Factors linked to lung cancer similar among adults with incidentally detected lung nodules

Key takeaways:

• Several characteristics known to be related to lung cancer were observed among those with incidentally detected nodules.
• Prediction models with these characteristics yielded “good discrimination.”

Age, sex, BMI, smoking history, nodule size and nodule location — notable factors linked to lung cancer — were also related to this diagnosis in adults with incidentally detected lung nodules, according to study results published in CHEST.

“Findings from this study may influence how clinicians care for and how practice guideline committees consider individuals with an incidentally detected lung nodule and a prior history of a nonlung malignancy,” Farhood Farjah, MD, MPH, FACS, associate professor of surgery at the University of Washington and associate medical director of the Surgical Outcomes Research Center at University of Washington Medicine, and colleagues wrote.

In a retrospective cohort study, Farjah and colleagues analyzed 7,240 adults (median age, 67 years; 56% women; 56% ever smokers) with incidentally detected lung nodules found on chest CT scans between 2005 and 2015 to see if patient and nodule characteristics in this sample are related to characteristics known to be linked to lung cancer through binomial regression.

Researchers additionally tested whether these characteristics could be incorporated into lung cancer prediction models, which they validated through bootstrap optimism correction.

Of the total cohort, 31% had prior nonlung malignancy, and this included patients with multiple nodules (57%) and an upper lobe nodule (40%). Lung cancer was observed in 265 patients (3.7%; 95% CI, 3.2%-4.1%) in the 2 years following discovery of a nodule.

Assessing 2-year probability for lung cancer, researchers found three factors linked to a greater chance for this diagnosis using multivariable analysis: sex (women vs. men, RR = 1.51; 95% CI, 1.21-1.88), smoking history (ever smokers vs. never smokers, RR = 3.22; 95% CI, 2.34-4.43), and nodule location (upper lobe vs. non-upper lobe, RR = 1.67; 95% CI, 1.34-2.09). Probability for lung cancer was also observed when evaluating age (P = .003), BMI (P = .02) and nodule size (P < .001).

Notable factors not related to this diagnosis included prior malignancy, nodule number, nodule laterality, health system and year of nodule diagnosis, according to researchers.

Using the factors found in multivariate analysis, researchers created and validated the performance of three different prediction models. The first model included all observed variables, regardless of significance, whereas the second model only included significant factors. In the third model, researchers only included nodule size.

In both the first and second models, area under the curve (AUC) was 0.75 (95% CI, 0.72-0.8), signaling “good discrimination,” according to researchers. However, the third model had a lower AUC of 0.7 (95% CI, 0.65-0.75).

Of the three models, deviation between predicted and observed probabilities of lung cancer was the lowest in the model including all variables and highest in the model that only included nodule size, according to researchers.

“Prediction models developed using readily available data from a population-based cohort are promising, but require external validation and comparison with existing models across various practice settings and patient populations before use in clinical practice,” Farjah and colleagues wrote.