Start Your Search
David F Yankelevitz
ES01 - Advances in Lung Cancer Screening Through Imaging (ID 769)
- Event: WCLC 2018
- Type: Educational Session
- Track: Screening and Early Detection
- Presentations: 1
ES01.02 - Image Interpretation and Advances from the Perspective of the Radiologist (ID 11352)
13:50 - 14:10 | Presenting Author(s): David F Yankelevitz
To a large extent the success of the screening process depends on the algorithm used to manage the findings. Therefore, even though the image production might be identical for two different screening programs, if they differ in the way they manage findings, especially nodules, the comparative results between them would be quite different. It is with this in mind that screening programs develop and choose their workup algorithm. In regard to how these algorithms are chosen, several features must be balanced including the rate of false positives and also potential delays in diagnosing lung cancer. Each of these is associated with a potential cost, the fewer the false positives might be associated with an increase in the number of cases where lung cancer diagnosis is delayed. As we have moved from 5 mm size thresholds to 8 mm size thresholds on the baseline scan we can clearly see how these factors are balanced against each other. The number of false positives dramatically declines as the number of cancers where diagnosis will be delayed by nine months will increase. While it is generally assumed that there is a substantial downside to delaying diagnosis, the challenge is understanding how much the delay actually costs in terms of decrease in the curability of the cancer and this is then considered in terms of how often this might occur.
Another major feature of management protocols is their dependence on change in size over time to guide the management protocol. Different protocols apply different criteria for measuring change. One of the main differences is that some recommend the use of 3D volumetric analysis while others still rely on 2D measurements. As a general rule, the 3D approach has inherent advantages in that boundaries for the nodule are automatically chosen by the computer, asymmetric growth can be more easily recognized, and the proportional change for a given amount of change is far greater for volume then for diameter measurements. Nevertheless, there may still be circumstances where volume measures can still be misleading and the radiologist still has a very important role in visually inspecting the nodule to confirm whether change has occurred. Along with the measurement of size change, the time interval between measurements is also important in determining whether growth is meaningful. It is not simply enough to say that a nodule is growing, but rather the intent is to understand its growth rate, and this depends on time, with shorter time intervals between scans introducing greater uncertainty.
There are currently many algorithms that have been developed. Some focus solely on screening such as Lung-RADS, while others are designed primarily for the incidentally detected nodules. Differences between the algorithms focus primarily on the size thresholds used to define a positive result, the time intervals between repeat scans, the choice of management for a positive finding, differences in the management of nodule subtypes (solid, part-solid, nonsolid), and differences between baseline rounds and repeat rounds. These different algorithms will be compared and data will be presented in terms of the influence on the rate of positive results. An additional consideration here is also how we define a positive result. Some algorithms define the positivity based on a size threshold, whereas others consider this based on a growth threshold or a combination of size and growth. When these growth thresholds are used, the rate of positive results dramatically decreases.
In addition to the finding of lung nodules there are many other findings that commonly occur on the scans such as micro-nodules, areas of atelectasis, perifissural nodules, waxing and waning nodules, endobronchial nodules, presumed pneumonias, that might be found by the radiologist but there are no specific guidance rules for management. Here again the radiologist is confronted with the challenge of attempting to balance excess workup against obtaining a firm clinical diagnosis. While many of these examples have no authoritative guidelines as to how they should be managed, some practical guidance is presented.
S01 - IASLC CT Screening Symposium: Forefront Advances in Lung Cancer Screening (Ticketed Session) (ID 853)
- Event: WCLC 2018
- Type: Symposium
- Track: Screening and Early Detection
- Presentations: 1
- Moderators:John Kirkpatrick Field, James L Mulshine
- Coordinates: 9/23/2018, 07:00 - 12:00, Room 203 BD
S01.04 - Lung Cancer Screening: 1999 to Date – What Have We Learnt? (ID 11885)
07:35 - 07:50 | Presenting Author(s): David F Yankelevitz
In 1999, ELCAP published their initial results from baseline screening. It found that in a cohort of 1000 participants approximately 85% of the cancers could be diagnosed as clinical Stage I, and that compared with chest radiography found many more of the cancers. In a subsequent study the expanded I-ELCAP found that the long term survival as a measure of cure rate approached 80%. The publicity associated with this initial study was quite large and led to the initiation of several other trials including the NLST. The NLST published their results in 2011 and based this, screening was endorsed by insurers in the US and now other countries are similarly following suit. However, despite the positive result of the NLST, and reimbursement from insurers, screening has had extremely limited uptake in the US, with only approximately 2% of those eligible (among a restricted population) are being screened. Thus, we face a situation where the most common cancer killer has been studied in the most expensive screening trial ever performed which had a positive result, insurers are reimbursing for it, and few people are having it done.
With lung cancer screening being touted as a major breakthrough in the war on cancer the question naturally arises as to why it is not being performed more frequently. There have been many reasons to explain the poor uptake, ranging from merely a slow start but expected steady increase, lack of awareness by the clinician or potential screenee, obstacles such as the shard decision making requirement, too many potential harms, and lack of significant benefits.
This lack of perceived significant benefit is perhaps the most important aspect, since without a substantial benefit, even if the harms were minimized, why would anyone get screened and why would a clinician recommend it. It seems that this is clearly influencing the decision not to be screened as many experts and even guideline organizations consider the benefits to not be sufficient enough so as to recommend the screening. Even CMS considered the balance of the risks and benefits so tenuous that they took the unique step of requiring a shared decision making process to be included as necessary for reimbursement so that a person could balance the risks and benefits.
It is this aspect of benefit that needs to be considered more carefully when explaining it to a potential screenee. Current decision aids, which are required as part of a shared decision making process, in the US and Canada rely almost exclusively on the NLST result and attempt to convert its findings into more visual aids. However, in translating those NLST results, it needs to be understood that they were highly dependent on the design parameters of the study itself, namely 3 rounds of screening and 6.5 years of follow-up. When these parameters change so do the benefits. In the US, current recommendations for screening include annual screening over the period of eligibility for the participant (although for Canada it is restricted to 3 years). Under the circumstance of continued annual screening, the reduction in mortality begins to approach the estimated cure rate for the cancer. It is this feature of cure rate that is really what is most important to any person interested in being screened, and it is substantially higher than the mortality reduction seen in a randomized trial where by necessity the mortality reduction is diluted by the time interval after screening has stopped and cancers are still being followed, and also by not including those cancers that are relatively slow growing and cured as a result of early treatment but not counted towards the mortality reduction because the trial has concluded before their counterpart in the control arm has died. Based on these considerations, it is possible to have a cancer that is 100% curable when screen detected, yet the trial may only show a 20% (or even lower) mortality reduction. Thus, there is inherently no incompatibility between the 80% cure rate seen in the I-ELCAP compared with the 20% mortality reduction seen in the NLST. The simple conversion of the 20% mortality reduction found in NLST into a cure rate as is so commonly done when explaining the benefit to a person interested in screening is highly misleading. The cure rate, which is the clinically relevant feature, is higher. This coupled with the way in which harms are currently expressed, based again almost solely on those NLST results has the effect of amplifying harms at the same time the benefits are being underestimated and surely affect the perception of overall value of CT screening both for physicians as well as people who might be interested.