Table 1 and Table 2. The mean±standard deviation (SD) is reported for age (years) and smoking status (pack-years), and the median, interquartile (Q1;Q3) and overall range (min-max) is reported for age. The information on pack-years was available for nearly all individuals except for: 25 patients (a) and 36 patients (h) (AC, lung adenocarcinoma; SCC, lung squamous cell carcinoma; SCLC, small cell lung cancer; LCC, large cell lung cancer; COPD, chronic obstructive pulmonary disease; NOD, stable solitary pulmonary nodules; PN, pneumonia; Benign, patients operated with benign pulmonary nodules (surgery false positives)). Tumor stage was defined based on the TNM Classification of Malignant Tumors published by the International Union Against Cancer (UICC), 7th Edition. Percentages could not add up to 100 due to rounding.
The 12 asymptomatic lung cancer patients in the ‘calibration-set’ were part of the original cohort of 174 individuals used to derive the miR-Test.
Tumor stage at the time of diagnosis was determined according to guidelines of the American Joint Committee on Cancer (http://www.cancerstaging.org/). Informed consent was obtained from all participants. Patients and individuals of the Calibration, Validation and Specificity sets were all enrolled in the COSMOS study, a screening program for the anticipation of lung cancer diagnosis in high-risk individuals. All enrolled individuals were smokers or former smokers with a smoking exposure of more than 20 pack-years, aged over 50.
For the experiments shown in
Blood samples (10 mL) were collected by standard phlebotomy before any analysis or instrumental procedure. The first 3 mL of blood were not used for serum preparation to prevent contamination by skin. Serum was prepared by collecting blood in tubes with clot activator (S-Monovette 7.5 ml, REF01.1601—Sarstedt), left at room temperature for 3 hours to clot, then spun at 3000 rpm (1000 g, Megafuge 2.0 Heraeus) for 10 minutes at RT. The serum was removed immediately after centrifugation, leaving a 0.5 cm leftover to avoid disturbing the serum-clot interface. Serum was then aliquoted in barcoded cryotubes and snap frozen in dry ice. Aliquots were stored in a dedicated −80° C. freezer.
Serum miRNA Purification and Expression Profiling
Total RNA purification, including miRNAs, was based on lysis with Guanidinium thiocyanate-phenol-chloroform extraction (TriZol-LS, Applied Biosystem) and Spin Column-based total RNA purification (MiRneasy Mini Kit, Qiagen). Briefly, 0.3 mL of serum were mixed with Trizol-LS in volumetric ratios of 3:1 for lysis. After denaturation, a spike of synthetic miRNA (5×108 copies of miR-34a) was added to the solution to monitor the extraction efficiency. A volume of 0.24 mL of Chloroform was added to the solution, which was then centrifuged for 15 min at 11,800 g at 4° C. A fixed amount (0.35 mL; around 70% of the final volume of the sample) of aqueous phase was recovered. This choice was preferred to limit contamination from the interphase.
The subsequent steps were performed in automation through a Qiacube machine (QIAGEN), according to the “miRNEASY Mini” standard protocol. RNA was eluted in 30 mL and stored at −80° C. until further analysis. Reverse transcription (RT) and pre-amplification reaction (Pre-AMP) protocols were optimized by using a MicroLab Star Liquid Handling. RT was performed from a fixed volume (3 μL) of total RNA with TaqMan MicroRNA Reverse Transcription Kit and a custom pool of Human miRNA-specific stem-loop primers (Applied Biosystem) according to the manufacturer's instruction, in a 2720 Thermo Cycler (Applied Biosystem). RT product was then diluted 1:2 with water; 5 μL of diluted RT product were pre-amplified (12 cycles of PCR) with TaqMan PreAmp Master Mix (2×) and Megaplex PreAmp Primers Human PoolA (Applied Biosystem), according to the manufacturer's instruction. The pre-amplification reaction was diluted 1:4 with Tris-EDTA buffer 0.1×. Next, 6 μL of diluted pre-amplification reaction were combined with 46.5 μL of water and with 52.5 μL of TaqMan Universal Master Mix II (Applied Biosystem). The final solution (105 μL) was loaded into one lane of a custom TaqMan®Low Density Array microRNA Custom Panel (Applied Biosystem). qRT-PCR was carried out on an Applied Biosystems 7900HT thermocycler using the manufacturer's recommended cycling conditions. A Hamilton MICROLAB® STAR liquid handling workstation was used to automate most of the described processes.
qRT-PCR Data Analysis—miRNA Profiling and Data Normalization
Data were exported into text format for further analysis. If qRT-PCR amplification curves presented a “Tholdfail” error flag (i.e. the automatic thresholding algorithm failed), data were discarded. Subsequently, raw Ct values were normalized.
