Journal of Cancer Research and Clinical Oncology

, Volume 138, Issue 9, pp 1475–1486

Randomized study on early detection of lung cancer with MSCT in Germany: study design and results of the first screening round

Authors

    • Division of Cancer EpidemiologyGerman Cancer Research Center
  • E. Motsch
    • Division of Cancer EpidemiologyGerman Cancer Research Center
  • M.-L. Gross
    • Division of Cancer EpidemiologyGerman Cancer Research Center
  • A. Eigentopf
    • Division of Cancer EpidemiologyGerman Cancer Research Center
  • C. P. Heussel
    • Department of Radiology, Thoraxklinik HeidelbergHeidelberg University
    • German Center for Lung Research
  • H. Dienemann
    • Department of Surgery, Thoraxklinik HeidelbergHeidelberg University
    • German Center for Lung Research
  • P. A. Schnabel
    • Institute of PathologyHeidelberg University
    • German Center for Lung Research
  • L. Pilz
    • Medical Faculty MannheimHeidelberg University
  • M. Eichinger
    • Department of RadiologyGerman Cancer Research Center
    • German Center for Lung Research
  • D.-E. Optazaite
    • Department of RadiologyGerman Cancer Research Center
  • M. Puderbach
    • Department of RadiologyGerman Cancer Research Center
    • Department of Radiology, Thoraxklinik HeidelbergHeidelberg University
  • J. Tremper
    • Department of RadiologyGerman Cancer Research Center
  • S. Delorme
    • Department of RadiologyGerman Cancer Research Center
Original Paper

DOI: 10.1007/s00432-012-1228-9

Cite this article as:
Becker, N., Motsch, E., Gross, M. et al. J Cancer Res Clin Oncol (2012) 138: 1475. doi:10.1007/s00432-012-1228-9

Abstract

Purpose

Low-dose multislice-CT (MSCT) detects many early-stage lung cancers with good prognosis, but whether it decreases lung cancer mortality and at which costs is yet insufficiently explored. Scope of the present study is to examine within a common European effort whether MSCT screening is capable to reduce the lung cancer mortality by at least 20 % and at which amount of undesired side effects this could be achieved.

Methods

Overall 4,052 heavy smoking men and women were recruited by a population-based approach and randomized into a screening arm with five annual MSCT screens and an initial quit-smoking counseling, and a control arm with initial quit-smoking counseling and five annual questionnaire inquiries.

Results

In the first screening round, 2,029 participants received a MSCT providing 1,488 negative and 540 suspicious screens with early recalls (early recall rate 26.6 %) leading to 31 biopsies (biopsy rate 1.5 %) and 22 confirmed lung cancers (detection rate 1.1 %). Among the lung cancers, 15 were adenocarcinomas, 3 squamous cell carcinomas, one small-cell lung cancer, and 3 others, whereby 18 were in clinical stage I, one in stage II, and 3 in stage III. One interval cancer occurred.

Conclusions

The indicated performance indicators fit into the range observed in comparable trials. The study continues finalizing the second screening round and for the first participants even the last screening round. The unresolved issue of the precise amount of side effects and the high early recall rate precludes currently the recommendation of MSCT as screening tool for lung cancer.

Keywords

EpidemiologyLung cancer screeningLUSIMultislice-CTRCT

Introduction

With almost 41,000 deaths in 2006, lung cancer is the leading cause of cancer death in Germany (29,000 men and 12,000 women), and with estimated 32,500 new cases in men and 14,600 new cases in women (about 47,000 overall), the second most frequent cancer site, respectively (http://www.canceratlas.de, GEKID 2010). The rates are declining in men since decades, but are still increasing in women (Fig. 1). These trends reflect long-term downwards trends of smoking prevalence in men, but increasing smoking habits in women (Heuer and Becker 1999).
https://static-content.springer.com/image/art%3A10.1007%2Fs00432-012-1228-9/MediaObjects/432_2012_1228_Fig1_HTML.gif
Fig. 1

