Optimising Cancer Therapy And Identifying Causes of Pneumonitis USing Artificial Intelligence (COVID-19) (OCTAPUS-AI)

• ClinicalTrials.gov Identifier: NCT04721444
• Recruitment Status: Recruiting
• First Posted: January 22, 2021
• Last Update Posted: June 9, 2021
• Sponsor: Royal Marsden NHS Foundation Trust
• Information provided by (Responsible Party): Royal Marsden NHS Foundation Trust

Study Description

Brief Summary:

Distinguishing changes on patients that have received thoracic radiotherapy and patients that are currently receiving or have recently received IO and presenting lung changes which will be identified using AI.

Condition or disease	Intervention/treatment
Lung Cancer; Covid19; Pneumonitis	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Diagnostic Test: Machine Learning Classification of recurrence and non-recurrence

Study Design

• Study Type: Observational
• Estimated Enrollment: 1500 participants
• Observational Model: Cohort
• Time Perspective: Retrospective
• Official Title: Optimising Cancer Therapy And Identifying Causes of Pneumonitis USing Artificial Intelligence
• Actual Study Start Date: January 27, 2021
• Estimated Primary Completion Date: August 1, 2022
• Estimated Study Completion Date: January 1, 2023

Groups and Cohorts

Group/Cohort	Intervention/treatment
Arm A - Cohort A1; Training set: Pneumonitis in the context of IO therapy and negative for infectious pneumonia (including COVID-19)	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Arms A & B: Radiomics and deep-learning approaches will be used on patient's imaging to develop a feature vector that can distinguish parenchymal lung changes, e.g. infection from drug-toxicity.
Arms A and B - Cohort B1; Training set B1: IO and RT naive and pneumonia (without COVID-19)	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Arms A & B: Radiomics and deep-learning approaches will be used on patient's imaging to develop a feature vector that can distinguish parenchymal lung changes, e.g. infection from drug-toxicity.
Arms A and B - Cohort B2; Training set B2: IO and RT naive and confirmed COVID-19 positive with pneumonia	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Arms A & B: Radiomics and deep-learning approaches will be used on patient's imaging to develop a feature vector that can distinguish parenchymal lung changes, e.g. infection from drug-toxicity.
Arm B - Cohort A2; Training set: Pneumonitis in the context of thoracic RT and negative for infectious pneumonia (including COVID-19)	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Arms A & B: Radiomics and deep-learning approaches will be used on patient's imaging to develop a feature vector that can distinguish parenchymal lung changes, e.g. infection from drug-toxicity.
Arm A - Test Cohort (Cohort C1); Test set C1: Patients on IO and with possible toxicity versus COVID-19 or other infective pneumonitis	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Arms A & B: Radiomics and deep-learning approaches will be used on patient's imaging to develop a feature vector that can distinguish parenchymal lung changes, e.g. infection from drug-toxicity.
Arm B - Test Cohort (Cohort C2); Test set C2: Patients with pneumonitis in context of thoracic RT with possible toxicity versus COVID-19 or other infective pneumonitis.	Diagnostic Test: Machine Learning Classification of parenchymal lung change cause; Arms A & B: Radiomics and deep-learning approaches will be used on patient's imaging to develop a feature vector that can distinguish parenchymal lung changes, e.g. infection from drug-toxicity.
Arm C; Patients with radiotherapy planning CT scans and post-treatment surveillance CT scans at 3, 6 and 12-months post treatment	Diagnostic Test: Machine Learning Classification of recurrence and non-recurrence; Arm C: Radiomics and deep-learning approaches will be used on patient's imaging to develop a risk-signature for recurrence of malignancy following radical treatment

Outcome Measures

Primary Outcome Measures :

1. Development of a Machine Learning model to distinguish parenchymal lung changes [ Time Frame: 3 years ]
   The development and validation of an ML/radiomic classifier to distinguish between Infective/COVID-19 pneumonia and cancer therapy induced lung changes
2. Development of a Machine Learning model to predict recurrence risk after radical radiotherapy for non-small cell lung cancer [ Time Frame: 3 years ]
   to develop a prognostic AI/radiomic signature for NSCLC recurrence after radical RT (Conventional fractionated RT +/- chemotherapy or stereotactic body RT (SBRT)) to stratify appropriate surveillance and onward care, thus minimising unnecessary hospital visits and resource use.

