Low Dose CT for Lung Cancer Screening: The Background, the Guidelines, and a Tailored Approach to Patient Care

Abstract

Despite lung cancer’s prevalence and high burden of mortality, a useful screening program has taken a long time to develop. Now, most of the associated national organizations recommend low dose CT screening in the appropriate population. However, since the USPSTF guidelines were published, implementation has been slow. This article outlines the current evidence and provides additional resources to help physicians tailor a screening program for their patients.

Introduction

Lung cancer is the second most prevalent cancer overall, and is the principal cause of cancer-related deaths in the United States.¹ Until the development of dyspnea from metastatic pleural effusions, hemoptysis and cough from invasion of the pulmonary vasculature and bronchial tree, or pain from bony metastasis, these cancers themselves do not tend to cause symptoms. Thus, two-thirds of patients are not diagnosed until they have Stage III or IV disease, at which time their 5-year survival is 16% and 4% respectively.¹

Given this heavy burden of disease and high mortality, much effort has been put forth into early recognition of pulmonary nodules to aid in early diagnosis. Attempts to identify an effective screening test have been underway for decades. Simple chest radiography and sputum cytology have proven ineffective as a screening method.² Beginning in the 1990s, there was increasing interest in CT imaging as a screening modality. These studies initially established superiority over chest radiography, with one study demonstrating that 73% of CT-detected cancers were not visualized on screening chest x-ray.³ Later studies reaffirmed a place for low-dose CT (LDCT) as a screening tool to aid in a more timely diagnosis for the growing number of patients falling victim to this growing epidemic.^2,4,5,6,7,8

LDCT Screening

Establishing an effective screening program has not been easy. Many of the studies in the early 2000’s were limited by small sample sizes. Over the last 10 years, several larger studies have been performed to analyze whether targeted CT screening programs would provide a mortality benefit. The Danish Lung Cancer Screening Trial (DLCST) compared 4,104 patients divided between an annual LDCT screening exam and no intervention. After five years, more cancers – primarily early stage – were diagnosed in the LDCT group. There was no significant difference in the rate of cancer-related or all-cause mortality.⁴ The Multi-centric Italian Lung Detection (MILD) trial randomized 4,099 patients amongst three groups – an annual and biannual LDCT group, as well as a control group. Over a five-year period, more lung cancers were diagnosed in the annual LDCT group, though there was no impact on lung cancer-related or overall mortality.⁵ The Detection and Screening of Early Lung Cancer (DANTE) trial was a study of men aged 60–75 with a 20+ pack-year history. Each patient had baseline chest radiography and sputum cytology. Thereafter, patients were randomized to annual medical exam versus LDCT. Unfortunately, the study was underpowered to detect a mortality difference.⁶ The U.K. Lung Screen (UKLS) trial randomized 4,055 patients between a CT arm and control arm. This study was also underpowered for detecting a mortality difference, but did result in more early-stage cancer diagnoses, and did support LDCT as a financially plausible screening tool for the diagnosis of lung cancer in the U.K.⁷

In 2011, the National Lung Screening Trial (NLST) research team published the largest trial to date with over 53,000 participants over 5 years.⁸ Study participants were between the ages of 55 and 74 with a greater than 30 pack-year smoking history who continued to smoke or quit within the last fifteen years. They were randomized to be screened annually for 3 years with LDCT versus standard chest radiography.⁸ In this study, there was a higher rate of both true positive and false positive tests in the CT arm, with a sensitivity of 93.8% and a specificity of 73.4%.⁹ This detection led to a significant decrease in cancer-related mortality with a relative risk reduction of 20%, as well as an overall decrease in all-cause mortality by 6.7%.⁸ Other important findings in this study included a relatively high rate of positive screening tests. Of the 25,309 patients in the CT arm, 27% had a positive screening CT, and 3.8% were ultimately diagnosed with cancer. For the majority, positive screens were followed with repeat imaging, but in 6% of patients with benign disease, invasive procedures were undertaken to obtain a diagnosis.⁸ This was not significantly different from the standard radiography arm, in which 5% of patients with benign disease underwent invasive procedures. Fears exist as to the detrimental effects of anxiety, as well as unnecessary costs as a result of false positive screening exams.^10,11 Despite this, the NLST study has been monumental in bringing about the lung cancer screening (LCS) guidelines in the United States as we know them. The limitations of the study reflect the importance of shared decision making and clear identification of a high-risk cohort who would benefit most from screening.