Incidence of and mortality from lung cancer in Germany among men and women. Incidence data (dotted lines) are taken from the Federal State Cancer Registry of Saarland (http://www.krebsregister-saarland.de) and mortality data (straight lines) from the German Cancer Atlas (http://www.canceratlas.de). Rates are standardized to World standard

Indicated by the close levels of incidence and mortality, only 10–15 % of those diagnosed with lung cancer do not die from the disease, although the 5-year survival is dependent upon stage at diagnosis. Since the disease remains asymptomatic for a long time, most of the clinically diagnosed lung cancers are detected in the late stage IV (Fig. 2), which implies a dismal prognosis. Patients diagnosed with a stage Ia disease have a 5-year survival of up to 70 %. Thus, early detection of lung cancer theoretically should improve the outcome of the disease.
https://static-content.springer.com/image/art%3A10.1007%2Fs00432-012-1228-9/MediaObjects/432_2012_1228_Fig2_HTML.gif
Fig. 2

Stage distribution at the time of clinical lung cancer diagnosis (n = 2,265). Data were made available from the Federal State Cancer Registry of the Saarland

Ten to fifteen years ago, evidence has accrued from uncontrolled studies in Japan and the United States that low-dose multislice-CT (MSCT) of the lung is capable of detecting approximately four times as many small stage 1 lung cancers as chest X-rays, and that these appear likely to have a good prognosis. However, before being performed population-wide, the new approach needed a careful examination for several reasons: (a) it is unclear, whether the technique detects selectively only those cancers with a good prognosis leaving those with a bad prognosis occurring at about the same time in their natural history as at present and thus leading to a little overall benefit; (b) a longer survival generally observed with early detection can be caused by the advanced diagnosis and not be a prolonged lifetime as expected from effective screening; (c) limited specificity of the MSCT test indicates to a large proportion of undesired side effects (particularly invasive diagnostic procedures), which had to be quantified carefully. Thus, a real benefit in terms of mortality reduction has to be demonstrated scientifically and potential harms be identified before MSCT can be recommended as screening option for lung cancer.

Sample size considerations showed that even risk-adapted screening confined to heavy smokers required large cohorts of at least 28,000 participants to confirm a 20 % reduction in lung cancer mortality within a reasonable time frame (van Iersel et al. 2007). Thus, large multicentre trials, one in Europe and two in the USA, have been set up to quantify the efficacy of MSCT screening for lung cancer (see, e.g., van Klaveren et al. 2001). The European initiative comprises running studies from the Netherlands and Belgium (NELSON; 15,422 participants; van Klaveren et al. 2009), Denmark (DLCST; 4,104 participants; Pedersen et al. 2009), Italy (DANTE; 2,472 participants; Infante et al. 2008; MILD; 3,997 participants; Pastorino 2006; ITALUNG; 3,206 participants; Lopez Pegna et al. 2009), and Germany (LUSI; 4,052 participants), and a one-screen trial in preparation (Baldwin et al. 2011). From the two US trials (NLST and I-ELCAP), the first one comprising about 53,000 participants is a randomized trial whose first results have recently been published (NLST Research Team 2011) and the second one a non-randomized trial (I-ELCAP Investigators 2006). A major difference between the randomized studies is that MSCT screening in the intervention arm is compared with no intervention in the European trials and with chest X-ray in the control arm in the US trials.

In the present article, the design of the German component of the European initiative will be outlined, the outcome of the first screening round reported, and the reasons addressed why the European studies are continued despite the availability of the main result of a significant lung cancer mortality reduction by MSCT screening in the US randomized trial.

Materials and methods

Basic study design

The German Lung Cancer Screening Intervention trial (LUSI) is an epidemiological study among 50–69-year-old men and women with a history of heavy smoking randomized into a screening intervention arm comprising a MSCT at the time of randomization and 4 subsequent annual MSCTs, and a control arm with no intervention. A quit-smoking counseling was offered to both arms during the recruitment period, and all participants were asked for giving blood for future biomarker research (22.5 ml in the screening arm and 10 ml in the control arm). The recruitment procedure was population-based in a study area mainly approximately 20–25 km around Heidelberg, but partially up to 70 km. Follow-up is being conducted actively by annual questionnaire mailing and passively by repeated linkage to the local population registers and cancer registries. Scope was to contribute data of about 4,000 subjects (2,000 randomized in the MSCT screening arm and 2,000 in the “usual care” control arm) to the European study in order to confirm an at least 20 % reduction in lung cancer mortality by MSCT screening. The study was approved by the local ethics committees (073/2001) and the radiation protection authority (22462/2 2006-045).

