Validation of the Diagnostic Accuracy of the Electronic Nose in the Detection of Thyroid Cancer
Validation of the Diagnostic Accuracy of the Electronic Nose in the Detection of Thyroid Cancer
Patients with a suspected thyroid nodule face an invasive and patient unfriendly diagnostic work-up to determine the risk of malignancy. Typically, patients undergo ultrasound of the thyroid gland followed by fine-needle aspiration cytology (FNAC). FNAC has been considered as a gold standard diagnostic procedure in suspected thyroid nodules. Unfortunately, both the negative- and positive predictive value of FNAC is poor, often resulting in the need for a diagnostic hemithyroidectomy for definite diagnosis . Approximately 40-94% of the suspected thyroid nodules appear to be benign after resection and thus exposes patients to unnecessary surgery with unnecessary risks. Therefore, a quick, non-invasive assessment of the risk of malignancy of thyroid nodules is of paramount importance. Such a novel test could fasten the diagnostic process for patients with malignancies and reduce the amount of 'unnecessary' surgeries for benign conditions.
A promising development in cancer detection is based on volatile organic compounds (VOCs), gaseous degradation products of biochemical processes detectable in exhaled breath. During pathophysiological processes related to tumor growth, alterations in cell metabolism lead to a shift in the production of VOCs. The VOCs' patterns can be detected by the Aeonose™ through their reaction with the metal-oxide sensors in this device. A pilot study conducted at the Maastricht University Medical Center demonstrated that, by creating an artificial neural network (ANN) from the VOC patterns of numerous patients and their specific histopathological diagnosis, the Aeonose™ has a high diagnostic accuracy to discriminate benign from malignant thyroid nodules.
The purpose of this study is to validate the accuracy of the Aeonose™, to prevent unnecessary surgery and to investigate the use of the Aeonose™ as a surveillance tool in the postoperative follow-up of differentiated thyroid cancer.
We hypothesize that the high negative predictive value of the pilot study will be confirmed in the validation study and expect that implementation of the Aeonose™ in clinical practice will subsequently reduce the number of unnecessary surgeries below 10% for patients with Bethesda ≥ III nodules and may provide an important role in non-invasive detection of recurrent disease.
Thyroid cancer is the most common endocrine cancer worldwide. In most developed countries, the incidence of thyroid cancer has rapidly increased in the last three decades In the Netherlands, 700 patients are diagnosed with thyroid cancer each year. The most common histological forms are the well-differentiated papillary and follicular cancer, comprising of 80-85% of all thyroid cancers
Patients often present themselves with a thyroid nodule, which can be benign or malignant. Current diagnostic work-up includes ultrasound of the neck, resulting in the Thyroid Imaging Reporting And Data System (TI-RADS). This risk-stratification system for thyroid nodules is based on ultrasound features. When indicated, fine-needle aspiration cytology (FNAC) is performed, resulting in the Bethesda classification to report the risk of malignancy. The FNAC has been considered as gold standard diagnostic procedure in suspected thyroid nodules . Unfortunately, both negative- and positive predictive value of FNAC is poor, often resulting in the need for a diagnostic hemithyroidectomy for definite diagnosis. 40-94% of the suspected thyroid nodules appear to be benign after resection. After surgical removal, patients with differentiated thyroid cancer sometimes receive adjuvant treatment with radioactive-iodine. Thereafter, patients face a long-term follow-up since recurrence of the disease may occur years after removal of the primary tumor. Based on risk stratification, serum thyroglobulin (Tg) levels are determined once or twice per year with or without additional imaging (e.g. thyroid ultrasound or total body scintigraphy). So, both diagnosis and follow-up of thyroid cancer are associated with invasive and unpleasant procedures for the patient.
After surgical removal, patients with differentiated thyroid cancer sometimes receive adjuvant treatment with radioactive-iodine. Thereafter, patients will be subjected to a long-term follow- up since recurrence of the disease may occur years after removal of the primary tumor. Based on risk stratification, serum thyroglobulin (Tg) levels are determined once or twice per year with or without additional imaging (e.g. thyroid ultrasound or total body scintigraphy). So, both diagnosis and follow-up of thyroid cancer are associated with invasive and unpleasant procedures for the patient.
