Diagnosing Non-Small Cell Lung Cancer by Exhaled Breath Profiling Using an Electronic Nose: A Multicenter Validation Study

BACKGROUND

Despite the potential of exhaled breath analysis of volatile organic compounds to diagnose lung cancer, clinical implementation has not been realized, partly due to the lack of validation studies.

RESEARCH QUESTION

This study addressed two questions. First, can we simultaneously train and validate a prediction model to distinguish patients with non-small cell lung cancer from non-lung cancer subjects based on exhaled breath patterns? Second, does addition of clinical variables to exhaled breath data improve the diagnosis of lung cancer?

STUDY DESIGN AND METHODS

In this multicenter study, subjects with non-small cell lung cancer and control subjects performed 5 min of tidal breathing through the aeoNose, a handheld electronic nose device. A training cohort was used for developing a prediction model based on breath data, and a blinded cohort was used for validation. Multivariable logistic regression analysis was performed, including breath data and clinical variables, in which the formula and cutoff value for the probability of lung cancer were applied to the validation data.

RESULTS

A total of 376 subjects formed the training set, and 199 subjects formed the validation set. The full training model (including exhaled breath data and clinical parameters from the training set) were combined in a multivariable logistic regression analysis, maintaining a cut off of 16% probability of lung cancer, resulting in a sensitivity of 95%, a specificity of 51%, and a negative predictive value of 94%; the area under the receiver-operating characteristic curve was 0.87. Performance of the prediction model on the validation cohort showed corresponding results with a sensitivity of 95%, a specificity of 49%, a negative predictive value of 94%, and an area under the receiver-operating characteristic curve of 0.86.

INTERPRETATION

Combining exhaled breath data and clinical variables in a multicenter, multi-device validation study can adequately distinguish patients with lung cancer from subjects without lung cancer in a noninvasive manner. This study paves the way to implement exhaled breath analysis in the daily practice of diagnosing lung cancer.

CLINICAL TRIAL REGISTRATION

The Netherlands Trial Register; No.: NL7025; URL: https://trialregister.nl/trial/7025.

Using a stepwise approach to simultaneously develop and validate machine learning based prediction models

Accurate diagnosis of a disease is essential in healthcare. Prediction models, based on classical regression techniques, are widely used in clinical practice. Machine Learning (ML) techniques might be preferred in case of a large amount of data per patient and relatively limited numbers of subjects. However, this increases the risk of overfitting, and external validation is imperative. However, in the field of ML, new and more efficient techniques are developed rapidly, and if recruiting patients for a validation study is time consuming, the ML technique used to develop the first model might have been surpassed by more efficient ML techniques, rendering this original model no longer relevant. We demonstrate a stepwise design for simultaneous development and validation of prediction models based on ML techniques. The design enables - in one study - evaluation of the stability and robustness of a prediction model over increasing sample size as well as assessment of the stability of sensitivity/specificity at a chosen cut-off. This will shorten the time to introduction of a new test in health care. We finally describe how to use regular clinical parameters in conjunction with ML based predictions, to further enhance differentiation between subjects with and without a disease.

Improving lung cancer diagnosis by combining exhaled-breath data and clinical parameters

Introduction

Exhaled-breath analysis of volatile organic compounds could detect lung cancer earlier, possibly leading to improved outcomes. Combining exhaled-breath data with clinical parameters may improve lung cancer diagnosis.

Methods

Based on data from a previous multi-centre study, this article reports additional analyses. 138 subjects with non-small cell lung cancer (NSCLC) and 143 controls without NSCLC breathed into the Aeonose. The diagnostic accuracy, presented as area under the receiver operating characteristic curve (AUC-ROC), of the Aeonose itself was compared with 1) performing a multivariate logistic regression analysis of the distinct clinical parameters obtained, and 2) using this clinical information beforehand in the training process of the artificial neural network (ANN) for the breath analysis.

Results

NSCLC patients (mean±sd age 67.1±9.1 years, 58% male) were compared with controls (62.1±7.0 years, 40.6% male). The AUC-ROC of the classification value of the Aeonose itself was 0.75 (95% CI 0.69–0.81). Adding age, number of pack-years and presence of COPD to this value in a multivariate regression analysis resulted in an improved performance with an AUC-ROC of 0.86 (95% CI 0.81–0.90). Adding these clinical variables beforehand to the ANN for classifying the breath print also led to an improved performance with an AUC-ROC of 0.84 (95% CI 0.79–0.89).

Conclusions

Adding readily available clinical information to the classification value of exhaled-breath analysis with the Aeonose, either post hoc in a multivariate regression analysis or a priori to the ANN, significantly improves the diagnostic accuracy to detect the presence or absence of lung cancer.

Adding readily available clinical information to the classification value of exhaled-breath analysis with the Aeonose significantly improves the diagnostic accuracy to detect the presence or absence of lung cancerhttp://bit.ly/38ps6fH

Multi-centre prospective study on diagnosing subtypes of lung cancer by exhaled-breath analysis

OBJECTIVES

Lung cancer is a leading cause of mortality. Exhaled-breath analysis of volatile organic compounds (VOC's) might detect lung cancer early in the course of the disease, which may improve outcomes. Subtyping lung cancers could be helpful in further clinical decisions.