Population-based recruitment

Study areas

In order to establish the cohort of overall 4,000 subjects with the specified conditions, a population-based recruitment was conducted in a local study area comprising the towns of Heidelberg, Mannheim, Ludwigshafen, all communities of the adjacent Rhine-Neckar-county and selected communities of the Neckar-Odenwald county. Identification of eligible subjects was carried out since September 2007 in several waves.

For this, the responsible population registers were asked for the selection of the total men and women population in the age range 50–69 years in which a pre-inquiry on basic socio-demographic characteristics, smoking habits, and relevant medical conditions were performed. In advance of data transfer, the authorities of all 88 involved communities were approached with a written request for agreement to data transmission for this dedicated purpose, and all provided written consent (based on a research regulation in the population register edict).

The data transfer was carried out immediately before mailing in order to maintain actuality of data. The mailing was done in weekly batches of 6,000 letters containing a personal letter with an invitation to attend this inquiry, a description of the purpose of the inquiry and the subsequent screening trial, a short self-administered questionnaire and a postpaid return envelope. Printing and mailing of the postal material were performed by a printing company.

The subjects were asked to return the filled-in questionnaire to the DKFZ within 2 weeks. The inquiry started in Mannheim, continued in Heidelberg and the Rhine-Neckar County, and ended in Ludwigshafen and two communities of the Necker-Odenwald County. In Mannheim and Heidelberg, a second wave of inquiry was conducted at the end of the recruitment period with an updated selection of the population register in order to further increase the number of eligible subjects. The entire process was effectuated with interruptions due to breaks in funding. At start or restart of the inquiry in the respective areas, a mass media information was released, which explained the purpose and relevance of the investigation that was mostly followed by additional journalist-initiated reports on lung cancer and the backgrounds of the project. In parallel, the physicians of the area, especially the pulmonologists, were directly contacted by letter and informed about the purpose and the design of the study, and they were asked to support the trial in case of involvement.

Population questionnaire

The questionnaire was designed with a system, which allowed for precise formatting and—after having been filled-in—the automatic scanner-based reading and data transmission into the study database. It comprised questions on basic socio-demographic characteristics (e.g., date of birth, place of birth, gender, ethnic origin, marital status, highest educational, and highest training qualification), smoking habits (e.g., smoking status (never, ever, and ex), start of smoking, average daily amount of smoking, periods of interruption of smoking, and date of smoking cessation), and relevant medical conditions (cancer diagnoses, any lung conditions, other serious illnesses, recent diagnostic procedures involving exposure to ionizing radiation). Finally, it comprised a question whether the subject would be interested to join the main screening trial if invited.

The filled-in questionnaire was scanned, and the accuracy of scanning and the completeness of data and its transfer into the study database checked by staff of the study. In case of irregularities, manual corrections were made, or the participants were approached by telephone (overall about 6,500) in order to complete missing information or clarify obvious inconsistencies. An anonymized version of the original questionnaire was scanned to an image, which was stored in the study archive for potential enquiries in case of later occurring data inconsistencies. Afterward, the paper questionnaires were destroyed and the data of the population registers deleted for those subjects which were not included in the main screening trial (see below).

Eligible subjects and recruitment invitation

Risk-adapted determination of eligibility was computed by the evaluation of the smoking-related information in the population questionnaire based on the procedures outlined in van Iersel et al. (2007). Eligibility was defined in the German study according to “option B” in that publication, that is, at least 25 years smoking of at least 15 cigarettes per day, or at least 30 years smoking of at least 10 cigarettes per day including ex-smokers who stopped smoking at most 10 years ago.

