Utility of a Molecular Classifier as a Complement to High-Resolution Computed Tomography to Identify Usual Interstitial Pneumonia.

Rationale: Usual interstitial pneumonia (UIP) is the defining morphology of idiopathic pulmonary fibrosis (IPF). Guidelines for IPF diagnosis conditionally recommend surgical lung biopsy for histopathology diagnosis of UIP when radiology and clinical context are not definitive. A "molecular diagnosis of UIP" in transbronchial lung biopsy, the Envisia Genomic Classifier, accurately predicted histopathologic UIP.Objectives: We evaluated the combined accuracy of the Envisia Genomic Classifier and local radiology in the detection of UIP pattern.Methods: Ninety-six patients who had diagnostic lung pathology as well as a transbronchial lung biopsy for molecular testing with Envisia Genomic Classifier were included in this analysis. The classifier results were scored against reference pathology. UIP identified on high-resolution computed tomography (HRCT) as documented by features in local radiologists' reports was compared with histopathology.Measurements and Main Results: In 96 patients, the Envisia Classifier achieved a specificity of 92.1% (confidence interval [CI],78.6-98.3%) and a sensitivity of 60.3% (CI, 46.6-73.0%) for histology-proven UIP pattern. Local radiologists identified UIP in 18 of 53 patients with UIP histopathology, with a sensitivity of 34.0% (CI, 21.5-48.3%) and a specificity of 96.9% (CI, 83.8-100%). In conjunction with HRCT patterns of UIP, the Envisia Classifier results identified 24 additional patients with UIP (sensitivity 79.2%; specificity 90.6%).Conclusions: In 96 patients with suspected interstitial lung disease, the Envisia Genomic Classifier identified UIP regardless of HRCT pattern. These results suggest that recognition of a UIP pattern by the Envisia Genomic Classifier combined with HRCT and clinical factors in a multidisciplinary discussion may assist clinicians in making an interstitial lung disease (especially IPF) diagnosis without the need for a surgical lung biopsy.

Analytical validation of the Percepta genomic sequencing classifier; an RNA next generation sequencing assay for the assessment of Lung Cancer risk of suspicious pulmonary nodules.

BACKGROUND: Bronchoscopy is a common procedure used for evaluation of suspicious lung nodules, but the low diagnostic sensitivity of bronchoscopy often results in inconclusive results and delays in treatment. Percepta Genomic Sequencing Classifier (GSC) was developed to assist with patient management in cases where bronchoscopy is inconclusive. Studies have shown that exposure to tobacco smoke alters gene expression in airway epithelial cells in a way that indicates an increased risk of developing lung cancer. Percepta GSC leverages this idea of a molecular "field of injury" from smoking and was developed using RNA sequencing data generated from lung bronchial brushings of the upper airway. A Percepta GSC score is calculated from an ensemble of machine learning algorithms utilizing clinical and genomic features and is used to refine a patient's risk stratification. METHODS: The objective of the analysis described and reported here is to validate the analytical performance of Percepta GSC. Analytical performance studies characterized the sensitivity of Percepta GSC test results to input RNA quantity, the potentially interfering agents of blood and genomic DNA, and the reproducibility of test results within and between processing runs and between laboratories. RESULTS: Varying the amount of input RNA into the assay across a nominal range had no significant impact on Percepta GSC classifier results. Bronchial brushing RNA contaminated with up to 10% genomic DNA by nucleic acid mass also showed no significant difference on classifier results. The addition of blood RNA, a potential contaminant in the bronchial brushing sample, caused no change to classifier results at up to 11% contamination by RNA proportion. Percepta GSC scores were reproducible between runs, within runs, and between laboratories, varying within less than 4% of the total score range (standard deviation of 0.169 for scores on 4.57 scale). CONCLUSIONS: The analytical sensitivity, analytical specificity, and reproducibility of Percepta GSC laboratory results were successfully demonstrated under conditions of expected day to day variation in testing. Percepta GSC test results are analytically robust and suitable for routine clinical use.

Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning: challenges and solutions.

