Prostate Cancer
Conditions
Keywords
Artificial Intelligence, Radiotherapy reactions, clinical decisions, predict outcomes, Patient individualised response, Machine Learning, AI, Personalized Radiotherapy, Radiotherapy Adaptation
Brief summary
Nearly half of all cancer patients receive radiotherapy as part of their treatment and although it is effective at destroying cancerous lesions deep within the body, this comes at the cost of damaging healthy, or normal, tissues. With 50% of cancer patients surviving for 10 years or more, these patients can be left with life-changing side effects from their radiotherapy. It is clear that more must be done to limit damage to normal healthy tissue without compromising annihilation of the tumour and curing patients. The key to this is personalising an individual's radiotherapy treatment, in other words rather than assuming that all tumours respond similarly to radiotherapy, the treatment is optimised for an individual. To date, approaches to do this have been restricted to small numbers of carefully selected patients, are inordinately expensive, and not suitable for rolling out into everyday practice across the NHS. There is however another way, namely using Artificial Intelligence (AI) combined with an individual's healthcare record. By linking together large numbers of healthcare records at a national level, combined with the power of AI, the PROSECCA project will transform radiotherapy and cancer care.
Detailed description
As an example of how this technology is beginning to emerge, proof of concept data from the team has shown that using AI it is possible to identify patients at increased risk of damage from radiotherapy long before they receive any radiation as part of their treatment. However, to move these AI-based approaches from the research domain into the clinic requires significant effort, which is the aim of PROSECCA study. The project will use AI to analyse healthcare records from up to 15,000 prostate cancer patients who underwent radiotherapy in Scotland. Through linkage of data obtained specifically for radiotherapy and data held within each patient's unique healthcare history it will be possible to establish new relationships between a patient's medical history and how well a patient responds to radiotherapy. By establishing what factors or information in a patient's complex healthcare record indicate that an individual may have a poor response to treatment, or an increased risk of side effects from radiation, it will be possible to identify these patients earlier than is currently possible and adapt treatment accordingly. Furthermore, identifying these important factors would improve radiotherapy treatment for prostate cancer patients in the future.
Interventions
None listed
Sponsors
University of Edinburgh
NHS Greater Glasgow and Clyde
NHS Grampian
NHS Highland
NHS Tayside
University of Glasgow
University of Lyon
NHS Lothian
NHS Research Scotland
Public Health Scotland
Prostate Cancer UK
Movember Foundation
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
MALE
Healthy volunteers
No
Inclusion criteria
* External beam radiotherapy delivered by a linear accelerator * Prostate Specific Antigen (PSA) recorded at regular intervals after radiotherapy * Minimum of 10 year survival post-radiotherapy * Diagnostic Computerised Tomography (CT) acquired * Radiotherapy planning CT acquired * Radiotherapy treatment planning data available * Corresponding healthcare data available to infer toxicity events (ref previous work by Lemanska et al)
Exclusion criteria
* Incomplete course of radiotherapy * No PSA data * No follow-up corresponding healthcare data available * No imaging data available
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| PSA relapse free survival | Plus/minus 10 years from date of radiotherapy treatment | PSA relapse free survival will be measured from the date of treatment. This information will be obtained from historical patient records. The expected number of events in this category was calculated using the following approach. 5 year endpoint, 10% event rate based on Conventional or hypofractionated high dose intensity modulated radiotherapy for prostate cancer (CHHIP) trial Data up to 2015 resulting in sample size of 11,250 (75% of 15,000) Total Events = 1125 |
| Overall Survival | Plus/minus 10 years from date of radiotherapy treatment | Overall survival will be measured from the date of treatment. This information will be obtained from historical patient records. The expected number of events in this category was calculated using the following approach. 10 year endpoint, 29% event rate based on RT01 radiotherapy trial Data up to 2010 resulting in sample size of 7,500 (50% of 15,000) Total Events = 2175 |
| Radiotherapy Toxicity | Plus/minus 10 years from date of radiotherapy treatment | Radiotherapy Toxicity was estimated based on the cohort toxicity reported in the CHHIP trial. The same estimates were applied to this cohort with the expected number of events as follows. Note that a higher attrition rate is expected for this endpoint, 60% of 15,000=10,000 5 year endpoint, 12% event rate for Grade 2+ toxicity based on CHHIP trial Total Events = 900 |
Countries
United Kingdom
Contacts
CONTACTBill Nailon
CONTACTDuncan McLaren
PRINCIPAL_INVESTIGATORBill Nailon
University of Edinburgh
PRINCIPAL_INVESTIGATORDuncan McLaren
University of Edinburgh
Outcome results
None listed