Further eligibility criteria were no cancer diagnosis within the past five years, no medical circumstances preventing surgical treatment in case of a lung cancer diagnosis in screening, and no serious illness shortening life expectancy below 10 years.

Subjects satisfying all eligibility criteria and having answered the question about interest at joining the main screening trial with “yes” were invited by letter with a proposed date of visit into the LUSI study center in the German Cancer Research Center (DKFZ). The invitation contained a comprehensive description of the study design as well as of the personal implication of being randomized, together with the potential benefits and harms of being assigned to either the screening intervention or the “usual care” control arm. The envelope also included a written consent form to be brought with into the study center. About 60–80 eligible subjects were invited about 2–3 weeks in advance of the proposed date of visit to four recruiting days per week in order to get about 30–40 weekly recruited participants. The others claimed for an alternative date of visit, and some refused participation. Since many of the eligible subjects were in an age of occupational activity, the time window for recruitment visits was between 2 pm and 6 pm.

Randomization, blood sampling, and MSCT screening

Randomization started at October 23, 2007 and ended at April 11, 2011. During that time, the participants arriving in DKFZ were escorted into the LUSI study center and asked to fill-in a short questionnaire initializing trial-specific information. The questionnaire intended essentially to update relevant information about medical conditions affecting eligibility, since temporal distance between the population inquiry and the invitation to the randomized main trial was partially up to 1 year. The study physicians checked this recent information and the potential appearance of so far unknown exclusion criteria and explained the conduct of the study together with potential benefits and harms at participation. Afterward, electronic randomization was carried out using the randomization tool RANDI developed in the Biostatistic Branch of the DKFZ (Hoffmann and Pilz 2005). The block randomization was stratified by age (<60 years vs. ≥60 years of age), gender, and smoking status (current vs former smoker). A quit-smoking counseling was offered to all participants of both study arms, and blood samples were also taken from all participants of both arms. The questionnaire was electronically processed as described above for the population questionnaire.

Blood protocol

Peripheral blood was collected in S-Monovetten (Sarstedt) that allow a one-step separation of blood cells. Among subjects in the screening arm, serum was obtained from one 7.5 ml S-Monovette and plasma and mononuclear cells from two 7.5 ml S-Monovetten with EDTA as anticoagulant. Among subjects in the control arm, serum was obtained from one 4.9 ml S-Monovette and plasma and mononuclear cells from one 4.9 ml S-Monovette with EDTA as anticoagulant. Fractionation of blood samples was aimed to be performed within 2 h after drawing, that is, mostly at late afternoon and early evening. On the average, 6–8 aliquots (à 500 μl) of serum and 12–16 aliquots of plasma, lymphocytes, and erythrocytes were obtained and frozen at −80 °C.

Quit-smoking counseling

Quit-smoking counseling was offered by two psychologists who were especially trained for this purpose in the WHO collaboration center for tobacco control located in the DKFZ. The counseling was offered in a separate room within the DKFZ temporally parallel to the recruitment hours. Interested participants were escorted by study staff to this room before or after randomization or screening where an approximately 20-min personal counseling took place followed by at least one subsequent telephone contact. Details about the precise procedure and results will be published elsewhere.

MSCT screening

Those participants having been randomized into the screening intervention arm were immediately escorted to the Radiology Department of the DKFZ. Here, subjects were asked whether they agreed to attend an additional pulmonary check within a separate study that will also be described in detail elsewhere. Then, the MSCT scan was performed by trained staff of the Radiology Department and passed either to the quit-smoking counseling or to the DKFZ portal.

For MSCT scanning, first a Toshiba 16 row scanner was used, and from 2010 on, a Siemens 128 row scanner running low-dose CTs with 1.6–2 mSv per scan maximally. Slice thickness was 1 mm and reconstruction interval 0.8 mm and 0.7 mm, respectively. The MSCT data were evaluated by especially trained radiologists mostly within 10 days according to the algorithm described below, and the results were reported in writing to the respective participants. Initial nodule evaluation was done with 2D image representation and subsequent evaluation with 3D representation using CAD (MEDIAN) with volumetric software.