BACKGROUND: We developed a classifier using RNA sequencing data that identifies the usual interstitial pneumonia (UIP) pattern for the diagnosis of idiopathic pulmonary fibrosis. We addressed significant challenges, including limited sample size, biological and technical sample heterogeneity, and reagent and assay batch effects. RESULTS: We identified inter- and intra-patient heterogeneity, particularly within the non-UIP group. The models classified UIP on transbronchial biopsy samples with a receiver-operating characteristic area under the curve of ~ 0.9 in cross-validation. Using in silico mixed samples in training, we prospectively defined a decision boundary to optimize specificity at ≥85%. The penalized logistic regression model showed greater reproducibility across technical replicates and was chosen as the final model. The final model showed sensitivity of 70% and specificity of 88% in the test set. CONCLUSIONS: We demonstrated that the suggested methodologies appropriately addressed challenges of the sample size, disease heterogeneity and technical batch effects and developed a highly accurate and robust classifier leveraging RNA sequencing for the classification of UIP.

Analytical performance of Envisia: a genomic classifier for usual interstitial pneumonia.

BACKGROUND: Clinical guidelines specify that diagnosis of interstitial pulmonary fibrosis (IPF) requires identification of usual interstitial pneumonia (UIP) pattern. While UIP can be identified by high resolution CT of the chest, the results are often inconclusive, making surgical lung biopsy necessary to reach a definitive diagnosis (Raghu et al., Am J Respir Crit Care Med 183(6):788-824, 2011). The Envisia genomic classifier differentiates UIP from non-UIP pathology in transbronchial biopsies (TBB), potentially allowing patients to avoid an invasive procedure (Brown et al., Am J Respir Crit Care Med 195:A6792, 2017). To ensure patient safety and efficacy, a laboratory developed test (LDT) must meet strict regulatory requirements for accuracy, reproducibility and robustness. The analytical characteristics of the Envisia test are assessed and reported here. METHODS: The Envisia test utilizes total RNA extracted from TBB samples to perform Next Generation RNA Sequencing. The gene count data from 190 genes are then input to the Envisia genomic classifier, a machine learning algorithm, to output either a UIP or non-UIP classification result. We characterized the stability of RNA in TBBs during collection and shipment, and evaluated input RNA mass and proportions on the limit of detection of UIP. We evaluated potentially interfering substances such as blood and genomic DNA. Intra-run, inter-run, and inter-laboratory reproducibility of test results were also characterized. RESULTS: RNA content within TBBs preserved in RNAprotect is stable for up to 14 days with no detectable change in RNA quality. The Envisia test is tolerant to variation in RNA input (5 to 30 ng), with no impact on classifier results. The Envisia test can tolerate dilution of non-UIP and UIP classification signals at the RNA level by up to 60% and 20%, respectively. Analytical specificity studies utilizing UIP and non-UIP samples mixed with genomic DNA (up to 30% relative input) demonstrated no impact to classifier results. The Envisia test tolerates up to 22% of blood contamination, well beyond the level observed in TBBs. The test is reproducible from RNA extraction through to Envisia test result (standard deviation of 0.20 for Envisia classification scores on > 7-unit scale). CONCLUSIONS: The Envisia test demonstrates the robust analytical performance required of an LDT. Envisia can be used to inform the diagnoses of patients with suspected IPF.

The diagnostic application of RNA sequencing in patients with thyroid cancer: an analysis of 851 variants and 133 fusions in 524 genes.

BACKGROUND: Thyroid carcinomas are known to harbor oncogenic driver mutations and advances in sequencing technology now allow the detection of these in fine needle aspiration biopsies (FNA). Recent work by The Cancer Genome Atlas (TCGA) Research Network has expanded the number of genetic alterations detected in papillary thyroid carcinomas (PTC). We sought to investigate the prevalence of these and other genetic alterations in diverse subtypes of thyroid nodules beyond PTC, including a variety of samples with benign histopathology. This is the first clinical evaluation of a large panel of TCGA-reported genomic alterations in thyroid FNAs. RESULTS: In FNAs, genetic alterations were detected in 19/44 malignant samples (43% sensitivity) and in 7/44 histopathology benign samples (84% specificity). Overall, after adding a cohort of tissue samples, 38/76 (50%) of histopathology malignant samples were found to harbor a genetic alteration, while 15/75 (20%) of benign samples were also mutated. The most frequently mutated malignant subtypes were medullary thyroid carcinoma (9/12, 75%) and PTC (14/30, 47%). Additionally, follicular adenoma, a benign subtype of thyroid neoplasm, was also found to harbor mutations (12/29, 41%). Frequently mutated genes in malignant samples included BRAF (20/76, 26%) and RAS (9/76, 12%). Of the TSHR variants detected, (6/7, 86%) were in benign nodules. In a direct comparison of the same FNA also tested by an RNA-based gene expression classifier (GEC), the sensitivity of genetic alterations alone was 42%, compared to the 91% sensitivity achieved by the GEC. The specificity based only on genetic alterations was 84%, compared to 77% specificity with the GEC. CONCLUSIONS: While the genomic landscape of all thyroid neoplasm subtypes will inevitably be elucidated, caution should be used in the early adoption of published mutations as the sole predictor of malignancy in thyroid. The largest set of such mutations known to date detects only a portion of thyroid carcinomas in preoperative FNAs in our cohort and thus is not sufficient to rule out cancer. Due to the finding that variants are also found in benign nodules, testing only GEC suspicious nodules may be helpful in avoiding false positives and altering the extent of treatment when selected mutations are found.