Current Guidelines

Many national organizations have published guidelines for LCS (Table 1). They are, for the most part, in agreement. Any patient over the age of 55 with a 30-pack-year smoking history who continues to smoke, or has quit within the last 15 years and has no symptoms of lung cancer may be considered for LDCT-LCS.¹² For patients meeting these criteria, LCS is covered under Medicare.¹³ Up for debate amongst these organizations is at what age this screening should stop (between 74–80 years of age – Medicare coverage up to age 77),¹³ and the role other risk factors such as non-smoking-related exposures and family history of lung cancer should play in qualifying a patient for screening.^{12,14,15,16,17} Most organizations also include a caveat that patients who have other comorbid conditions which would make them unable to tolerate surgical resection or other treatment should not be screened.^12,14,15,16 Additionally, given the relatively high false-positive rate, particularly on initial screening, as well as the potential for over-diagnosis (finding indolent cancers that are not life-limiting),² shared decision making is also emphasized as an important condition for enrollment.^{12,13,14,15,16,17}

Table 1.

General Recommendations:

• Start Screening: Age 55
• Smoking history: 30 + pack years. Current or former smoker within the last 15 years
• Asymptomatic for lung cancer
• General health status which would tolerate treatment if cancer were identified

U.S. Preventative Services Task Force	Stop Screening: Age 8012
Centers for Medicare and Medicaid Services	Stop Screening: Age 7713
American College of Chest Physicians	Stop Screening: Age 7714 Should not be performed in those thought to be high risk but do not meet above criteria Could be considered, however, in those thought to be healthy enough to benefit.
National Comprehensive Cancer Network	Stop Screening: Age 7415 Consider screening in patients >50 years of age with 20+ pack year history and additional risk factors that make the individual ‘high risk’ Consider screening those thought to be high risk are evaluated using the Tammemagi lung cancer risk calculator and receive a risk score >1.3% Risk factors include: radon exposure, occupational exposure, personal history of cancer or first degree relative with history of lung cancer, second hand smoke exposure, or personal history of COPD or pulmonary fibrosis.
American Association for Thoracic Surgery	Stop Screening: Age 7916 Consider screening lung cancer survivors after 4 years of negative surveillance Consider screening in patients >50 years of age with 20 pack year history and other factors that increase risk of developing cancer to >5% in the next 5 years

Critical exceptions to these guidelines include: patients who have symptoms suggestive of lung cancer, such as unintended weight loss or hemoptysis should have a full dose diagnostic CT chest,¹⁴ and patients with signs or symptoms suggestive of infection should have their screening test postponed several weeks until the infection has been treated or resolved to minimize false positive exams.

Barriers to Lung Cancer Screening

There are a number of reasons why physicians are slow to refer their patients for lung cancer screening – unsurprisingly, time constraints play a major factor. ^18,19,20 Shared decision making is key for quality patient-centered care. Unfortunately, studies have shown that providers as a whole are not participating in quality shared decision making discussions. In one study of providers enrolling patients for LCS, discussions were scored on a scale from 0 to 100, with 100 being a comprehensive discussion. The average score of all providers was 6, with the highest score being 17. Most discussions were less than one minute in length.¹⁸ The time for shared decision making is noted as a barrier in many studies,^18,19,20 and few are aware of Medicare billing code G0296, which reimburses for time spent on shared decision making.²⁰ Of note, to be reimbursed, the physician must document a patient’s eligibility, including pack-year smoking history, current smoking status including years since quitting, as well as provide clear documentation of the discussion of risks, benefits, ability and willingness to undergo treatment, smoking cessation education if applicable, and any decision aid utilized in guiding the choice to enroll.¹³

Apart from insufficient time with patients, a number of other barriers to LCS exist on the part of physicians. These include prior authorizations, failure of insurance coverage, and concern over financial burden or potential physical and emotional harm to patients who require additional workup.^10,11 Concern that a negative result will interfere with smoking cessation efforts is a reported concern as well.^10,12 As the guidelines note, screening is not meant to replace smoking cessation efforts.^12,14,15,17 The ultimate goal of lung cancer prevention via smoking cessation clearly supersedes that of early diagnosis.¹² Studies have not upheld the idea that an unremarkable LDCT hinders efforts to quit – in fact a trend towards higher rates of tobacco cessation have been reported.¹⁷ Confidence in the utility of LDCT for lung cancer screening is also lacking, despite the data that shows the number needed to screen after 3 years is less than that of mammography, and sigmoidoscopy.¹⁹ Unfortunately many physicians continue to order simple chest radiography for cancer screening.^11,19