MATERIALS AND METHODS

In a prospective, multi-centre study, using 10 electronic nose devices, 144 subjects diagnosed with NSCLC and 146 healthy subjects, including subjects considered negative for NSCLC after investigation, breathed into the Aeonose™ (The eNose Company, Zutphen, Netherlands). Also, analyses into subtypes of NSCLC, such as adenocarcinoma (AC) and squamous cell carcinoma (SCC), and analyses of patients with small cell lung cancer (SCLC) were performed.

RESULTS

Choosing a cut-off point to predominantly rule out cancer resulted for NSCLC in a sensitivity of 94.4%, a specificity of 32.9%, a positive predictive value of 58.1%, a negative predictive value (NPV) of 85.7%, and an area under the curve (AUC) of 0.76. For AC sensitivity, PPV, NPV, and AUC were 81.5%, 56.4%, 79.5%, and 0.74, respectively, while for SCC these numbers were 80.8%, 45.7%, 93.0%, and 0.77, respectively. SCLC could be ruled out with a sensitivity of 88.9% and an NPV of 96.8% with an AUC of 0.86.

CONCLUSION

Electronic nose technology with the Aeonose™ can play an important role in rapidly excluding lung cancer due to the high negative predictive value for various, but not all types of lung cancer. Patients showing positive breath tests should still be subjected to further diagnostic testing.

Quality of life and adherence to inhaled corticosteroids and tiotropium in COPD are related

Background

Poor adherence to inhaled medications in COPD patients seems to be associated with an increased risk of death and hospitalization. Knowing the determinants of nonadherence to inhaled medications is important for creating interventions to improve adherence.

Objectives

To identify disease-specific and health-related quality of life (HRQoL) factors, associated with adherence to inhaled corticosteroids (ICS) and tiotropium in COPD patients.

Methods

Adherence of 795 patients was recorded over 3 years and was deemed optimal at >75%–≤125%, suboptimal at ≥50%–<75%, and poor at <50% (underuse) or >125% (overuse). Health-related quality of life was measured with the Clinical COPD Questionnaire and the EuroQol-5D questionnaire.

Results

Patients with a higher forced expiratory volume in 1 second (FEV₁)/vital capacity (VC) (odds ratio [OR] =1.03) and ≥1 hospitalizations in the year prior to inclusion in this study (OR =2.67) had an increased risk of suboptimal adherence to ICS instead of optimal adherence. An increased risk of underuse was predicted by a higher FEV₁/VC (OR =1.05). Predictors for the risk of overuse were a lower FEV₁ (OR =0.49), higher scores on Clinical COPD Questionnaire-question 3 (anxiety for dyspnea) (OR =1.26), and current smoking (OR =1.73). Regarding tiotropium, predictors for suboptimal use were a higher FEV₁/VC (OR =1.03) and the inability to perform usual activities as asked by the EuroQol-5D questionnaire (OR =3.09). A higher FEV₁/VC also was a predictor for the risk of underuse compared to optimal adherence (OR =1.03). The risk of overuse increased again with higher scores on Clinical COPD Questionnaire-question 3 (OR =1.46).

Conclusion

Several disease-specific and quality of life factors are related to ICS and tiotropium adherence, but a clear profile of a nonadherent patient cannot yet be outlined. Overusers of ICS and tiotropium experience more anxiety.

Abdominal aortic aneurysms and thoracic aortic atheromas

BACKGROUND

Abdominal aortic aneurysm (AAA) is associated with atherosclerosis elsewhere. Thoracic aortic atheromas (ATHs) seen on transesophageal echocardiography (TEE) are an important cause of stroke and peripheral embolization. The purposes of this study were to determine whether an association exists between AAA and ATHs and to assess the importance of screening patients with ATHs for AAA.

METHODS

For the retrospective analysis, 109 patients with AAA and 109 matched controls were compared for the prevalence of ATHs on TEE and for historical variables. For the prospective analysis, screening for AAA on ultrasonography was performed in 364 patients at the time of TEE.

RESULTS

Results of the retrospective analysis showed that ATHs were present in 52% of patients with AAA and in 25% of controls (odds ratio [OR] = 3.3; P =.00003). There was a significantly higher prevalence of hypertension, myocardial infarction, heart failure, smoking, and carotid or peripheral arterial disease in patients with AAA. However, only ATHs were independently associated with AAA on multivariate analysis (P =.001). Results of the prospective analysis showed that screening at the time of TEE in 364 patients revealed AAA in 13.9% of those with ATHs and in 1.4% of those without ATHs (P <.0001; OR = 11.4).

CONCLUSIONS

(1) There is a strong, highly significant association between abdominal aneurysm and thoracic atheromas. (2) Patients with AAA may be at high risk for stroke because of the concomitance of thoracic aortic atheromas. (3) The high prevalence of abdominal aneurysm in patients with thoracic atheromas suggests that screening for abdominal aneurysm should be carried out in all patients with thoracic atheromas identified by TEE.

