To ultimately find what we are actually looking for, the invasive malignant nodule in a haystack of benign lesions, new strategies and qualitative and quantitative tools are needed to propel noninvasive evaluation of solitary pulmonary nodules into the 21st century.
The article by Drs. Brawley and Flenaugh in this issue of ONCOLOGY deals with one of the most common findings in thoracic radiology: the incidental, solitary pulmonary nodule (SPN). Not only in the lung cancer screening environment, but also in imaging in oncology and general medicine, SPNs are a frequent finding and cause a lot of concern for the reading radiologist and treating physicians. On one hand, early detection of a possibly harmful pulmonary nodule offers the chance for an earlier, potentially curative treatment. On the other hand, the sheer number of SPN cases calls for pragmatic, noninvasive strategies for evaluation and follow-up.
The authors present the impact of SPNs predominantly from a screening point of view, and discuss several tools that are used in the evaluation of SPNs. The challenges around SPNs are manifold, and have evolved over the last decade. “Detection” alone used to be a key target; then, over time, as the technology improved, slice thickness was decreased to the point where lung cancer screening protocols now recommend 1-mm images. Computer-aided detection (CAD) software has also been developed and tested. Publications discussing CAD applications for lung nodule detection have declined considerably in the last few years, partly because the ratio of true positives to false positives could no longer be improved, and partly because of radiologists’ declining interest in methods to further increase the number of tiny indeterminate nodules detected.
The next challenge is “characterization,” and that is the topic covered in greatest depth in the article. After all of the nodules are detected, most need to be quickly and pragmatically discarded as nonrelevant, and only those that are potentially malignant need to be filtered out. To date, nodule characterization has been based on simple morphologic features such as location, size, margin, and density. Most commonly, these features are assessed and combined by the reading radiologist in the final report; in a more sophisticated technique, they can be combined to produce a “quantitative prediction model.” In this context, “quantitative” refers to an estimated risk that a nodule is malignant. Computer aided diagnostic (CADx) software has been developed that combines morphologic features and highlights the likelihood of a nodule being malignant, but such programs have not been widely accepted. Beyond the standard morphologic features, no truly quantitative features are available. As discussed in the article, positron emission tomography (PET) has only a very limited role in the evaluation of SPNs. Three-dimensional (3D) measurements of a nodule volume work very well in phantoms, but biology (breathing, heart pulsations), data acquisition (noise, pixel size, filters, etc), and segmentation (attachment and inclusion of adjacent anatomy such as vessels, different algorithms from different vendors) limit their accuracy and reproducibility in real life, and comparison of two subsequent scans to quantify the growth rate is often not very accurate. Quantitative nodule evaluation has not made it into routine clinical practice, and true quantitative measurements are still lacking.
The most recent challenge in the context of SPNs that has also been mentioned in the article is the “management” of SPNs. It has only been very recently appreciated that not every malignant nodule actually has to be resected or treated; the concept of overdiagnosis and overtreatment arose mainly from lung cancer screening studies through the detection of asymptomatic, noninvasive or minimally invasive adenocarcinomas that are unlikely to become relevant in a patient’s lifetime. Image-guided biopsy does not really help with this management, as the mere presence of malignant cells does not reveal any information about invasiveness. This area is a wide-open field in radiology, and currently available morphologic tools are too coarse to discern what structure would differentiate an invasive lesion from an noninvasive one that could be left alone.
Just as the authors mention that pathology remains rooted in the 19th century and needs to move forward, radiology seems to be trapped by 20th century designations ascribed to specific morphologic features. Morphology is of limited utility in both characterization and management of malignancy, and radiology needs to branch out into entirely different areas. Promising tools include perfusion imaging, the kinetic analysis of the passage of contrast material through a nodule. This technique is currently under research evaluation and is not yet ready for clinical use, therefore it is not mentioned in the article.
New strategies and qualitative and quantitative tools are needed to propel noninvasive evaluation of SPNs into the 21st century. These will help us to avoid an inundation of detected SPNs of little or no clinical consequence, to decrease patient harm, and to ultimately find what we are actually looking for: the invasive malignant nodule in a haystack of benign lesions.
Financial Disclosure:The author has no significant financial interest or other relationship with the manufacturers of any products or providers of any service mentioned in this article.
1. Low dose spiral CT screening and evaluation of the solitary pulmonary nodule. Oncology (Williston Park). 2014;28:441-6.