Lung Cancer Screening With Low-Dose Computed Tomography.

Background: Approximately 21 900 women and 35 300 men developed lung cancer in Germany in 2018, and 16 514 women and 28 365 men died of it. The outcome mainly depends on the tumor stage. In early stages (stage I or II), treatment can be curative; unfortunately, because early-stage lung cancers are generally asymptomatic, 74% of women and 77% of men already have advanced-stage disease (stage III or IV) at the time of diagnosis. Screening with low-dose computed tomography is an option enabling early diagnosis and curative treatment.

Methods: This review is based on pertinent articles retrieved by a selective search of the literature on screening for lung cancer.

Results: In the studies of lung cancer screening that have been published to date, sensitivity ranged from 68.5% to 93.8%, and specificity from 73.4% to 99.2%. A meta-analysis by the German Federal Office for Radiation Protection revealed a 15% reduction in lung cancer mortality when low-dose computed tomography was used in persons who were judged to be at high risk for lung cancer (risk ratio [RR] 0.85, 95% confidence interval [0.77; 0.95]). 1.9% of subjects died in the screening arm of the metaanalysis, and 2.2% in the control group. The observation periods ranged from 6.6 to 10 years; false-positive rates ranged from 84.9% to 96.4%. Malignant findings were confirmed in 45% to 70% of the biopsies or resective procedures that were performed.

Conclusion: Systematic lung cancer screening with low-dose CT lowers mortality from lung cancer in (current or former) heavy smokers. This benefit must be weighed against the high rate of false-positive findings and overdiagnoses.

Reck M, Dettmer S, Kauczor HU, Kaaks R, Reinmuth N, Vogel-Claussen J:

Lung cancer screening with low-dose computed tomography: current status in Germany.

Dtsch Arztebl Int 2023; 120: 387-92. DOI: 10.3238/arztebl.m2023.0099

In 2018, 21 930 women and 35 290 men developed lung cancer in Germany, and 16 514 women and 28 365 men died of it. Lung cancer is therefore the most common cancer-related cause of death in men and the second leading cause in women (1, 2). The majority of lung cancer cases in Germany are attributable to smoking tobacco; it is estimated that smoking causes over 60% of all lung cancer cases in women and up to 90% of all cases in men. The risk of developing lung cancer is significantly higher for smokers and former smokers than for never- smokers. Male cigarette smokers have an up to 24-fold higher risk of developing lung cancer than never-smokers, while the risk for former smokers is increased up to 7.5-fold. The risk for female smokers is increased up to 8.7-fold and up to twice as much for former smokers as compared with never-smokers (3). The risk of lung cancer increases with the total duration of smoking (years) and with increasing smoking intensity (cigarettes per day) (4, 5). The relative risk of lung cancer decreases for former smokers with increasing years of abstinence but never reaches the low level of a person who has never smoked.

About 85% of all lung cancer incidences in Germany are non-small cell lung cancers (NSCLC), while the remaining 15% are small-cell cancers (SCLC). The three main subtypes of non-small cell lung cancer are squamous cell carcinoma (about 30% of all lung cancer cases), adenocarcinoma (about 40%), and large-cell carcinoma (about 10 to 15% of all lung cancer cases). In Germany, the majority of lung cancer patients, i.e., 74% women and 77% of men, already have an advanced, prognostically unfavorable tumor stage at the time of diagnosis, with involvement of the mediastinal lymph nodes (stage 3) or metastatic spread to other organs (stage 4). By contrast, a diagnosis at an early tumor stage with the possibility of surgical resection and potential follow- up treatment has a much more favorable prognosis. This applies especially for NSCLC, with a 5-year survival rate of up to 80 to 93% for small, localized (stage 1A) tumors in comparison with a one-year survival rate of less than 20% in patients with advanced and metastatic (stage 4) cancer. However, only 17% of cancers in women and 15% of cancers in men are diagnosed at stage 1 (1).

Given the large prognostic, tumor stage-dependent differences and the often very late diagnosis, lung cancer is particularly suitable for a screening program with low-dose computed tomography (LD-CT).

The aim of this review article is to present the current status of CT screening with LD-CT and its implementation in Germany.

Because of the large cancer stage-related differences in survival time, effective early detection screening is the most efficient measure to significantly improve healing prospects and survival time. Both the Institute for Quality and Economic Efficiency in Health Care (IQWIG) and the German Federal Office for Radiation Protection (BfS) have now confirmed in two expert reports, based on a systematic evaluation of available evidence, a positive impact for lung cancer screening with low-dose computed tomography (LD-CT) (6, 7).