To run the miR-Test, data were normalized using mean CTs (HK-mean) of six “housekeeping” miRNAs genes (miR-197, miR-19b, miR-24, miR-146, miR-15b, miR-19a) using the following procedure: a scaling factor (SF) was calculated for each sample by subtracting the “HK-mean” to a constant value (K=21.646). Data were then normalized by these SFs to eliminate technical fluctuations, using the formula:
SF+miRNA CTraw=CTnormalized (1)
If CT raw was greater than 30.01 or ‘undetermined’ by qRT-PCR, data was set to 30.01 and normalization skipped. This normalization strategy assures an independency from the original set of samples used to train the miR-Test, and allow new samples classification without the need of an internal reference.
The calibration set was used to: i) refine diagonal linear discriminant analysis (DLDA) weight coefficients (wi) of the miRNA signature by using DLDA as class metric (BRB-ArrayTools version 4.3.0—Beta_2 Release); ii) reduce the number of miRNAs in the signature. The leave-one-out (LOO) cross-validation procedure was used to cross-validate the classifier and estimate sensitivity and specificity in the calibration set. Parametric t-test set at 0.05 significance level was used for feature selection. Statistical significance of the performance of the classifier was assessed by performing 1000 permutations of class labels. Samples were then classified as positive or negative if the absolute value of the inner sum of weighted expressions (CTnorm*wi) was greater than the absolute value of the threshold. Since both weights and threshold are negative, the risk score was calculated as:
RS=(Σiwixi−THRES), for i=1 . . . 13 (13-miRNA signature) (2)
where THRES=−261.779, wi is the weight coefficient of the ith miRNA (Table 3), and xi is the expression value (CT) of the ith miRNA.
All analyses were automated using a custom R script developed in-house using the R statistical software version 2.14.1. The script allows to read the information content of the output documents from the Applied Biosystems ViiA 7™ and provides miR-Test risk scores automatically.
miRNA Signature Refinement
The Calibration set was used to refine a 34-miRNA signature (see, Bianchi F, Nicassio F, Marzi M, et al. EMBO molecular medicine 2011; 3(8):495-503) by employing an optimized serum miRNA detection protocol. Briefly, a pre-amplification step (PreAMP) that improved the detection of circulating miRNAs [˜23 versus ˜31 cycle threshold (Ct), on average] was added, and all steps for purification and sample preparation (including PreAMP) were automated. This procedure minimized technical variability that resulted in a lower number of miRNAs in the signature from 34 to 13, while preserving the performance of the original model (Tables 3 and 4;
Table 3 reports the miRNA assay, accession numbers (Acc.), sequence, and miRbase nomenclature (release 20) for the 13 miRNAs that comprise the miR-Test. Fold-change (Fold) and p-values (parametric t-test) refer to the expression of miRNAs in the 12 tumor sera versus the 12 normal sera of the Calibration set. Wi is the weight coefficient computed by diagonal linear discriminant analysis (DLDA) and used in the miR-Test to compute risk scores.
The sensitivity (SEN) and specificity (SPE) of the original 34-miRNA signature and the reduced 13-miRNA signature (the miR-Test) in the Calibration Set is depicted in Table 4. Sensitivity and specificity are based on cross-validation results of the diagonal linear discriminant analysis (DLDA) classifier (see Example 1). A risk score of 0 was used to compute sensitivity and specificity.
Analyses were performed using JMP 10 software (SAS) and SAS 9.3 (SAS Institute, Cary N.C.). Cumulative lung cancer-specific mortality was estimated through the competing risks approach.
A multi-tiered study was designed on high-risk individuals (heavy smokers, aged>50) enrolled in the lung cancer screening trial COSMOS, and on lung cancer patients diagnosed outside of screening (
The miR-Test was then validated in an independent “Validation Set” of 1008 subjects enrolled in the COSMOS trial (
Next, a simulation of a “clinical” setting was attempted, in which the miRNA-Test would be used as a triage screening to identify individuals who should subsequently undergo LDCT. The Validation Set was stratified into three risk categories: high, intermediate, and low (see Example 1). By grouping together the high and intermediate risk categories, the sensitivity of the test increased to 86% (31 of 36 tumors, Table 5). Analysis of individuals in the low risk category (533 of 1008 (53%) including 5 tumors; Table 5), who would not be required to undergo LDCT, revealed a negative predictive value (NPV) of >99% for the miR-Test. The fact that 5 of 36 LDCT-detected tumors were classified as low risk (false negatives) by the miR-Test could reflect an intrinsic limitation of the test in its present form. However, in the case of the false negatives, we did not observe any death event out of the 15 deaths registered in all individuals screened (Validation and Clinical Sets). In addition, the death rate per 1000 patients/year was 0, 51, and 71, in the low, intermediate and high-risk categories, respectively. While the relatively low number of deaths in our cohort limits the statistical power of these results, it is plausible that tumors missed by the miR-Test may be rather indolent or even represent overdiagnosis by LDCT. Notably, however, none of the individuals in the Validation Set who were assigned a low risk score, either in the two- or three-category stratification, died of lung cancer in the follow-up period (>30 months) (Table 5).