Analytical performance of a bronchial genomic classifier.

BACKGROUND: The current standard practice of lung lesion diagnosis often leads to inconclusive results, requiring additional diagnostic follow up procedures that are invasive and often unnecessary due to the high benign rate in such lesions (Chest 143:e78S-e92, 2013). The Percepta bronchial genomic classifier was developed and clinically validated to provide more accurate classification of lung nodules and lesions that are inconclusive by bronchoscopy, using bronchial brushing specimens (N Engl J Med 373:243-51, 2015, BMC Med Genomics 8:18, 2015). The analytical performance of the Percepta test is reported here. METHODS: Analytical performance studies were designed to characterize the stability of RNA in bronchial brushing specimens during collection and shipment; analytical sensitivity defined as input RNA mass; analytical specificity (i.e. potentially interfering substances) as tested on blood and genomic DNA; and assay performance studies including intra-run, inter-run, and inter-laboratory reproducibility. RESULTS: RNA content within bronchial brushing specimens preserved in RNAprotect is stable for up to 20 days at 4 °C with no changes in RNA yield or integrity. Analytical sensitivity studies demonstrated tolerance to variation in RNA input (157 ng to 243 ng). Analytical specificity studies utilizing cancer positive and cancer negative samples mixed with either blood (up to 10 % input mass) or genomic DNA (up to 10 % input mass) demonstrated no assay interference. The test is reproducible from RNA extraction through to Percepta test result, including variation across operators, runs, reagent lots, and laboratories (standard deviation of 0.26 for scores on > 6 unit scale). CONCLUSIONS: Analytical sensitivity, analytical specificity and robustness of the Percepta test were successfully verified, supporting its suitability for clinical use.

Preoperative diagnosis of benign thyroid nodules with indeterminate cytology.

BACKGROUND: Approximately 15 to 30% of thyroid nodules evaluated by means of fine-needle aspiration are not clearly benign or malignant. Patients with cytologically indeterminate nodules are often referred for diagnostic surgery, though most of these nodules prove to be benign. A novel diagnostic test that measures the expression of 167 genes has shown promise in improving preoperative risk assessment. METHODS: We performed a 19-month, prospective, multicenter validation study involving 49 clinical sites, 3789 patients, and 4812 fine-needle aspirates from thyroid nodules 1 cm or larger that required evaluation. We obtained 577 cytologically indeterminate aspirates, 413 of which had corresponding histopathological specimens from excised lesions. Results of a central, blinded histopathological review served as the reference standard. After inclusion criteria were met, a gene-expression classifier was used to test 265 indeterminate nodules in this analysis, and its performance was assessed. RESULTS: Of the 265 indeterminate nodules, 85 were malignant. The gene-expression classifier correctly identified 78 of the 85 nodules as suspicious (92% sensitivity; 95% confidence interval [CI], 84 to 97), with a specificity of 52% (95% CI, 44 to 59). The negative predictive values for "atypia (or follicular lesion) of undetermined clinical significance," "follicular neoplasm or lesion suspicious for follicular neoplasm," or "suspicious cytologic findings" were 95%, 94%, and 85%, respectively. Analysis of 7 aspirates with false negative results revealed that 6 had a paucity of thyroid follicular cells, suggesting insufficient sampling of the nodule. CONCLUSIONS: These data suggest consideration of a more conservative approach for most patients with thyroid nodules that are cytologically indeterminate on fine-needle aspiration and benign according to gene-expression classifier results. (Funded by Veracyte.).

A Nasal Swab Classifier to Evaluate the Probability of Lung Cancer in Patients With Pulmonary Nodules.