The meta-analysis by the Federal Office for Radiation Protection demonstrated a reduction in lung cancer mortality of 15% (risk ratio [RR] = 0.85; 95% confidence interval: [0.77; 0.95]) after screening with low-dose computed tomography in high-risk patients. Thus, 1.9 of participants died in the screening arm and 2.2% in the control group. The observation period was 6.6 to 10 years (7).

A meta-analysis by Hunger et al. concludes that LD-CT screening can reduce lung cancer mortality of (former) smokers provided the same stringent quality assurance and quality control measures are in place as in randomized controlled trials (RCT) (8).

So far, over 46 000 high-risk participants in ten RCTs have been screened with LD-CT and compared with control groups who were either examined only clinically (seven studies) or using standard chest radiographs (three studies). The number of examined participants in the individual studies varied widely between 53 452 (NLST) and 765 (DEPISCAN). Apart from total mortality, the primary endpoint was primarily lung cancer-specific mortality (9-21). The Table provides an overview of the various studies on lung cancer screening.

The majority of studies used age (between 50 and 60 years to 70 or 75 years) and smoking consumption

(20 to 30 pack-years) to define the risk cohort. This risk stratification certainly requires improvement and should be adapted to predict individual high-risk persons. This could improve both the efficiency and cost- effectiveness of screening (22).

With the exception of one study, subjects were screened multiple times, with between two and seven rounds of screening scheduled with an interval of usually one year, although this varied in some studies. So far, eight randomized studies (ITALUNG, LUSI, MILD, NELSON, NLST, DANTE, DLCST, LSS) (9-12, 14, 16, 18, 23) have demonstrated reduction in lung cancer mortality of 12%, with a risk ratio of 0.88 [0.79; 0.97] (24). A sex-specific difference was noticeable, showing a higher reduction in lung cancer mortality among women in the LUSI trial (RR 0.31 versus 0.92 among men), the NELSON trial (RR 0.67 versus 0.76), and the NLST (RR 0.74 versus 0.92) (9, 10, 21). This difference is possibly due to sex-specific differences in the occurrence of histological subtypes, with a predominance of adenocarcinomas among women. There is a clinically relevant shift in tumor stages at the time of diagnosis, with a higher rate of early tumor stages in the screening arms (Stage 1: 44% vs. 26% in the control arms) with a simultaneous reduction of advanced stages (Stage 4: 29 % versus 41 % in the control arms) (7). In addition, a reduction in overall mortality was also observed which, however, with a risk reduction of 6%, was lower than for lung cancer mortality alone (RR = 0.94 [0.89; 0.99]), (9, 11, 12, 14-16, 20, 21). With its high sensitivity, LD-CT is suitable for the early detection of lung cancer as it can also detect small (<10 mm) and usually asymptomatic lesions. Apart from the quality of LD-CT screening, other relevant factors for the implementation of screening for lung cancer include the frequency of false positive results, possible complications in the course of further diagnostic workup, and the risk of overdiagnosis, i.e., diagnoses that would not have manifested during the lifetime of the screening participants and would not have reduced survival time without screening.

Very heterogeneous sensitivities and specificities were observed in the individual studies, partly due to the different definitions of positive LD-CT results. Range of sensitivity varied between 68.5% in the MILD trial and 93.8 % (NLST) (25), and that of specificity between 73.4% (NLST) (25) and 99.2% (MILD) (14, 25, 26). A clinically significant point is the high rate of false-positive LD-CT results ranging between 84.9% (MILD) and 96.4% (NLST) (25, 26). This is the percent proportion of false-positive results from the cohort of all positives. This can create mental stress for those affected, as well as give rise to stressful follow-up examinations and sometimes invasive diagnostic procedures with a subsequent risk of complications.