Unusual echocardiographic views of bicuspid and tricuspid pulmonic valves

The pulmonary artery is typically seen on transthoracic echocardiography in its longitudinal axis. Therefore, short axis views of the pulmonic valve leaflets are not generally obtained, and the distinction between tricuspid and bicuspid pulmonic valves is difficult or impossible. Bicuspid pulmonic valve is one cause of pulmonic stenosis, which is especially common in tetralogy of Fallot. Presented here are 2 patients in whom the orientation of the pulmonary artery was unusual, and the pulmonic valve was seen en face. The first patient had tetralogy of Fallot and a bicuspid pulmonic valve. The severe obstruction to right ventricular outflow was infundibular. The second patient had severe stenosis of a tricuspid pulmonic valve, which was treated with balloon valvuloplasty. These unusual views of the pulmonic valve leaflets were obtained because of anterior displacement of the pulmonary artery, and precise anatomic delineation of the problem in each case was possible with transthoracic echocardiography.

Minimally invasive aortic valve replacement: echocardiographic and clinical results

BACKGROUND

Port access has been described for mitral and bypass surgery. The purpose of this study was to review the clinical and echocardiographic outcomes of aortic valve replacement by use of port access.

METHODS

Between 1996 and 1999, 153 port-access aortic valve replacements were performed at our institution. The mean age was 63 years (range 16-91 years); 58% were male. The New York Heart Association mean class was III; 18% were in class IV. Thirteen percent had diabetes, 42% hypertension, 7% prior transient ischemic episode or stroke, 7% lung disease, 3% renal failure, and 13% previous surgery. Echocardiograms were obtained after valve replacement in 125 patients (96 intraoperative transesophageal and 97 transthoracic echoes).

RESULTS

Median length of stay was 8 days. There were no intraoperative deaths; 10 patients (6.5%) died in the postoperative period. Stroke occurred in 4 (2.6%), sepsis in 5 (3.3%), renal failure in 5 (3.3%), pneumonia in 3 (2%), and wound infection in 1 (0.7%). Tissue prosthesis was present in 83 and a mechanical prosthesis in 42. No or trace regurgitation was seen on 94 of 96 (98%) postbypass intraoperative echocardiograms and mild on 2. On follow-up echocardiograms, moderate regurgitation was seen in 4 of 97 (4.1%), mild-to-moderate in 2 (2.1%), mild in 18 (18.6%), and no or trace in 71 (73.2%). Of those who had aortic regurgitation on intraoperative or follow-up echocardiograms, it was paravalvular in 8.

CONCLUSIONS

Minimally invasive aortic valve replacement with a port-access approach is feasible, even in high-risk patients. Small incisions, a low infection rate, and a short length of stay are attainable. However, the complications associated with traditional aortic valve replacement still occur. Echocardiography is valuable both for intraoperative monitoring and follow-up of this new procedure.

Acquired aorta-pulmonary artery fistula: diagnosis by multiple imaging modalities

Acquired communication between the aorta and the pulmonary artery is a rare phenomenon. We describe two patients with a thoracic aortic aneurysm in whom the diagnosis of a communication with the pulmonary artery was first made on transthoracic echocardiography and then more completely elucidated by means of multiple imaging modalities: transesophageal echocardiography, epiaortic ultrasound, computed tomography, and magnetic resonance imaging. Representative images from these complementary studies are presented. A successful repair of the fistula was subsequently accomplished in both patients.

Thromboembolus from a ligated left atrial appendage

The left atrial appendage of patients with mitral valve disease is commonly a source of thromboembolus and is often ligated during mitral valve surgery to diminish this risk. However, ligation is often incomplete. We describe a patient with a stroke whose only source of embolus was an incompletely ligated left atrial appendage. Attempts to exclude the left atrial appendage from the arterial circulation by suture ligation may not decrease the risk of thromboemboli and instead may increase such risk.

Hyperglycemic athymic nude mice: factors affecting in vitro insulin secretion

The strain of athymic nude male mice (ANM) developed at the University of Southern California (USC) exhibits spontaneous hyperglycemia and relative hypoinsulinemia in vivo. To investigate factors that influence insulin secretion in this animal model of non-insulin-dependent diabetes mellitus, we utilized the isolated perfused mouse pancreas of the ANM-USC and control BALB/c mice. We compared in vitro glucose-induced insulin secretion in ANM-USC and control mice, inhibition of secretion by somatostatin, and variability of insulin secretion over the two-year period it took to complete these experiments. Glucose-induced insulin secretion from the isolated pancreas was biphasic in both ANM-USC and controls. Insulin secretion was quantitatively equal to or greater than control mice, depending on the phase of secretion analyzed and the source of the control mice. In contrast to pancreases of control mice, insulin secretion from ANM-USC pancreases was relatively resistant to inhibition of insulin secretion by somatostatin. Variability in insulin secretion over the two years in which these experiments were performed was greater from pancreases of control than that observed from pancreases of the ANM-USC. The hyperglycemic ANM-USC mouse does not demonstrate diminished insulin secretion in vitro yet is relatively hypoinsulinemic in vivo. Thus circulating factors other than somatostatin might contribute to the insulinopenic stage in this animal model.