Care was taken that reports on suspicious findings were sent out at week days, which allowed the respective screenee to contact a physician of choice to get further advice. In case of negative findings, subjects were informed and alluded to the next screening round in 12 months.

MSCT evaluation algorithm

The MSCT evaluation algorithm follows largely that of Henschke et al. (1999) and specifies interventions according to nodule size and nodule growth (Table 1). Initial MSCTs with no nodules or nodules below 5 mm in diameter were considered negative implying continuation of routine screening after 12 months. All others were classified as suspicious implying early recall depending upon size of the largest observed nodule: re-invitation with repeat MSCT after 6 months in case of a largest nodule of 5–7 mm in diameter, after 3 months in case of a largest nodule of 8–10 mm in diameter, and immediate recall with nodules larger than 10 mm in diameter. The radiologists were free to deviate from this algorithm in case of morphological characteristics of the nodules, which allowed for a more specific decision. This would be the case, for example in typical inflammatory infiltrates, pleural lymph nodes, or subpleural scars in the lung apex or dorsal recessus, which would be deemed benign. Nodules less than 10 mm in diameter might likewise be recalled earlier or immediately if they showed features highly suggestive of malignancy, like a stellance appearance, semi-solid pattern, and central lucency.
Table 1

MSCT evaluation algorithm

MSCT screening

Early recall

Outcome

Action

Outcome

growth of nodule

Action

Without abnormality or nodules <5 mm

Back to routine screening (12 months)

Nodules 5–7 mm

Early recall (6 months)

>600 VDT

400–600 VDT

∅ < 7.5 mm

∅ ≥ 7.5 mm–10 mm

≤400 VDT or

∅ > 10 mm

Back to routine

6-months early recall

3-months early recall

 

 

Immediate recall

Nodules 8–10 mm

Early recall (3 months)

Nodules >10 mm/malignity high malignity low

Immediate recall

Early recall (3 months)

Malignant

Non-malignant

Treatment

Back to routine screening

The repeat MSCTs of the early recalls were used to check the presence or disappearance of the incriminated nodule, and in case of presence to quantify potential growth in terms of volume doubling time (VDT). Disappearance of nodules or VDT above 600 days was considered negative and reported accordingly to the participants implying filing in routine 12 months screening.

VDT between 400 and 600 days was set into the early recall status in 6 or 3 months according to nodule diameter less than 7.5 or 7.5–10 mm, and VDT below 400 days and all MSCTs with nodules exceeding 10 mm in diameter was immediately recalled.

Immediate recalls led to the recommendation to the participants of individual assessment by an office-based pulmonologist. This could imply antibiotic treatment and CT control after several weeks, PET imaging, or immediate biopsy. Due to the organization of the German health care system based on the principle of free choice of physician by the patient and private office-based medical care, the respective participants could not be sent directly to the collaborating Thoraxklinik, but were free to seek their physician of choice. Due to the pre-information of the physicians of the study area (see above), the workup and the information on the results of the workup performed rather well, and most of the respective participants arrived at the Thoraxklinik Heidelberg.

Clinical workup and treatment for malignancies followed the respective guidelines and were not affected by the trialists.

Database

A SQL database was developed, which allowed for data protection-adjusted storage of the population data for the population-based recruitment, the storage of the obtained information of the filled-in questionnaires, and the computational determination of inclusion into the smoking-based high risk group as well as satisfying the other eligibility criteria. Based on this result, the database supported the preparation of the specific invitation of the subset of eligible subjects to the randomized trial and the control of the time table for the visits of the participants in the German Cancer Research Center. It also included the radiological information of the MSCT screening scans and the computational assistance of preparing the respective medical reports to the screenees of the intervention arm and the results of the annual active follow-up inquiries.

Statistical methods

The data of the first screening round were presented descriptively in the form of absolute figures and percentages. Tests on difference between screening and control arm were based on the chi-square test. All evaluations were done with the statistical analysis system SAS (SAS Institute, Inc., Cary, North Carolina) using the procedure FREQ.