The proportion of positive LD-CT results requiring invasive clarification within the studies DANTE, ITALUNG, LSS, LUSI, MILD, NELSON, NLST, and

Number of participants Criterion Result Design Screening effect ■ Trial Randomized Screened Tobacco consumption Cessation of smoking Age (yrs) (median) Proportion of males Screening rounds Interval Lung cancer mortality DANTE (11) 2450 1264 >20 py <10 yrs 64 100% 5 1 HR 0,99 95% CI [0.69; 1.43] Depiscan (17) 765 385 >15 cigs/d* 20 yrs <15 yrs 56 71% 3 1 n.s. z DLCST (15, 16) 4104 2052 >20 py <10 yrs 58 56% 5 1 HR 1.03 95% CI [0.66; 1.6] ITALUNG (12) 3206 1613 >20 py n.s. 61 65% 4 1 RR 0.7 95% CI [0.47; 1.03] LUSI (9) 4052 2029 >15 cigs/d* 25 yrs, or >10 cigs/d* 30 yrs <10 yrs 50-54: 46% 65-69: 11% 65% 5 1 HR 0.74 95% CI [0.46; 1.19] MILD (13, 14) 4099 1190/1186 (annually, biannually) >20 py <10 yrs 58 68% 7/4 0.5 HR 0.61 95% CI [0.39; 0.95] NELSON (10) 15 822 7915 >15 cigs/d* 25 yrs, or >10 cigs/d* 30 yrs <10 yrs 59 84% 4 1/2/2.5 RR 0.76 95% CI [0.61; 0.91] after 10 years UKLS (40) 4055 2028 LLP risk >5% in 5 yrs 67 6.3% 1 (baseline) RR 0.65 [0.41; 1.02] o ■- LSS (18) 3318 1660 >30 py <10 yrs <64: 68% 59% 2 1 RR 1.24 95% CI [0.74; 2.08] NLST (20) 53 452 26 722 >30 py <15 yrs <59: 43% 65 +: 6% 59% 3 1 RR 0.84 95% CI [0.75; 0.95]

HR, hazard ratio; yrs, years, CI, confidence interval; LLP, Liverpool Lung Project; py, pack years (average); RR, risk ratio; d, day; cigs, cigarettes

UKLS ranged from 5.1% in the ITALUNG trial to 19.4% in the NELSON trial, and the proportion of complications during invasive diagnostic clarification (at least one) was reported to range from 0.2% (LSS) to 1.7% (DANTE) (11-13, 16, 17, 19, 20, 27, 28).

The positive predictive value for positive screening results which were clarified by biopsy or surgical resection varied in the NLST, NELSON, and LUSI trials and in other large-scale screening studies between about 45 and 70%. That means that 45 to 70% of the biopsies or surgical procedures were performed for malignant results (9, 27, 29).

Given the short follow-up times, only few valid data are available at present on overdiagnosis of lung cancer after LD-CT screening. In a meta-analysis of five studies, (DLCST, ITALUNG, LUSI, MILD, NELSON), in which a risk calculation was performed, an average surplus of cancer diagnoses of 38% was identified in comparison with the total number of lung cancers detected by LD-CT. With a very long follow-up time after the end of screening, this excess bias can be interpreted as a measure of overdiagnosis (30, 31). However, the risk varied greatly between the individual studies, ranging from 18.5% (NLST) to 87.5 % (DEPISCAN) (17, 20). The risk of overdiagnosis fluctuates due to individual life expectancy and competing death risks. The follow-up times adopted by the studies varied considerably and are not long enough in the majority of the studies to make a valid estimation of overdiagnosis.

Calculation of radiation cancer risk is performed using various risk models, and the Federal Office for Radiation Protection has chosen modified risk models based on the data of the American BEIR (Biological Effect of Ionizing Radiation) VII Committee (32) and the Japanese Life Span Study of atomic bomb survivors (33). The potential risks for developing breast cancer and lung cancer are of particular importance when it comes to screening for lung cancer. This risk decreases with increasing age at the time of exposure, with a gender-specific comparison showing a higher risk of lung cancer for women. The estimated additional lifetime risk (LAR) for the German population attributable to radiation exposure of 10 mSV organ equivalent dose for women aged 60 years is 0.027% for lung cancer and 0.08% for breast cancer and is 0.015% for lung cancer in 60-year-old men. The additional LAR attributable to annual LD-CT screening is 0.11% for women in the same age group and below

Standardized requirements, based on the already analyzed RCTs, need to be created to ensure that the advantages of the screening examination outweigh radiation risks.

Participants must be given detailed information on the advantages of screening and its associated risks (34).