Eligibility Criteria

• Ages Eligible for Study: 18 Years and older (Adult, Older Adult)
• Sexes Eligible for Study: All
• Sampling Method: Non-Probability Sample

Study Population

Arms A and B: Adult patients with pneumonitis that have previously been treated with thoracic radiotherapy or immunotherapy Arm C: Adult patients that have received radical radiotherapy for NSCLC

Criteria

Arms A & B:

Inclusion Criteria:

Cohort A1 (from Arm A) - Immunotherapy (IO) pneumonitis cases: patients currently on or having received ICI IO in the last 3 months of presentation with:

• New radiological lung changes on CT/CXR (confirmed on report) of a severity and distribution consistent with IO pneumonitis. These changes should be of severity and distribution that are not incompatible with viral or lower respiratory tract infection.

AND Must not have had RT involving the thorax (unless this was breast/chest wall RT more than 5 years ago, which is permissible) AND

• Where there is documented clinical concern for infection, have undergone one or more laboratory investigations for viral or lower respiratory tract infection including, but not limited to Nasopharyngeal aspirate or swab for respiratory virus by PCR; Sputum sample or bronchial washings MCS with no organism(s) consistent with lower respiratory tract infection, cytology or beta-glucan/galactomannan for PCP or fungal infection; broncho-alveolar lavage for markers of infection such as MCS, PCR, fungal culture, beta-glucan/galactomannan for PCP or evidence of lower respiratory tract infection (including invasive fungal infection) by cytology, none of which were considered positive for infection by the clinical team.
• Where empirical antibiotics were prescribed, patients must either have had a negative BAL infection screen or may be included at the discretion of the local site PI and local radiologist with lung interest or two members of the trial management group, one of whom must be a radiologist with lung interest or respiratory physician or oncologist with suitable experience of thoracic CT imaging, after after review of the case-notes.
• Prophylactic co-trimoxazole prescribed in the context of high-dose steroid therapy is permitted.

Cohort A2 (from Arm B) - Radiotherapy (RT) pneumonits cases: Patients that have completed a course of RT involving the thorax (e.g. lung, breast, oesophageal RT) in the last 12 months prior to presentation, that have not received immunotherapy, with:.

• New radiological lung changes on CT/CXR (confirmed on report) of a severity and distribution consistent with radiation pneumonitis or early fibrosis (should not include established fibrosis). These changes should be of severity and distribution that are not incompatible with viral or lower respiratory tract infection.

AND

• Where there is documented clinical concern for infection, have undergone one or more laboratory investigations for viral or lower respiratory tract infection including, but not limited to Nasopharyngeal aspirate or swab for respiratory virus by PCR; Sputum sample or bronchial washings MCS with no organism(s) consistent with lower respiratory tract infection, cytology or beta-glucan/galactomannan for PCP or fungal infection; broncho-alveolar lavage (BAL) for markers of infection such as MCS, PCR, fungal culture, beta-glucan/galactomannan for Pneumocystis Pneumonia (PCP) or evidence of lower respiratory tract infection (including invasive fungal infection) by cytology, none of which were considered positive for infection by the clinical team. Where empirical antibiotics were prescribed, patients must either have had a negative BAL infection screen or may be included at the discretion of the local site PI and local radiologist with lung interest or two members of the trial management group, one of whom must be a radiologist with lung interest or respiratory physician or oncologist with suitable experience of thoracic CT imaging, after review of the case-notes.
• Prophylactic co-trimoxazole prescribed in the context of high-dose steroid therapy is permitted.