One of the other important limitations for widespread LDCT-LCS is geographic. According to the CDC, as of 2017, 28% of qualified patients did not live within a 30-minute drive of a screening center.²¹ While this is a problem throughout the country, the majority of those patients were located in the northern Midwest. Missouri is a high mortality, medium access state. The surrounding states are all medium access as well, with Arkansas being the exception. Arkansas is part of a southeastern band of states with high lung cancer related mortality. However, Arkansas has disproportionately low screening access.²¹

More recent studies show some improvement in the numbers of physicians referring their patients for screening, though rates of enrollment in LDCT-LCS across the country remain suboptimal.^11,18,19,22 Patients themselves may be hesitant to enroll. Reasons for this are not consistent in studies. A study from France found that current smokers were more likely to desire screening if they were male, and had an intention to quit smoking.²³ Those with few or no comorbid conditions are also more likely to desire LCS, as well as those seeing their own primary care provider.^22,23 This highlights the importance of the relationship between patients and their primary care physician, as well as the need for primary care providers to be the forerunners in promoting LCS for the appropriate population.²² Radiation dose is occasionally a concern for patients and providers, but it is important to note that LDCT provides approximately a 1.4 mSv radiation dose, as opposed to 7mSv delivered by conventional chest CT.²⁴ For patients or physicians who desire more information on the benefits and risks associated with LCS, www.shouldiscreen.com is a useful resource created by the University of Michigan to help guide decision making.²⁵

My Patient Has a Positive Screen – Now What?

The American College of Radiology has developed a standardized process for reading LDCT entitled LungRADS, similar to that which has been published for mammography.²⁶ Their website features a comprehensive table illustrating the various radiographic appearances of lung nodules, and the recommended follow up interval based on a nodule’s malignant probability.²⁶ Based on these characteristics, a screening CT scan is given a category 0–4B: “0” meaning there is insufficient data for interpretation and comparison images are required, “1” being negative, and “4 A/B” categorized as suspicious and requiring short-term follow up with CT, PET/CT or biopsy (Figure 1).^26,27 For example, ground glass nodules, even those larger than 2cm, are often benign or represent indolent cancers, and are given a category 2 or 3 with a follow up interval of no sooner than 6–12 months.²⁶ Further, part-solid nodules are higher risk than completely solid nodules of the same size, and new small nodules are higher risk than their larger, stable counterparts. Higher risk nodules are of course prescribed closer follow up intervals.²⁶ Notably, a PET-CT is only recommended for nodules with an 8mm or larger solid component, due to the test’s limited sensitivity for smaller nodules. Other categories, such as 4X, S and C are additional findings or modifiers which may guide further testing.²⁶

Figure 1.

Suspicious spiculated pulmonary nodules – each >1cm which require further evaluation with follow-up CT versus PET-CT versus biopsy, depending on patient-specific and nodule-specific factors (symptoms, comorbid conditions, evolution of the nodule, presence of other nodules or adenopathy, etc.)

Currently, guidelines support annual screening exams, with interval low-dose diagnostic scans in the event of a positive screen. Aside from scheduling ease, there is no clear basis for this recommendation, and studies are underway to scrutinize this practice.²⁸ A recent review article suggests that in patients with a negative initial screen and reassuring risk profile, a longer interval between screening exams (lying somewhere between 1 and 2 years) may be reasonable.²⁸ However, in order for a screening program to be successful, ease of application is key. A complex screening algorithm is unlikely to be optimally implemented and will thus not bring about the benefits that make a screening program worthwhile – i.e. those of early detection, early treatment and decreased mortality. As more patients are enrolled in LCS, we will hopefully be better able to identify this optimal timeline to minimize the financial burdens of screening, while maintaining its benefits.