BACKGROUND: Accurate assessment of the probability of lung cancer (pCA) is critical in patients with pulmonary nodules (PNs) to help guide decision-making. We sought to validate a clinical-genomic classifier developed using whole-transcriptome sequencing of nasal epithelial cells from patients with a PN ≤ 30 mm who smoke or have previously smoked. RESEARCH QUESTION: Can the pCA in individuals with a PN and a history of smoking be predicted by a classifier that uses clinical factors and genomic data from nasal epithelial cells obtained by cytologic brushing? STUDY DESIGN AND METHODS: Machine learning was used to train a classifier using genomic and clinical features on 1,120 patients with PNs labeled as benign or malignant established by a final diagnosis or a minimum of 12 months of radiographic surveillance. The classifier was designed to yield low-, intermediate-, and high-risk categories. The classifier was validated in an independent set of 312 patients, including 63 patients with a prior history of cancer (other than lung cancer), comparing the classifier prediction with the known clinical outcome. RESULTS: In the primary validation set, sensitivity and specificity for low-risk classification were 96% and 42%, whereas sensitivity and specificity for high-risk classification was 58% and 90%, respectively. Sensitivity was similar across stages of non-small cell lung cancer, independent of subtype. Performance compared favorably with clinical-only risk models. Analysis of 63 patients with prior cancer showed similar performance as did subanalyses of patients with light vs heavy smoking burden and those eligible for lung cancer screening vs those who were not. INTERPRETATION: The nasal classifier provides an accurate assessment of pCA in individuals with a PN ≤ 30 mm who smoke or have previously smoked. Classifier-guided decision-making could lead to fewer diagnostic procedures in patients without cancer and more timely treatment in patients with lung cancer.

Analytical Validation of a Telomerase Reverse Transcriptase (TERT) Promoter Mutation Assay.

CONTEXT: TERT promoter mutated thyroid cancers are associated with a decreased rate of disease free and disease specific survival. High quality analytical validation of a diagnostic test promotes confidence in the results which inform clinical decision making. OBJECTIVE: To demonstrate the analytical validation of the Afirma TERT promoter mutation assay. METHODS: TERT promoter C228T and C250T variant detection in genomic DNA (gDNA) was analyzed by assessing variable DNA input and the limit of detection (LOD) of variant allele frequency (VAF). The negative and positive percent agreement (NPA and PPA) of the Afirma TERT test was examined against a reference primer pair as was the analytical specificity from potential interfering substances (RNA and blood gDNA). Further, the intra-run, inter-run and inter-laboratory reproducibility of the assay were tested. RESULTS: The Afirma TERT test is tolerant to variation in DNA input amount (7-13 ng) and can detect expected positive TERT promoter variants down to 5% VAF LOD at 7ng DNA input with > 95% sensitivity. Both NPA and PPA were 100% against the reference primer pair. The test remains accurate in presence of 20% RNA or 80% blood gDNA for an average patient sample that typically has 30% VAF. The test also demonstrated a 100% confirmation rate when compared with an external NGS-based reference assay executed in a non-Veracyte laboratory. CONCLUSION: The analytical robustness and reproducibility of the Afirma TERT test support its routine clinical use among thyroid nodules with indeterminate cytology that are Afirma GSC suspicious or among Bethesda V/VI nodules.

Maximizing Small Biopsy Patient Samples: Unified RNA-Seq Platform Assessment of over 120,000 Patient Biopsies.

Despite its wide-ranging benefits, whole-transcriptome or RNA exome profiling is challenging to implement in a clinical diagnostic setting. The Unified Assay is a comprehensive workflow wherein exome-enriched RNA-sequencing (RNA-Seq) assays are performed on clinical samples and analyzed by a series of advanced machine learning-based classifiers. Gene expression signatures and rare and/or novel genomic events, including fusions, mitochondrial variants, and loss of heterozygosity were assessed using RNA-Seq data generated from 120,313 clinical samples across three clinical indications (thyroid cancer, lung cancer, and interstitial lung disease). Since its implementation, the data derived from the Unified Assay have allowed significantly more patients to avoid unnecessary diagnostic surgery and have played an important role in guiding follow-up decisions regarding treatment. Collectively, data from the Unified Assay show the utility of RNA-Seq and RNA expression signatures in the clinical laboratory, and their importance to the future of precision medicine.

Clinical validation and utility of Percepta GSC for the evaluation of lung cancer.