When the effects of LD-CT for Germany were extrapolated based on the NLST study (35), the data showed that, with a nationwide participation of 50% of the 55 to 74-year-old heavy smokers, around 1.3 million of this cohort would receive screening with LD-CT. After three rounds of screening performed once a year, 916 918 positive results and 32 826 confirmed lung cancers would be expected. Of the screening participants, 519 837 (39.1%) would have to expect at least one positive screening result. 12 449 screened individuals would experience (at least) one complication in the course of further diagnostic clarification, and 569 with a positive screening result, in whom lung cancer was suspected but not verified, would suffer a major complication. After an observation period of 6.5 years, 4155 (2.6%) deaths from lung cancer would be prevented (35). However, these calculations were made before 2015 and are dependent on the radiological classification of pulmonary nodules at that time and the subsequent multidisci- plinary management. Radiological LD-CT reporting, in particular, has evolved further since then with the advent of new volume-based detection and reporting software using artificial intelligence and of the radiological classification of pulmonary nodules.

In the USA, the US Preventive Services Task Force (USPSTF) recommends annual lung cancer screening with LD-CT for smokers aged 50 through 80 years who have at least a 20 pack-year smoking history (one pack-year is equal to smoking 20 cigarettes/one pack per day for one year) and those who have quit within the past 15 years (36). The recommendations are based on comparative simulation modeling for individuals from the 1950 and 1960 US birth cohorts followed from ages 45 to 90 years. It has been shown that selecting individuals for lung cancer screening based on their personal lung cancer risk has improved compared with the USPSTF criteria of 2013. Sociodemographic and clinical risk factors are incorporated in the risk prediction models and include age, smoking history, sex, ethnicity, family history, and previous history of emphysema and chronic obstructive lung disease.

In Europe, Poland, Croatia, and Italy are currently implementing national screening programs for high- risk individuals (38). It is planned to engage primary care physicians to recruit patients in these countries.

Despite the confirmed efficacy of lung cancer screening, only in two to eight percent of subjects included in the controlled studies analyzed by the BfS were diagnosed with lung cancer. Supplementary biomarker analyses, for example, using blood or respiratory air, could possibly increase the efficacy of radiological screening. This could reduce the number of false positive results and correct any false negatives. Tumor cells, proteins, auto-antibodies, nucleic acids, and other molecules are very promising in this respect. They can either be taken directly from the tumor or its surrounding tissue or obtained during liquid biopsy as circulating markers from blood, urine, or other bodily fluids. While the spectrum of potential biomarkers is developing very dynamically, the obligatory controlled validation in the form of defined clinical cohorts, especially the potential effect on sensitivity by a sequential approach, is a challenging and lengthy process (39).

No national screening program for lung cancer has so far been implemented in Germany. Based on the scientific evidence on the reduction of lung cancer mortality using LD-CT screening, the Federal Office for

Radiation Protection (BfS) and the Institute for Quality and Economic Efficiency in Health Care (IQWiG) have published positive assessments (6, 7). Based on these reports, a legal regulation is currently being prepared for the Federal Ministry for the Environment, Nature Conservation, Nuclear Safety and Consumer Protection (BMUV).

Apart from a clear definition of the high-risk population, decisive points in its implementation will be the organization and content of patient information, integration of an offer for smoking cessation treatment, as well as the quality assurance of the screening and all subsequent diagnostic or therapeutic steps.

The professional societies involved are to be responsible for developing the specific recommendations for implementation. Of particular importance in this regard is the feasibility of adequate and widespread recruitment by pulmonologists and retention of the target population in a future national program with adequate infrastructure and budget to deal with the approximately six million people at risk (7). A clear distinction must be made here between early detection as part of lung cancer screening and further clinical management as a patient in the event of a positive result (Figure).

Two studies are currently being conducted to gain new insights into the specific implementation and future development of screening for lung cancer. The holistic implementation of quality-assured lung cancer screening in conjunction with a better definition of the target population and other issues is currently being assessed in the multifocal, North German Hanse trial (Box). Initial results are expected by the end of 2023 (www.hanselungenscheck.de). In addition, the multicenter European 4-In-The-Lung-Run project with involvement of the German Cancer Research Center in Heidelberg and the University Hospital of Essen is examining innovative approaches to optimize the screening interval and other issues in a randomized trial (eFigure, eBox).

Given the scientific evidence of the reduction of lung cancer mortality by LD-CT screening, it is to be expected that screening for lung cancer will also be introduced for the high-risk group in Germany.

The authors declare that they have no conflict of interest.

Manuscript received on 15 March 2022, revised version accepted on 5 April 2023.

Translated from the original German by Dr Grahame Larkin MD.