Lung cancer better predicted with CT tool
Model held up for toughest cases for doctors
Predicting whether lung lesions are cancerous can be more accurate using new software, Canadian researchers have shown.
Lung cancer has a poor survival rate because by the time symptoms emerge, it's usually too late to treat. Doctors have been testing CT scans to help diagnose the disease.
In Wednesday's online New England Journal of Medicine, Canadian, U.S. and Argentine researchers used CT data from a total of 12,029 lung nodules from 2,961 current and former smokers in Canada to test the accuracy of a prediction tool.
"We already know that CT screening saves lives," co-principal investigator Dr. Stephen Lam, chair of BC's Provincial Lung Tumour Group at the BC Cancer Agency, said in a release.
"Now, we have evidence that our model and risk calculator can accurately predict which abnormalities that show up on a first CT require further follow up, such as a repeat CT scan, a biopsy, or surgery, and which ones do not."
Current guidelines assess nodules based on size. But investigators found that in 20 per cent of participants, the largest lung nodule wasn't the one that was malignant on follow-up.
The new tool considers other factors besides size, including:
- Older age.
- Female sex.
- Family history of lung cancer.
- Location of the nodule in the upper lobe of the lung.
Lam said the prediction tool held up even for the toughest cases, such as nodules that are one centimetre or smaller.
Each nodule in the study was followed for at least two years.
Participants were at high risk and were from either part of a pan-Canadian study or the BC Cancer Agency. The study did not consider the model for those at low-risk.
The research was funded by the research arm of the Terry Fox Foundation, the Canadian Partnership Against Cancer and the National Cancer Institute.
Lung cancer kills over 20,000 Canadians each year.