The Percepta Genomic Sequencing Classifier (GSC) was developed to up-classify as well as down-classify the risk of malignancy for lung lesions when bronchoscopy is non-diagnostic. We evaluated the performance of Percepta GSC in risk re-classification of indeterminate lung lesions. This multicenter study included individuals who currently or formerly smoked undergoing bronchoscopy for suspected lung cancer from the AEGIS I/ II cohorts and the Percepta Registry. The classifier was measured in normal-appearing bronchial epithelium from bronchial brushings. The sensitivity, specificity, and predictive values were calculated using predefined thresholds. The ability of the classifier to decrease unnecessary invasive procedures was estimated. A set of 412 patients were included in the validation (prevalence of malignancy was 39.6%). Overall, 29% of intermediate-risk lung lesions were down-classified to low-risk with a 91.0% negative predictive value (NPV) and 12.2% of intermediate-risk lesions were up-classified to high-risk with a 65.4% positive predictive value (PPV). In addition, 54.5% of low-risk lesions were down-classified to very low risk with >99% NPV and 27.3% of high-risk lesions were up-classified to very high risk with a 91.5% PPV. If the classifier results were used in nodule management, 50% of patients with benign lesions and 29% of patients with malignant lesions undergoing additional invasive procedures could have avoided these procedures. The Percepta GSC is highly accurate as both a rule-out and rule-in test. This high accuracy of risk re-classification may lead to improved management of lung lesions.

Preoperative Identification of Medullary Thyroid Carcinoma (MTC): Clinical Validation of the Afirma MTC RNA-Sequencing Classifier.

Background: Cytopathological evaluation of thyroid fine-needle aspiration biopsy (FNAB) specimens can fail to raise preoperative suspicion of medullary thyroid carcinoma (MTC). The Afirma RNA-sequencing MTC classifier identifies MTC among FNA samples that are cytologically indeterminate, suspicious, or malignant (Bethesda categories III-VI). In this study we report the development and clinical performance of this MTC classifier. Methods: Algorithm training was performed with a set of 483 FNAB specimens (21 MTC and 462 non-MTC). A support vector machine classifier was developed using 108 differentially expressed genes, which includes the 5 genes in the prior Afirma microarray-based MTC cassette. Results: The final MTC classifier was blindly tested on 211 preoperative FNAB specimens with subsequent surgical pathology, including 21 MTC and 190 non-MTC specimens from benign and malignant thyroid nodules independent from those used in training. The classifier had 100% sensitivity (21/21 MTC FNAB specimens correctly called positive; 95% confidence interval [CI] = 83.9-100%) and 100% specificity (190/190 non-MTC FNAs correctly called negative; CI = 98.1-100%). All positive samples had pathological confirmation of MTC, while all negative samples were negative for MTC on surgical pathology. Conclusions: The RNA-sequencing MTC classifier accurately identified MTC from preoperative thyroid nodule FNAB specimens in an independent validation cohort. This identification may facilitate an MTC-specific preoperative evaluation and resulting treatment.

Improving lung cancer risk stratification leveraging whole transcriptome RNA sequencing and machine learning across multiple cohorts.

BACKGROUND: Bronchoscopy for suspected lung cancer has low diagnostic sensitivity, rendering many inconclusive results. The Bronchial Genomic Classifier (BGC) was developed to help with patient management by identifying those with low risk of lung cancer when bronchoscopy is inconclusive. The BGC was trained and validated on patients in the Airway Epithelial Gene Expression in the Diagnosis of Lung Cancer (AEGIS) trials. A modern patient cohort, the BGC Registry, showed differences in key clinical factors from the AEGIS cohorts, with less smoking history, smaller nodules and older age. Additionally, we discovered interfering factors (inhaled medication and sample collection timing) that impacted gene expressions and potentially disguised genomic cancer signals. METHODS: In this study, we leveraged multiple cohorts and next generation sequencing technology to develop a robust Genomic Sequencing Classifier (GSC). To address demographic composition shift and interfering factors, we synergized three algorithmic strategies: 1) ensemble of clinical dominant and genomic dominant models; 2) development of hierarchical regression models where the main effects from clinical variables were regressed out prior to the genomic impact being fitted in the model; and 3) targeted placement of genomic and clinical interaction terms to stabilize the effect of interfering factors. The final GSC model uses 1232 genes and four clinical covariates - age, pack-years, inhaled medication use, and specimen collection timing. RESULTS: In the validation set (N = 412), the GSC down-classified low and intermediate pre-test risk subjects to very low and low post-test risk with a specificity of 45% (95% CI 37-53%) and a sensitivity of 91% (95%CI 81-97%), resulting in a negative predictive value of 95% (95% CI 89-98%). Twelve percent of intermediate pre-test risk subjects were up-classified to high post-test risk with a positive predictive value of 65% (95%CI 44-82%), and 27% of high pre-test risk subjects were up-classified to very high post-test risk with a positive predictive value of 91% (95% CI 78-97%). CONCLUSIONS: The GSC overcame the impact of interfering factors and achieved consistent performance across multiple cohorts. It demonstrated diagnostic accuracy in both down- and up-classification of cancer risk, providing physicians actionable information for many patients with inconclusive bronchoscopy.