A promising development in cancer detection is based on volatile organic compounds (VOCs), gaseous degradation products of biochemical processes detectable in exhaled breath. During pathophysiological processes related to tumor growth, alterations in cell metabolism lead to a shift in the production of VOCs. The VOCs' patterns can be detected by the Aeonose™ through their reaction to the metal-oxide sensors in this device. A pilot study conducted at the Maastricht University Medical Center+ demonstrated that, by creating an artificial neural network (ANN) from the VOC patterns of numerous patients and their specific histopathological diagnosis, the Aeonose™ has a high diagnostic accuracy to discriminate benign from malignant thyroid nodules.
The purpose of this study is to validate the accuracy of the Aeonose™, to prevent unnecessary surgery and to investigate the use of the Aeonose™ as a surveillance tool in the postoperative follow-up of differentiated thyroid cancer.
Aim
This study consists of three aims:
We plan to conduct an international prospective, observational multicenter study in twelve hospitals in the Netherlands, Belgium and Germany to validate the diagnostic performance of the Aeonose™ and to investigate its efficacy in the follow-up of differentiated thyroid cancer. First, a database of breathing patterns will be built to develop an ANN. An external validation study will be conducted using exhaled breath patterns of a total of 500 patients to address the first aim. Data from the validation study together with clinical parameters will be used to determine the necessity for a diagnostic hemithyroidectomy. The number of unnecessary surgeries will be evaluated. To address the third aim, a pilot study will be conducted. Patients with histologically proven differentiated thyroid cancer are asked to breathe in the Aeonose™ parallel to every regular follow-up moment. The Aeonose™ will be trained by the input of data of patients developing a recurrence or not to develop an ANN that can be used as a surveillance tool to detect recurrence non-invasively.
Before starting the validation study, the artificial neural network (ANN) that has been built in the pilot study will be expanded in the training study. The more patients included with a variety in breathing patterns (as a result of e.g. gender, age, smoking-status, food intake, medication use), the more stable and robust the model becomes. For this training study, a total of 200 patients with histopathological proven differentiated thyroid cancer is needed. Assuming that the prevalence of thyroid cancer among suspected thyroid nodules is 20%, a total of 1000 patients would be required for the training study. Data of patients who participated in the pilot study will be used in the training study. The training study will be carried out in all 12 hospitals.
Validation study Directly after finishing the training study, the validation study will be started in 'new' patients that have not participated in the training study.
Every patient with a suspected thyroid nodule, indicated for additional diagnostic procedures including an ultrasound of the thyroid gland (resulting in the TIRADS classification) followed by a fine-needle aspiration for cytology (resulting in the Bethesda classification), will be asked to participate. We will provide information about the study and obtain a signed informed consent. We expect that in particular, the FNAC might influence the patterns of the VOCs in the exhaled breath, measured by the Aeonose™. For this reason, we will include patients in this study before FNAC will take place.
Patients in the validation study will be asked to participate only once by breathing in the Aeonose™ for five minutes, followed by completing a short questionnaire of known factors that can influence the breathing patterns.
A minimum of 500 patients will be necessary to validate the sensitivity of the pilot study.
This study has no specific benefits for the participating patients. The risks of using the Aeonose are very small. Possible side effects during measurements are dizziness and nausea, usually due to hyperventilation. Other side effects are hypo- or hyper salivation during measurements. Blood will be collected during regular pre-operative blood collection, resulting in no extra skin puncture. Furthermore two quick to answer questionnaires as well as a short case report form (CRF) will be filled in by the participating patients. Using the Aeonose as a non-invasive, rapid and inexpensive diagnostic tool could be a major benefit for patients with thyroid nodules due to the faster and less invasive diagnostic process. Benefits for patients with benign thyroid diseases include the possibility to resign from unnecessary invasive treatments such as (diagnostic) surgery.
We hypothesize that the high negative predictive value of the pilot study will be confirmed in the validation study and expect that implementation of the Aeonose™ in clinical practice will subsequently reduce the number of unnecessary surgeries below 10% for patients with Bethesda ≥ III nodules. For the pilot study on the follow-up, we hypothesize that the Aeonose™ can play an important role in non-invasive detection of recurrent disease. Within several years, analysis of exhaled breath using the Aeonose™ will play an important role in the detection of thyroid cancer and/or its recurrence and will, therefore, have a position in regular clinical decision making
Inclusion Criteria:
Exclusion Criteria:
n.bouvy@mumc.nl+31 43 3876543
z.al-difaie@maastrichtuniversity.nl+31640075718