NIHR | Manchester Biomedical Research Centre

PhDs in Focus: Cancer of Unknown Primary – ‘The nameless cancer’

Holly Cassell

Author: Holly Cassell View Profile

PhD student

Welcome to our PhDs in Focus blog series, where our PhD students are showcasing their pioneering research projects at the National Institute for Health and Care Research (NIHR) Manchester Biomedical Research Centre (BRC).

In this blog, Manchester BRC PhD student Holly Cassell outlines how their project aims to shed light on the identities of unknown cancers, within the Cancer Precision Medicine theme as part of work from a team at the Cancer Research UK National Biomarker Centre.


What is CUP?

Cancer of unknown primary (CUP) is a lesser-known cancer, associated with unanswered questions and poor survival. It is the 6th most common cause of cancer related death in the United Kingdom (UK), despite only forming around 2% of cancer diagnoses.

CUP is an aggressive, metastatic cancer, which means it has spread to other parts of the body. For most metastatic cancers, doctors use diagnostic tests to tell them in which tissue the cancer started (tissue-of-origin), including examining a sample of tissue from the tumour (a tumour biopsy). This information helps medical professionals to choose the best treatments. However, in CUP the usual tests cannot identify where the cancer started, making it more difficult to treat.

Diagnosis and treatment 

After clinical investigations, patients with CUP are divided into two categories: favourable or unfavourable. For favourable CUP there is a strong suspicion of the tumours tissue-of-origin (where the cancer started), so these patients can be treated as per that tumour type. This results in a higher chance of responding to treatment. Unfortunately, this is only the case for 20% of CUP diagnoses.

The remaining 80% of patients with CUP fall into the unfavourable CUP category, where no likely tissue-of-origin can be identified. There are very few treatments available to patients with unfavourable CUP and unfortunately the average survival time for these patients is only 9 months.

However, if the tissue-of-origin can be discovered for patients with CUP, more effective, tissue-specific treatments could be used to help increase survival time. To achieve this, the team at the Cancer Research UK National Biomarker Centre, which opened in Manchester in June 2024, are developing a blood test that uses DNA released by cancer cells into a patient’s blood to predict where the cancer originally came from.

CUP classification

Markers on DNA can be associated with certain types of cancer and analysing these differences can be used to discover the tissue-of-origin in patients with CUP. One such marker on DNA is methylation, where a small chemical group called a methyl group is added to DNA, which alters how the DNA is read by the cell. The patterns of methylation in the DNA vary between different tissues and tumour types.

Using a computational approach called machine learning, this tissue-specific methylation can be used to build a classifier to identify tissue-of-origin in CUP. By providing examples of DNA methylation profiles from lots of different cancer types, the machine learning classifier can learn the methylation patterns for each tumour type and then apply them to DNA samples from patients with CUP to predict the most likely tumour type.

Using blood tests to determine the tissue-of-origin could avoid using traditional tumour samples which are difficult and invasive to collect and in patients with CUP there is often not enough of the tumour biopsy to perform genetic tests on. Blood tests are quicker, less invasive and easier to collect.

CUPiD and My PhD Project

At the Cancer Research UK National Biomarker Centre, a machine learning classifier called CUPiD has been developed to identify tissue-of-origin in patients with CUP, using DNA methylation patterns from blood samples.

CUPiD was trained to identify 29 different cancer types and was tested first in 143 blood samples from patients with cancer where we know where the cancer started. CUPiD could make a classification for 87% of the samples tested. 97% of these classifications were correct. CUPiD was then tested using blood samples from 41 patients with CUP.  CUPiD made a tissue-of-origin classification for 78% of patients, with 89% of these classifications being consistent with the clinically suspected tissue-of-origin.

My PhD project aims to develop the CUPiD classifier to improve its performance for use clinically. Areas of development include:

  • Including more cancer types.
  • Testing the classifier on more blood samples from different cancer types.
  • Improving the ability to identify the tumour type by adding in extra DNA changes such as mutations.
  • Confirming how effective CUPiD is at identifying tissue-of-origin in patients with CUP by testing on more CUP blood samples.

These developments aim to improve the ability of CUPiD to identify the tumour type in patients with CUP. The CUPiD blood test will then be tested in a clinical trial to see whether it leads to improved treatment options and better survival for patients with CUP.


Read more blogs in the PhDs in Focus series to find out how we’re developing a research talent pipeline.

Follow Manchester BRC on X/Twitter and LinkedIn to keep updated with the series.