Analytical Verification Performance of Afirma Genomic Sequencing Classifier in the Diagnosis of Cytologically Indeterminate Thyroid Nodules.

Background: Fine needle aspiration (FNA) cytology, a diagnostic test central to thyroid nodule management, may yield indeterminate results in up to 30% of cases. The Afirma® Genomic Sequencing Classifier (GSC) was developed and clinically validated to utilize genomic material obtained during the FNA to accurately identify benign nodules among those deemed cytologically indeterminate so that diagnostic surgery can be avoided. A key question for diagnostic tests is their robustness under different perturbations that may occur in the lab. Herein, we describe the analytical performance of the Afirma GSC. Results: We examined the analytical sensitivity of the Afirma GSC to varied input RNA amounts and the limit of detection of malignant signals with heterogenous samples mixed with adjacent normal or benign tissues. We also evaluated the analytical specificity from potential interfering substances such as blood and genomic DNA. Further, the inter-laboratory, intra-run, and inter-run reproducibility of the assay were examined. Analytical sensitivity analysis showed that Afirma GSC calls are tolerant to variation in RNA input amount (5-30 ng), and up to 75% dilution of malignant FNA material. Analytical specificity studies demonstrated Afirma GSC remains accurate in presence of up to 75% blood or 30% genomic DNA. The Afirma GSC results are highly reproducible across different operators, runs, reagent lots, and laboratories. Conclusion: The analytical robustness and reproducibility of the Afirma GSC test support its routine clinical use among thyroid nodules with indeterminant FNA cytology.

Identification of Hürthle cell cancers: solving a clinical challenge with genomic sequencing and a trio of machine learning algorithms.

BACKGROUND: Identification of Hürthle cell cancers by non-operative fine-needle aspiration biopsy (FNAB) of thyroid nodules is challenging. Resultingly, non-cancerous Hürthle lesions were conventionally distinguished from Hürthle cell cancers by histopathological examination of tissue following surgical resection. Reliance on histopathological evaluation requires patients to undergo surgery to obtain a diagnosis despite most being non-cancerous. It is highly desirable to avoid surgery and to provide accurate classification of benignity versus malignancy from FNAB preoperatively. In our first-generation algorithm, Gene Expression Classifier (GEC), we achieved this goal by using machine learning (ML) on gene expression features. The classifier is sensitive, but not specific due in part to the presence of non-neoplastic benign Hürthle cells in many FNAB. RESULTS: We sought to overcome this low-specificity limitation by expanding the feature set for ML using next-generation whole transcriptome RNA sequencing and called the improved algorithm the Genomic Sequencing Classifier (GSC). The Hürthle identification leverages mitochondrial expression and we developed novel feature extraction mechanisms to measure chromosomal and genomic level loss-of-heterozygosity (LOH) for the algorithm. Additionally, we developed a multi-layered system of cascading classifiers to sequentially triage Hürthle cell-containing FNAB, including: 1. presence of Hürthle cells, 2. presence of neoplastic Hürthle cells, and 3. presence of benign Hürthle cells. The final Hürthle cell Index utilizes 1048 nuclear and mitochondrial genes; and Hürthle cell Neoplasm Index leverages LOH features as well as 2041 genes. Both indices are Support Vector Machine (SVM) based. The third classifier, the GSC Benign/Suspicious classifier, utilizes 1115 core genes and is an ensemble classifier incorporating 12 individual models. CONCLUSIONS: The accurate algorithmic depiction of this complex biological system among Hürthle subtypes results in a dramatic improvement of classification performance; specificity among Hürthle cell neoplasms increases from 11.8% with the GEC to 58.8% with the GSC, while maintaining the same sensitivity of 89%.

Analytical and Clinical Validation of Expressed Variants and Fusions From the Whole Transcriptome of Thyroid FNA Samples.