Results

Almost 300,000 questionnaires were sent out from September 2007–December 2010 from which 95,797 were filled-in and sent back to the DKFZ (Table 2). The average compliance was 33 % with a range from 26 % (Mannheim) to 38 % (Rhine-Neckar County). Overall, 4,913 subjects (1.7 % of the approached 292,440 persons) answered the questionnaire and fulfilled the inclusion criteria (willingness to attend the screening trial, belonging to the smoking-related high risk group, and matching the specified medical profile). At the end, 4,052 (1.4 %) joined the screening trial.
Table 2

Study areas, periods of recruitment, numbers of population questionnaires sent out and returned, eligiblea and participating subjects, and overall yield of population-based recruitment

Study area

Period of recruitment

Number of questionnaires

Number of subjects matching the inclusion criteria (%)

Number of definitively recruited subjects (overall yield of recruitment in %)

Number of subjects matching the inclusion criteria (%)

Number of subjects matching the inclusion criteria (%)

Mannheim

2007–2008

2009–2010

79,423

20,261

(26 %)

930

(1.2 %)

847

(1.1 %)

Heidelberg

2008–2009

2010

30,699

11,217

(37 %)

525

(1.7 %)

444

(1.4 %)

Rhine-Neckar-County

2008–2010

145,271

54,933

(38 %)

2,891

(2.0 %)

2,380

(1.6 %)

Ludwigshafen

2010–2011

35,047

8,670

(25 %)

534

(1.5 %)

358

(1.0 %)

Mosbach/Osterburken

2010

2,000

676

(34 %)

33

(1.7 %)

23

(1.2 %)

Total

2007–2011

292,440

95,757

(33 %)

4,913

(1.7 %)

4,052

(1.4 %)

aEligibility criteria for inclusion into the study were as follows: (a) willingness to attend the screening study and (b) fulfilling specific medical conditions

From these 4,052 participants, 2,029 (50.1 %) were randomized into the MSCT screening arm and 2,023 (49.9 %) into the control arm (Table 3). About two-thirds (2,622) were men and one-third (1,430) was women. More than 46 % of the participants were in the age of 50–54 years, about 26 % in the age group 55–59 years, and only 17 % (age group 60–64) and 11 % (age group 65–69) in the upper age groups. The high p-values for test on difference for the randomization characteristics presented in the first column indicate that the randomization was well balanced.
Table 3

Sociodemographic characteristics of the study participants of the German lung cancer screening study LUSI

Characteristic (p-value for test on difference between study arms)

Intervention arm (% of characteristic)

Control arm (% of characteristic)

Total (% of entire cohort)

Total number

 (0.925)

2,029 (50.1)

2,023 (49.9)

4,052 (100)

Gender

 Males (0.876)

1,315 (50.1)

1,307 (49.9)

2,622 (64.7)

 Females (0.958)

714 (49.9)

716 (50.1)

1,430 (35.3)

Age

 50–54 (0.817)

942 (50.3)

932 (49.7)

1,874 (46.3)

 55–59 (0.757)

518 (49.5)

528 (50.5)

1,046 (25.8)

 60–64 (0.909)

344 (50.2)

341 (49.8)

685 (16.9)

 65–69 (0.887)

225 (50.3)

222 (49.7)

447 (11.0)

Smoking status

 Current smoker (0.811)

1,259 (50.2)

1,247 (49.8)

2,506 (61.9)

 Ex-smoker (0.896)

770 (49.8)

775 (50.2)

1,545 (38.1)

From the 2,029 screenees, 1,488 (73.4 %) were screened negative implying invitation to routine screening round 2 one year later (“back to routine”, Table 4). For one participant, the CT could not be evaluated due to numerous hemangiomas. 540 screenees (26.6 %) were screened positive with largest suspicious nodules of size 5–7 mm in 393 subjects, 8–10 mm in 78 subjects, and >10 mm in 69 subjects. Among these, 72 subjects, 7 and 11, respectively, were sent back into the routine scheme by radiology assessment of the suspicious nodule, while the others were assigned to an early recall after 6 or 3 months, respectively. All suspicious nodules were double-read by a senior radiologist. The table describes in detail the results of the early recalls leading either to a “back to routine” decision or a confirmed lung cancer diagnosis. Overall, 22 lung cancers were detected in the first screening round, four in the largest suspicious nodule group 5–7 mm, one in the 8–10 mm group, and 17 in the >10 mm group.
Table 4

