Do Canadians Have a Better Lung Cancer Screening Approach?

Kristin Jenkins

November 01, 2017

A highly predictive lung cancer risk model can identify candidates for low-dose computed tomography (LDCT) screening who are at high-risk for early, potentially curable disease, Canadian researchers say.

The incidence of cancers detected and the proportion of early-stage cancers in the screened population was higher than observed in previous studies, including the National Lung Screening Trial (NLST) in the United States, say Stephen Lam, MD, from the British Columbia Cancer Agency in Vancouver, Canada, and colleagues, in report published online October 18 in Lancet Oncology.

"This approach should be considered for adoption in lung cancer screening programmes," the authors say, suggesting it is better than the current approaches.

Results from the prospective, single-arm Pan-Canadian Early Detection of Lung Cancer (PanCan) study show that after a median follow-up of 5.5 years and 3 LDCT scans among 2537 enrolled participants, 172 lung cancers were diagnosed in 164 individuals.

Overall, the incidence of lung cancer was 6.5% among the PanCan study participants, which is significantly higher than the 4.0% (P < .0001) achieved in the NLST after three LDCT scans and 6.5 years of follow-up, the study authors point out. Notably, compared with the NLST, the PanCan model also detected a higher proportion of early-stage (I or II) cancers, which is desirable: 77% versus 57%.

The PanCan approach is also less costly.

The cost of screening per year of life saved was CAN$20,724 with PanCan versus US$81,000 with NLST criteria, which is based on classic risk factors such as age and smoking history.

As previously reported by Medscape Medical News, the NLST criteria demonstrated a 20% relative reduction in lung cancer mortality with annual LDCT compared with chest radiography. However, these criteria don't quantify risk, the Canadian authors of the new study say, noting that many high-risk individuals who are later diagnosed with lung cancer don't meet the criteria for LDCT screening.

Lung cancer screening rates among current and former heavy smokers in the United States remain extremely low, despite US Preventive Services Task Force recommendations for annual screening. In 2015, only 262,700 received LDCT screening out of an estimated 6.8 million eligible smokers.

The PanCan model, which has an estimated sensitivity of 92.7%, is an early prototype of the "extensively validated" PLCOm2012 lung cancer risk prediction model developed using data from the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial.

Earlier this year, a retrospective validation study of nine risk-prediction models showed that the PLCOm2012 model has better discrimination, calibration, and clinical usefulness than other prediction models, the investigators note.

Currently, ongoing international prospective clinical trials are comparing the effectiveness and cost of different selection methods for low-dose CT screening, the authors add.

When asked to comment, Lecia V. Sequist, MD, MPH, director of the Center for Innovation in Early Cancer Detection at Massachusetts General Hospital Cancer Center, Boston, said she found the PanCan study "quite helpful" — if the ultimate goal was to make smoking-associated lung cancer screening "as cost-effective as possible" by shrinking the screening population to those at the very highest risk.

However, the study can't definitively demonstrate the harms and benefits of the PanCan model relative to the standard NLST method in terms of finding cancers, survival, or cost, observed Dr Sequist, who is also an associate professor of medicine at Harvard Medical School.

Study Details

The PanCan study enrolled 2537 eligible ever-smokers between September 24, 2008, and December 17, 2010, at eight Canadian centers. Participants answered a structured epidemiological questionnaire before undergoing LDCT and spirometry. Some 1300 participants also chose to undergo autofluorescence bronchoscopy. Serum biomarker studies were conducted at baseline, and at 1 and 4 years postbaseline.

172 lung cancers were diagnosed, 161 participants had a single lung cancer, and there were 137 surgical resections. Of these, 121 (88%) were early-stage lung cancer and 16 (12%) were benign lesions.

Risk thresholds for tailored screening programs need to be determined, Dr Lam told Medscape Medical News. "Can we improve the sensitivity of tailored screening by inclusion of other factors such as air pollution, genetic susceptibility or biomarkers such as a blood test or breath test?"

The PanCan predictive risk model is being used in pilot screening programs in Ontario and Vancouver, and is publicly available, said Dr Lam, who is professor of medicine at the University of British Columbia in Vancouver.

In an accompanying editorial, Harry J. de Koning, MD, PhD, an epidemiologist and professor of public health and screening evaluation at Erasmus Medical Center in Rotterdam, the Netherlands, agreed that risk cutoffs need to be determined and applied to predictive tools. More than 50% of willing individuals in the PanCan study weren't considered to have a high enough risk to be screened, he pointed out.

The PanCan results substantiate data from both the NLST and the Dutch-Belgian Lung Cancer Screening trial (NELSON) about the negative predictive value for individuals with no nodules at baseline, said Dr Koning, who is principal investigator of the NELSON trial.

In PanCan, for individuals with no nodules at baseline their risk of lung cancer in the next 2 years was only 1 in 25, compared with individuals with nodules ≥ 1 mm (.0018 vs .046). This is line with data from NLST and NELSON. Thus, in the United Kingdom, small nodules at baseline are left untouched and participants are invited back for another CT scan one or two years later, said the editorialist.

More information about how best to reduce the number of referrals and interventions for benign lesions will be forthcoming from the NELSON trial, Dr Koning predicted. "The final results of the NELSON trial, which had a cumulative incidence of lung cancer of 3.2% after four CT scans, 69% of lung cancers detected in stage I, a referral rate of 2.1%, and which is reaching its 10-year follow-up, are eagerly awaited."

Harvard's Dr Sequist commented that to improve lung cancer screening over the long-term, test specificity should be increased so that screening can be offered to the entire general population, not just those at highest risk.

"It is important to consider that the number of patients with non-smoking associated cancers is growing each year and we aren't as clear as to what the risk factors may be for this disease," Dr Sequist told Medscape Medical News.

"Ideally," she added, this would also "lower the cost of follow-up testing."

The study was funded by the Terry Fox Research Institute and the Canadian Partnership Against Cancer. The study authors have disclosed no relevant financial relationships. Dr Koning is principal investigator of the NELSON trial. Dr Sequist has relationships with AstraZeneca, Pfizer, BMS, Ariad, and Genentech.

Lancet Oncol. Published online October 18, 2017. Abstract, Editorial

For more from Medscape Oncology, follow us on Twitter: @MedscapeOnc


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.