Introduction: The Afirma® Xpression Atlas (XA) detects gene variants and fusions in thyroid nodule FNA samples from a curated panel of 511 genes using whole-transcriptome RNA-sequencing. Its intended use is among cytologically indeterminate nodules that are Afirma GSC suspicious, Bethesda V/VI nodules, or known thyroid metastases. Here we report its analytical and clinical validation. Methods: DNA and RNA were purified from the same sample across 943 blinded FNAs and compared by multiple methodologies, including whole-transcriptome RNA-seq, targeted RNA-seq, and targeted DNA-seq. An additional 695 blinded FNAs were used to define performance for fusions between whole-transcriptome RNA-seq and targeted RNA-seq. We quantified the reproducibility of the whole-transcriptome RNA-seq assay across laboratories and reagent lots. Finally, variants and fusions were compared to histopathology results. Results: Of variants detected in DNA at 5 or 20% variant allele frequency, 74 and 88% were also detected by XA, respectively. XA variant detection was 89% when compared to an alternative RNA-based detection method. Low levels of expression of the DNA allele carrying the variant, compared with the wild-type allele, was found in some variants not detected by XA. 82% of gene fusions detected in a targeted RNA fusion assay were detected by XA. Conversely, nearly all variants or fusions detected by XA were confirmed by an alternative method. Analytical validation studies demonstrated high intra-plate reproducibility (89%-94%), inter-plate reproducibility (86-91%), and inter-lab accuracy (90%). Multiple variants and fusions previously described across the spectrum of thyroid cancers were identified by XA, including some with approved or investigational targeted therapies. Among 190 Bethesda III/IV nodules, the sensitivity of XA as a standalone test was 49%. Conclusion: When the Afirma Genomic Sequencing Classifier (GSC) is used first among Bethesda III/IV nodules as a rule-out test, XA supplements genomic insight among those that are GSC suspicious. Our data clinically and analytically validate XA for use among GSC suspicious, or Bethesda V/VI nodules. Genomic information provided by XA may inform clinical decision-making with precision medicine insights across a broad range of FNA sample types encountered in the care of patients with thyroid nodules and thyroid cancer.

Use of a molecular classifier to identify usual interstitial pneumonia in conventional transbronchial lung biopsy samples: a prospective validation study.

BACKGROUND: In the appropriate clinical setting, the diagnosis of idiopathic pulmonary fibrosis (IPF) requires a pattern of usual interstitial pneumonia to be present on high-resolution chest CT (HRCT) or surgical lung biopsy. A molecular usual interstitial pneumonia signature can be identified by a machine learning algorithm in less-invasive transbronchial lung biopsy samples. We report prospective findings for the clinical validity and utility of this molecular test. METHODS: We prospectively recruited 237 patients for this study from those enrolled in the Bronchial Sample Collection for a Novel Genomic Test (BRAVE) study in 29 US and European sites. Patients were undergoing evaluation for interstitial lung disease and had had samples obtained by clinically indicated surgical or transbronchial biopsy or cryobiopsy for pathology. Histopathological diagnoses were made by experienced pathologists. Available HRCT scans were reviewed centrally. Three to five transbronchial lung biopsy samples were collected from all patients specifically for this study, pooled by patient, and extracted for transcriptomic sequencing. After exclusions, diagnostic histopathology and RNA sequence data from 90 patients were used to train a machine learning algorithm (Envisia Genomic Classifier, Veracyte, San Francisco, CA, USA) to identify a usual interstitial pneumonia pattern. The primary study endpoint was validation of the classifier in 49 patients by comparison with diagnostic histopathology. To assess clinical utility, we compared the agreement and confidence level of diagnosis made by central multidisciplinary teams based on anonymised clinical information and radiology results plus either molecular classifier or histopathology results. FINDINGS: The classifier identified usual interstitial pneumonia in transbronchial lung biopsy samples from 49 patients with 88% specificity (95% CI 70-98) and 70% sensitivity (47-87). Among 42 of these patients who had possible or inconsistent usual interstitial pneumonia on HRCT, the classifier showed 81% positive predictive value (95% CI 54-96) for underlying biopsy-proven usual interstitial pneumonia. In the clinical utility analysis, we found 86% agreement (95% CI 78-92) between clinical diagnoses using classifier results and those using histopathology data. Diagnostic confidence was improved by the molecular classifier results compared with histopathology results in 18 with IPF diagnoses (proportion of diagnoses that were confident or provisional with high confidence 89% vs 56%, p=0·0339) and in all 48 patients with non-diagnostic pathology or non-classifiable fibrosis histopathology (63% vs 42%, p=0·0412). INTERPRETATION: The molecular test provided an objective method to aid clinicians and multidisciplinary teams in ascertaining a diagnosis of IPF, particularly for patients without a clear radiological diagnosis, in samples that can be obtained by a less invasive method. Further prospective clinical validation and utility studies are planned. FUNDING: Veracyte.