Results of the first screening round of the German lung cancer screening trial LUSI

Size of largest nodule (mm) and suggested mode of action

Outcome

1. Screening rounda

1st early recall

Outcome in terms of volume doubling time (VDT) and suggested mode of action

2nd early recall

Outcome according to the type of recall, suggested mode of action, and definite outcome

VDT > 600

600 ≥ VDT > 400

400 ≥ VDT or

∅ > 10 mm

6-months

2nd recall

3-months

2nd recall

∅ < 7.5 mm

∅ ≥ 7.5 mm

0 to <5 mm

1,488

      

 Back to routine

1,487

      

 Immediate recall

1

Back to routine

n = 1

     

5–7 mm

393

      

 Back to routine

72

      

 Recall after 6 months

313

Non-compliant

n = 6

Back to routine

n = 287

Early recall after 6 months

n = 6

Early recall after 3 months

n = 7

Immediate recall

n = 7

Back to routine

n = 4

LC: n = 3

n = 6

Back to routine

n = 6 (all)

n = 7

Back to routine

n = 6

LC: n = 1

 Recall after 3 months

6

 

Back to routine

n = 6

    

 Immediate recall

n = 2

Back to routine

n = 2

      

8–10 mm

78

      

 Back to routine

7

      

 Recall after 6 months

9

Non-compliant

n = 1

Back to routine

n = 7

 

Early recall after 3 months

n = 1

  

n = 1

Back to routine

n = 1 (all)

 Recall after 3 months

57

Back to routine

n = 51

Early recall after 6 months

n = 1

Early recall after 3 months

n = 3

Immediate recall

n = 2

Back to routine

n = 2 (all)

n = 1

Back to routine

n = 1 (all)

n = 3

Back to routine

n = 2

Immediate recall and then back to routine

n = 1

 Immediate recall

5

Back to routine

n = 3

Non-compliant

n = 1

LC: n = 1

      

>10 mm

69

      

 Back to routine

11

      

 Recall after 6 months

8

Back to routine

n = 7

  

Immediate recall

n = 1

Back to routine

n = 1 (all)

  

 Recall after 3 months

5

Back to routine

n = 5 (all)

     

 Immediate recall

n = 45

Back to routine

n = 24

Non-compliant

n = 5 (1 LC in 2nd screening round)

LC: 16

      

aOne screenee was excluded after MSCT since the scan could not be evaluated due to numerous hemangiomas

Overall, 12 participants did not comply with the early recall invitation after a suspicious nodule, six in the 5–7 mm group, two in the 8–10 mm group, and five in >10 mm group. From the six in the 5–7 mm group, two attended the next screening round, one omitted the 2nd but attended the 3rd round (all “back to routine”), two have until now not attended any further screening round, and one died. From the two non-complaints in the 8–10 mm group, one did not attend the early recall but attended the next screening round (“back to routine”), and the other having not attended the immediate recall recommendation did also not attend the further screening rounds but continues to fill-in the questionnaire. In the >10 mm group, one non-compliant omitted the 2nd screening round, attended the 3rd, where the suspicious nodule had disappeared. Two attended the next screening round with the suspicious nodule unchanged. One attended the 2nd and 3rd screening rounds, where the suspicious nodule continued to grow. Despite immediate recalls on each occasion, the participant has so far not responded. One participant with a suspicious nodule above 10 mm diameter did not attend the immediate recall recommendation, but attended the 2nd screening round, complied there with a further immediate recall due to nodule growth, and a lung cancer was diagnosed, which we counted thus as being detected in the 1st screening round. In the table, the participant was counted as “non-compliant,” but the lung cancer is notified in parentheses.