Going Forward – Minimizing Unnecessary Workups

A number of risk calculators have been developed to help physicians identify patients in whom an invasive workup might be more appropriate. The Mayo Calculator was developed in the 1990’s and identified a handful of variables which are independent predictors of lung cancer. These predictors include a patient’s age, smoking status, history of cancer, as well as nodule-specific characteristics such as size, spiculation, and location (upper vs lower lobe).²⁹ The Brock calculator was developed in 2013 and incorporated additional factors including the presence of emphysema, radiographic qualities of the nodule, i.e. ground glass, solid or part solid, total number of nodules, and family history of lung cancer.³⁰ When compared, the Brock and Mayo calculators perform similarly, with an AUC of around 0.9 for each.^31,32 In 2015, the Bayesian Inference Malignancy Calculator (BIMC) was created, which took into account a nodule’s volume doubling time³² – a factor which has been shown to be a strong predictor of malignancy.³³ The BIMC also utilizes PET-CT data for further prognostication.³⁴ Though the BIMC has not been compared directly to the Brock model, the BIMC has been shown to be more predictive of risk than the Mayo calculator.³⁵ Although no studies have compared these predictive models together, any of them may be utilized to help guide decision making when developing a plan of care for patients with pulmonary nodules.

Genetic and serologic methods for noninvasive evaluation of pulmonary nodules have also been developed. EarlyCDT-lung test measures a panel of auto-antibodies found in patients with lung cancer.³⁶ In a validation study of the 7 antibody assay, the test was only 47% sensitive, but 90% specific. Whereas the majority of patients with a negative assay did not have a cancer diagnosis, 89.7% of those with a positive antibody assay also did not have a diagnosis of cancer. Of the 19 with a cancer diagnosis, only 9 had a positive test.³⁶ In a larger study of 1613 patients, sensitivity was similar at 41%, and again of the 35 patients with lung cancer, only 13 had a positive antibody test.³⁷

Percepta is a genetics-based study using normal bronchial cells obtained at the time of bronchoscopy for suspected lung cancer.³⁸ Despite a normal macroscopic appearance, this test helps identify a pattern of abnormal gene expression in cells neighboring a malignant nodule. The specimen is stable for up to 20 days in refrigeration, and may be sent after bronchoscopy is ruled indeterminant.³⁹ In the initial validation study, the addition of genomic testing to bronchoscopy had a 97% sensitivity for detecting cancer.³⁸ Those with a low to intermediate probability of cancer had a negative predictive value of 100% and 91% respectively. It is important to note, however, that in those with a high pretest probability, the negative predictive value was only 38%.³⁸ Of those diagnosed with cancer despite initially indeterminant bronchoscopy, 87.4% had a positive genomics test. Just over half of those with benign disease, however, will also return positive.³⁸ When applied to clinical practice, Percepta decreases the rate of recommendation for invasive procedures in patients with low risk genetic profiles.⁴⁰ As with any test, it is important to identify the key cohort for implementation. Those patients with low risk for malignancy and high risk for invasive procedures might be best suited for genomic evaluation as a means to mitigate the risk of potentially unnecessary procedures.

Discussion

Since the USPSTF guidelines were published in 2016, LDCT for LCS has been a slowly growing trend. Many studies have demonstrated its utility in diagnosing more early-stage cancers than standard clinical practice and plain chest radiography.² Chest radiography has been proven ineffective as a lung cancer screening method and should no longer be utilized, although the prevalence of this remains surprisingly high.^11,19

There are a number of caveats to the basic guidelines for LDCT screening (ages 55–77 with >30 pack year history who smoke or quit within the last 15 years), which all primary care and pulmonary physicians should be well acquainted with. Firstly, physicians must involve their patients in a detailed shared decision making process prior to enrolling them in a lung cancer screening program. Additionally, physicians should be mindful of the patients that they are referring. Those who would be unwilling or unable to tolerate treatment should not be enrolled. While the incidence of pulmonary nodules is high, particularly in the Midwest, the relative incidence of lung cancer is low, and it is important to discuss the risk of a false positive screening test with patients. In most cases, a repeat CT scan is all that is necessary to follow these abnormalities, though invasive procedures are sometimes required for confirmation. Thankfully, there are a number of tools available to help guide further workup, including risk calculators, antibody testing, and genomics, each with their own limitations. Of course, smoking cessation counseling is crucial, and pharmacologic therapy to assist patients in their efforts to stop smoking should be considered.

Although some of the smaller studies did not demonstrate a mortality benefit with LDCT, nearly all of them successfully demonstrated a higher rate of early-stage cancer diagnosis. Five-year survival for stage I and stage II disease is 55% and 35%, respectively.¹ If observed for a longer period of time, a mortality difference perhaps could have been identified in these smaller studies, as fewer patients were allowed to progress to late-stage disease. With time and education, more physicians will become active in patient enrollment for LCS, and this mortality benefit found in trials may in fact be realized.