Performance of a Genomic Sequencing Classifier for the Preoperative Diagnosis of Cytologically Indeterminate Thyroid Nodules.

Importance: Use of next-generation sequencing of RNA and machine learning algorithms can classify the risk of malignancy in cytologically indeterminate thyroid nodules to limit unnecessary diagnostic surgery. Objective: To measure the performance of a genomic sequencing classifier for cytologically indeterminate thyroid nodules. Design, Setting, and Participants: A blinded validation study was conducted on a set of cytologically indeterminate thyroid nodules collected by fine-needle aspiration biopsy between June 2009 and December 2010 from 49 academic and community centers in the United States. All patients underwent surgery without genomic information and were assigned a histopathology diagnosis by an expert panel blinded to all genomic information. There were 210 potentially eligible thyroid biopsy samples with Bethesda III or IV indeterminate cytopathology that constituted a cohort previously used to validate the gene expression classifier. Of these, 191 samples (91.0%) had adequate residual RNA for validation of the genomic sequencing classifier. Algorithm development and independent validation occurred between August 2016 and May 2017. Exposures: Thyroid nodule surgical histopathology diagnosis by an expert panel blinded to all genomic data. Main Outcomes and Measures: The primary end point was measurement of genomic sequencing classifier sensitivity, specificity, and negative and positive predictive values in biopsies from Bethesda III and IV nodules. The secondary end point was measurement of classifier performance in biopsies from Bethesda II, V, and VI nodules. Results: Of the 183 included patients, 142 (77.6%) were women, and the mean (range) age was 51.7 (22.0-85.0) years. The genomic sequencing classifier had a sensitivity of 91% (95% CI, 79-98) and a specificity of 68% (95% CI, 60-76). At 24% cancer prevalence, the negative predictive value was 96% (95% CI, 90-99) and the positive predictive value was 47% (95% CI, 36-58). Conclusions and Relevance: The genomic sequencing classifier demonstrates high sensitivity and accuracy for identifying benign nodules. Its 36% increase in specificity compared with the gene expression classifier potentially increases the number of patients with benign nodules who can safely avoid unnecessary diagnostic surgery.

Usual Interstitial Pneumonia Can Be Detected in Transbronchial Biopsies Using Machine Learning.

RATIONALE: Usual interstitial pneumonia (UIP) is the histopathologic hallmark of idiopathic pulmonary fibrosis. Although UIP can be detected by high-resolution computed tomography of the chest, the results are frequently inconclusive, and pathology from transbronchial biopsy (TBB) has poor sensitivity. Surgical lung biopsy may be necessary for a definitive diagnosis. OBJECTIVES: To develop a genomic classifier in tissue obtained by TBB that distinguishes UIP from non-UIP, trained against central pathology as the reference standard. METHODS: Exome enriched RNA sequencing was performed on 283 TBBs from 84 subjects. Machine learning was used to train an algorithm with high rule-in (specificity) performance using specimens from 53 subjects. Performance was evaluated by cross-validation and on an independent test set of specimens from 31 subjects. We explored the feasibility of a single molecular test per subject by combining multiple TBBs from upper and lower lobes. To address whether classifier accuracy depends upon adequate alveolar sampling, we tested for correlation between classifier accuracy and expression of alveolar-specific genes. RESULTS: The top-performing algorithm distinguishes UIP from non-UIP conditions in single TBB samples with an area under the receiver operator characteristic curve (AUC) of 0.86, with specificity of 86% (confidence interval = 71-95%) and sensitivity of 63% (confidence interval = 51-74%) (31 test subjects). Performance improves to an AUC of 0.92 when three to five TBB samples per subject are combined at the RNA level for testing. Although we observed a wide range of type I and II alveolar-specific gene expression in TBBs, expression of these transcripts did not correlate with classifier accuracy. CONCLUSIONS: We demonstrate proof of principle that genomic analysis and machine learning improves the utility of TBB for the diagnosis of UIP, with greater sensitivity and specificity than pathology in TBB alone. Combining multiple individual subject samples results in increased test accuracy over single sample testing. This approach requires validation in an independent cohort of subjects before application in the clinic.