B1 (Utilised in Arms A & B) Non-COVID-19 infective cases:

• New radiological lung changes on CT/CXR (confirmed on report) of a severity and distribution consistent with lower respiratory tract infection but compatible with the grade and nature of pneumonitis seen with IO or RT
• AND
• Laboratory findings that fulfil one or more of the following criteria of infection: Nasopharyngeal aspirate or swab positive for a respiratory virus by PCR; Sputum sample or bronchial washings positive MCS for an organism(s) consistent with lower respiratory tract infection, cytology or beta-glucan/galactomannan positive for PCP or fungal infection, positive urine legionella/pneumococcal antigen screen, positive serology for mycoplasma pneumonia; broncho-alveolar lavage for markers of infection (MCS, PCR, fungal culture, beta-glucan/galactomannan for PCP or other evidence of lower respiratory tract infection (including invasive fungal infection) by cytology. Where no such laboratory findings were positive but the patient improved with anti-microbial therapy, such cases may be included at the discretion of the local site PI and local radiologist with lung interest or two members of the trial management group two members of the trial management group, one of whom must be a radiologist with a lung interest or respiratory physician or oncologist with suitable experience of thoracic CT imaging, after review of the case-notes and imaging.
• Not previously treated with immunotherapy OR
• Must not have had RT involving the thorax (unless this was breast/chest wall RT more than 5 years ago, which is permissible)
• First assessed prior to 1st January 2020 (and therefore not attributable to COVID-19)

B2 (Utilised in Arms A & B) COVID-19 cases:

• Laboratory findings that fulfil one or more of the following criteria of COVID-19 infection: positive COVID-19 PCR test and/or antigen test or other suitable assay that indicates current infection or previous exposure (including serology tests) as determined by the trial management group (TMG).

AND

• New radiological lung changes on CT/CXR (confirmed on report) of a severity and distribution consistent with COVID-19. These changes should be of severity and distribution that is not incompatible with the grade of pneumonitis seen with IO or RT
• Not previously treated with immunotherapy OR
• Must not have had RT involving the thorax (unless this was breast/chest wall RT more than 5 years ago, which is permissible)
• Assessed after 1st January 2020 (and therefore contemporaneous with COVID-19)

Exclusion Criteria:

• Patients with documented past medical history of congestive cardiac failure or other cause for interstitial lung disease

Arm C:

Inclusion Criteria:

• Adult patients (aged 18 or over) treated with radical thoracic RT (conventional fractionated RT +/- chemotherapy or SBRT) for NSCLC
• RT planning scan imaging and labelled structure set data available from participating centre
• Minimum 2 years of post-RT follow-up data including clinical or histological confirmation in the case of recurrence and whether the patient is alive as available from primary care or hospital records.
• Patients with post-treatment surveillance CT imaging (minimum of first scan post-treatment and where available +/- further scans within 2 years post-RT, e.g. at 3/6/12 months post-treatment).

Exclusion Criteria:

• Any patient that does not have a primary lung mass e.g. Tx disease
• Any patient being treated for recurrence of a previously treated lung cancer
• Any patient that did not have radical RT e.g. patients that had high dose palliative RT
• Any patient that does not have imaging that meets technical requirements within the imaging processing and analysis manual

Contacts and Locations

Contacts

Contact: Richard Lee, MBBS, PhD; 02073528171; richard.lee@rmh.nhs.uk

Contact: Sejal Jain; sejal.jain@rmh.nhs.uk

Locations

United Kingdom

Guys and St. Thomas' NHS Foundation Trust; Recruiting

London, United Kingdom

Contact: Shahreen Ahmed

Imperial College Healthcare NHS Trust; Recruiting

London, United Kingdom

Contact: Danielle Power

Royal Marsden NHS Foundation Trust; Recruiting

London, United Kingdom

Contact: Bianca Peet

Sponsors and Collaborators

Royal Marsden NHS Foundation Trust

Investigators

• Principal Investigator: Richard Lee; Royal Marsden NHS Foundation Trust

More Information

• Responsible Party: Royal Marsden NHS Foundation Trust
• ClinicalTrials.gov Identifier: NCT04721444
• Other Study ID Numbers: 5293
• First Posted: January 22, 2021
• Last Update Posted: June 9, 2021
• Last Verified: June 2021

• Studies a U.S. FDA-regulated Drug Product: No
• Studies a U.S. FDA-regulated Device Product: No

Additional relevant MeSH terms:

COVID-19; Pneumonia; Respiratory Tract Infections; Infections; Pneumonia, Viral; Virus Diseases; Coronavirus Infections; Coronaviridae Infections; Nidovirales Infections; RNA Virus Infections; Lung Diseases; Respiratory Tract Diseases