The investigators are developing a non-invasive breath test to help us detect pancreatic cancer earlier. The test detects small molecules called volatile organic compounds that are made by pancreatic cancers.

Pancreatic cancer is a rare disease but patients are often diagnosed at a late stage because their symptoms are the same as those of many common illnesses. This makes it hard for doctors to know which patients need to be tested for pancreatic cancer. If the investigators find pancreatic cancer at a late stage, it reduces the number of treatment choices for patients.

Our test could be offered to patients who are experiencing vague symptoms, which might be caused either by pancreatic cancer or a common illness. This test could help doctors to identify which of those patients may have pancreatic cancer, and ensure they get referred for specialised pancreatic cancer tests. The investigators hope that this will allow us to diagnose pancreatic cancer earlier, increasing treatment choices for patients and improving survival from pancreatic cancer.

The investigators have previously conducted a study (VAPOR1) which collected breath samples from people with and without pancreatic cancer. When the investigators analysed these samples, they found that there is a difference in the volatile organic compounds breathed out by people who have pancreatic cancer compared to those that do not. The investigators used these 'markers' to develop a breath test to diagnose pancreatic cancer. In VAPOR2, the investigators will study our breath test in a much larger group of patients who have been referred for further investigations for potential underlying pancreatic cancer to see how accurately it can pick up the small percentage of people who have pancreatic cancer.

A multicentre double-blind external validation study to validate the breath test to detect PDAC in 6079 adult participants referred from primary care with potential underlying pancreatic cancer, whereby the VOC analyst in the laboratory and the statistician and bioinformatician classifying the VOC profile into cancer/no-cancer will be blinded to the definitive diagnosis of cancer/no-cancer at the time of analysis. Additionally, the radiologist and/or pathologist reporting the reference test (CT, MRI, biopsy) will not have access to the breath test results at any point. Breath test analysis and definitive diagnoses will be held in separate datasets, which will be cleaned and locked before being combined for analysis. The lead statistician will be unblinded at the interim analysis and at the end of the study after the final database lock once all the study data has been collected and cleaned, to calculate the accuracy of the prediction models.

Inclusion Criteria:

• Adult participants ≥ 18 years old
• Referral from primary care according to the urgent suspected cancer referral guidelines for potential underlying pancreatic cancer, or referral directly to a pancreatic cancer multidisciplinary team meeting

Exclusion Criteria:

• Previous pancreatic resection
• History of another cancer (other than non-melanoma skin cancers) within three years
• Pregnant participants (pregnancy status to be confirmed verbally with the participant)
• Participants with co-morbidities preventing breath collection
• Unable or unwilling to provide informed consent

Participants above the age of 18 years with potential underlying pancreatic cancer referred from primary care according to urgent suspected cancer referral guidelines will be invited to participate in the study. In addition, participants referred directly to a pancreatic cancer multidisciplinary team meeting will also be invited to participate in the study.