A bewildering number of potential diagnostic biomarkers have already been examined in studies aimed at better assessing the malignancy risk of such focal shadowing: Protein-based biomarkers, auto-antibodies, models with clinical and demographic variables, multidimensional radiological features ("radiomics") and signatures using proteomics, genomics, transcriptomics, metabolomics, etc.2
A team from the "Vanderbilt-Ingram Cancer Center" succeeded in identifying from the large number of parameters those that do not represent additional stress for the patient (CT and blood samples were taken anyway). From these available parameters, they developed a prediction model that was also validated in an independent cohort. Their conclusion was that non-cancer patients in one of eight cases were spared an unnecessary invasive procedure.3,4
Incidental discovery of a lung nodule is so common that, unless there is a positive history of cancer or of smoking, people are often sent home for six months. "Biopsy is the only way to make a conclusive diagnosis, but bronchoscopy or needle biopsy come with costs and risks, especially for peripheral nodules, for which the pneumothorax rate is about 20%," explains first author Dr Michael Kammer. He says these risks of a biopsy can be seen as possibly unjustified as only about 5% of all lesions turn out to be malignant. "But those five per cent have the potential to be aggressive, and six months later it may already be too late."4
The team analysed data from 456 individuals who had a randomly detected lung nodule and who would have been considered at intermediate risk for lung cancer according to previously published standards (Mayo Clinic model). That risk was then assessed using the new model. In the benign intermediate-risk population, invasive procedures were reduced from 63% to 51% and the average time to cancer diagnosis was reduced from 60 days to 21 days for the intermediate-risk malignant cases, according to a clinical cost benefit analysis in two cohorts.
The limitations of the analysis involve the fact that CYFRA 21-1 is not (yet) a routine marker and that Vanderbilt, being a maximum care hospital, may have a biased composition of the study population. A published commentary on the study addresses the issue of subsolid nodules.5 These latter are being detected with increasing frequency in CTs and require a conventionally different approach. They may appear as pure ground glass opacities or show additional solid parts. Lesions that are purely ground glass opacities are often preinvasive (for example, in atypical adenomatous hyperplasia or adenocarcinoma in situ). Nodules with an additional solid component more often correspond to invasive changes. However, according to several preliminary papers, both forms share such a slow growth rate that many are unlikely to ever harm those affected.1,6 They state that a diagnostic examination is urgently required if the solid part grows by more than 2 mm.
Kammer et al. responded that their centre tends to monitor more aggressively than the national average, even significantly longer than two years in the case of subsolid nodules. But they also point out that many less suspicious subsolid nodules that remain stable over time are probably never referred to their centre (selection bias).7
Exciting times lie ahead: the next step will be to test the approach prospectively over two years at Vanderbilt or at the study's affiliate, the University of Colorado.