Table 5 shows the performance of the miR-Test in various tests. The number of individuals assigned to difference miR-Test categories is reported. In brackets, percentage out of total. Tumor stage was defined based on the TNM Classification of Malignant Tumors published by the International Union Against Cancer (UICC), 7th edition. §miR-Test performance using two-category stratification Pos: ≧0; Neg: <0. In Validation Set, sensitivity (SE) was 78% and specificity (SP) was 75%. In Specificity Set, SP was 87%. In Clinical Set, SE was 70%. * miR-Test performance using three-category stratification: High: ≧5; Intermediate: <5 and ≧−5; Low: <−5.
A series of experiments to analyze the performance of the miR-Test in various clinically relevant settings was performed next. To assess the ability of the miR-Test to distinguish between non-malignant lung disease (NMD) and lung cancer, a “Specificity Set” was assembled selecting individuals from the COSMOS study (
The miR-Test was then applied to a third independent set, the “Clinical Set”, composed of patients who were diagnosed with lung tumors outside of the COSMOS trial (
Finally, serum samples of a group of stage I, non-small cell lung cancer (NSCLC) patients before and after surgery were analyzed, to evaluate miR-Test tumor specificity (see Example 1). At one month post-surgery, there was no significant overall decrease in the miR-Test risk scores (p=0.39;
An interesting biological question pertains to the origin of circulating miRNAs and of their fluctuations in cancer patients. miRNA might be present in the serum due to their passive release from apoptotic cells, or active release from in microvesicles or in complex with miRNA-binding proteins, which protect them from degradation.
The expression of circulating miRNAs in the NCI-60 tumor cells dataset was analyzed. While such an approach has obvious caveats (e.g., expressed miRNAs are not necessarily secreted, and even if they are, their levels of expression might not reflect their levels of secretion), it was a feasible initial step which could direct further analyses.
Initially, an unsupervised clustering analysis of all miRNAs that can be reliably detected in serum (147-miRNA, see Bianchi et al.) was performed. This yielded three main groups: i) miRNAs preferentially expressed in lines of hematopoietic origin, which we named the “inflammatory-like” cluster; ii) miRNAs preferentially expressed in lines of epithelial origin, which we named the “epithelial-like” cluster; iii) miRNAs without a clear differential pattern of expression (
The above findings suggest a dual origin for the miRNAs of our signatures. Thus, the impact of the two components on the performance of the signature was investigated. First, it was established that both components are needed to maintain a good compromise between sensitivity and specificity (Table 7). Next, the levels of the epithelial- and inflammatory-like miRNAs, present in our signatures, in sera from individuals in the Validation Set were analyzed. The “inflammatory-like” miRNAs displayed a quasi-stereotypical behavior, with 5 of 6 miRNAs showing an increase in cancer patients vs. healthy individuals (
Finally, the expression of the epithelial- and inflammatory-like miRNAs in miR-Test-positive individuals in the Validation and Specificity Sets (
Significantly regulated miRNAs of the 34-miRNA (or 13-miRNA) signature in the NCI-60 cell line panel, restricted to lung cancer (LC) and leukemic cell lines (LE) are shown in Table 6. Gene expression data were centered on their means (miRNA ID, microRNA OSU V3 chip relative pre-miRNA ID; Accession, miRBASE accession number; miRNA, mature miRNA name; in bold, mature miRNAs (N=16) belonging to the original 34-miRNA signature; miR-32 was excluded from the analysis because of its presence in both clusters; Cluster Type, identification of miRNA based on clustering analysis as described in the main text). P-values were calculated by the two-tailed t-test.
The sensitivity (SEN) and specificity (SPE) of the complete 13-miRNA miR-Test, and the miR-Test without the inflammatory-like (Epithelial-like) or epithelial-like (Inflam-like) components, in the Validation set (N=1008) is shown in Table 7. A risk score of 0 was used as cutoff to assign positive and negative test results.
The details of one or more embodiments of the disclosure are set forth in the accompanying description above. Any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present methods and materials now described. Other features, objects, and advantages of the invention will be apparent from the description and from the claims. In the specification and the appended claims, the singular forms include plural referents unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. All patents and publications cited in this specification are incorporated by reference.
The foregoing description has been presented only for the purposes of illustration and is not intended to limit the invention to the precise form disclosed, but by the claims appended hereto.
This application claims priority to and benefit of provisional application U.S. Ser. No. 62/047,932 filed on Sep. 9, 2014, the contents of which are incorporated herein by reference in their entirety.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2015/070664 | 9/9/2015 | WO | 00 |
| Number | Date | Country | |
|---|---|---|---|
| 62047932 | Sep 2014 | US |
