Lynn T. Tanoue


Curr Opin Pulm Med. 2016;22(4):327-335. 

In This Article

Future Directions

Screening has sometimes been likened to the search for a needle in a haystack with a pitchfork. It is clear that we need to make the definition of a positive finding more rigorous, analogous to making the needle bigger. The NLST definition of a positive baseline screening result was nodule diameter at least 4 mm; the LungRADS threshold of 6 mm should result in fewer small nodules being identified as abnormal, and a decrease in the number of false-positive results.[40] The NELSON trial, whose results are due shortly, takes nodule measurement into a third dimension.[25] The definition of a positive result in NELSON is a noncalcified nodule with solid component volume greater than 500 mm3. Indeterminate nodules with solid component volume 50–500 mm3 are considered positive if a volume doubling time of less than 400 days is identified at 3-month follow-up. This aggressive strategy decreases the false-positive rate and, when combined with models predicting likelihood of malignancy, increases the ability to identify true malignancy.[43,44]

There is much interest in increasing the efficiency of screening by defining populations at higher risk, that is, making the haystack smaller. Several validated lung cancer risk assessment models are already available.[45–49] Applying such models to the NLST participants demonstrated that, even within that high-risk population, there was a 10-fold difference in the number of lung cancer deaths averted in the highest compared with lowest quintiles of risk.[50]

There is an inherent tension in the concept that risk assessment models should be used to help identify and perhaps limit screening to individuals at higher risk. On the one hand, focusing screening in highest risk groups will lower the false-positive rate, decrease the number needed to screen to prevent one lung cancer death, and increase cost-effectiveness. However, the USPSTF already recommends screening only in older persons with heavy smoking histories. This group represents a fraction of the much larger population with other lung cancer risk factors, including less intense smoking, chronic obstructive pulmonary disease or interstitial lung disease, positive family history, exposure to domestic or occupational carcinogens, and others. The appreciation that the majority of lung cancers occur in individuals who do not meet the USPSTF criteria is reflected in recommendations to expand screening to a much larger group of at-risk individuals by the National Comprehensive Cancer Network and Association of American Thoracic Surgeons.[51,52]

The quandary is that, using a validated assessment tool, one may appreciate that an individual's lung cancer risk exceeds that of the lowest or even highest risk NLST participant, but that individual may not meet the USPSTF criteria for age and smoking intensity. The sheer scale of RCTs large enough to rigorously study the effect of screening in all populations known to have any lung cancer risk makes it impractical that they will ever be done, the consequence of which is we will never know whether screening these other high-risk patients is clinically or cost-effective.

With progress being achieved in decreasing the false-positive rate and drilling down on populations at higher risk, there is intense interest now in developing better tools for screening, that is, improving the pitchfork. These include efforts to more accurately identify high-risk individuals likely to benefit from screening, and assess whether a pulmonary nodule identified by screening is malignant or benign.

The rapidly growing field of lung cancer biomarkers includes promising work involving genomics, proteomics, and the analysis of exhaled breath.[53–58] Much of this builds on the concept of field cancerization, wherein many cells in an organ with common exposure to carcinogens develop premalignant changes. This is applicable to the effect of cigarette smoke, or other inhaled carcinogens, on the bronchial epithelium. Cells throughout the lung may exhibit shared genetic changes related to the exposure; one or a few may compile sufficient changes to become frankly malignant. Biospecimens retrieved from within the exposed field may yield information about the spectrum and intensity of genetic alteration, which may then be relevant to predicting the likelihood of malignant transformation. A recent study of a genomic classifier incorporating age and the expression of 23 cancer-related genes in bronchial epithelial cells demonstrated promising results in patients undergoing evaluation of lung nodules.[54] The sensitivity of bronchoscopy combined with the classifier was better than either bronchoscopy or the classifier alone in patients with an intermediate or high pretest probability of lung cancer, and the classifier also had a high negative predictive value in patients with an intermediate probability of lung cancer. There is at present no evidence to support the use of routine bronchoscopy for the evaluation of screen-detected nodules. However, anticipating the plethora of lung nodules that will be identified with implementation of LDCT screening, this type of multidimensional tool incorporating patient-specific clinical and biologic information merits further study, given the need to improve our accuracy in determining which nodules merit further invasive evaluation and which can be safely monitored.

Exhaled breath, which includes both volatile and nonvolatile compounds, is another appealing biospecimen, as it can be obtained noninvasively and samples the whole lung. Volatile organic compounds (VOCs) have been of particular interest; dogs trained to accurately 'smell' different cancers in the exhaled breath of patients are presumably identifying VOCs.[59,60] VOCs can be measured by a variety of techniques, including gas chromatography, mass spectrometry, colorimetry, and electronic nose instruments, with distinct patterns identified for different lung diseases, including lung cancer.[61,62]

Beyond clinical expertise, the tools we currently use in lung cancer screening are limited to patient age, smoking history, and LDCT scanning. Using validated predictive models and biomarkers to focus on populations at high enough risk to justify screening, and then to more accurately assess the likelihood that a positive finding is an aggressive cancer, will improve screening immensely. These advances are literally occurring at the present time, and will continue to redefine state of the art screening.