Among the 540 screenees with suspicious nodules, 165 (31 %) had a solitary nodule, 189 (35 %) had two to four nodules, 145 (27 %) five to nine, and 41 (7 %) had 10 or more nodules. The number of suspicious nodules appeared unrelated to the category of the largest nodule (data not shown).

Blood samples were obtained from almost all participants from the screening arm (98.5 %), but only about 85 % of those of the control arm (Table 5). The early recall rate was 26.6 %, and the immediate recall rate 2.6 %. The lung cancer detection rate was 1.1 %. Biopsies were taken from 31 screenees from which 9 turned out to be benign and 22 malignant, implying a benign: malignant ratio for the biopsy of roughly 2:5. The biopsies which led to the 22 histologically confirmed diagnoses of lung cancer were obtained by 3 bronchoscopies, 8 VATS, and 11 thoracotomies. The VATS and thoracotomies were led over to immediate surgical treatment. The other 3 patients received chemotherapy and radiation. Among the nine biopsies for benign conditions, five were thoracotomies, three were video-assisted thoracoscopies, and one was a bronchoscopy.
Table 5

Performance indicators for the first screening round in the German lung cancer screening study LUSI

Characteristic

Number/proportion (%)

Assigned to the MSCT/control arm

2,029 (50.1)/2,023 (49.9)

Blood samples taken: MSCT/control arm

1,998 (98.5)/1,731 (85.6) (p < 0.0001)

Early recalls in the MSCT arm: total/immediate

540/53

Number of biopsies

31

Lung cancers (MSCT arm)

22

Early recall rate (total)

26.6 % (24.7–28.6 %)

Immediate recall rate

2.6 % (2.0–3.4 %)

Detection rate

1.1 % (0.7–1.6 %)

Positive predictive value for immediate recalls

34.6 (22.0–49.1 %)

Interval cancers

1

Benign: malignant biopsy ratio

9:22 (≈1:2.5)

From the 22 screen-detected lung cancers, 15 were adenocarcinoma, three squamous cell carcinoma, two carcinoids, one a small-cell lung cancer, and one a mixed small-cell lung cancer combined with adenocarcinoma (Table 6). Eighteen were in clinical stage I at the time of screen detection, one in stage II, and three in stage III.
Table 6

Characteristics of the lung cancers detected in the first screening round of the German lung cancer screening trial LUSI and the interval cancer after the first screening round

#

Date of CT

Date of diag.

Histology

ICD-O3

TNM/Grade

Stage

Topography

1

23-10-07

10-01-08

Adeno-Ca

8255/3

pT1pN0cM0/G1

IA

UL le

2

10-01-08

15-02-08

Adeno-Ca

8255/3

pT2pN0cM0/G3

IB

UL le

3

15-01-08

24-04-09

Adeno-Ca

8250/3

pT1pN0cM0/G2

IA

UL le

4

01-04-08

04-06-08

Adeno-Ca

8250/3

pT2pN0cM0/G1

IB

UL ri

5

17-04-08

29-05-08

Adeno-Ca

8480/3

pT1pN0cM0/G1

IA

LL ri

6

16-03-09

27-05-09

Squamous cell

8083/3

pT2pN1cM0/G3

IIB

UL ri

7

08-04-09

07-12-09

Adeno-Ca

8255/3

pT1pN0cM0/G3

IA

UL ri

8

13-05-09

02-07-09

Adeno-Ca

8250/3

pT2pN0cM0/G2

IB

LL le

9

14-05-09

30-06-10

Adeno-Ca

8255/3

pT1pN0cM0/G2

IA

LL le

10

26-05-09

02-07-09

Adeno-Ca

8255/3

pT2pN0cM0/G3

IB

UL le

11

07-07-09

30-07-09

Small cell

8041/3

cT1cN2cM0/G2

IIIA

LL ri

12

16-07-09

12-07-11

Adeno-Ca

8310/3

pT2pN0cM0/G3

IB

UL ri

13

27-07-09

14-10-10

Adeno-Ca

8288/3

pT1apN0cM0/G3

IA

UL ri

14

28-07-09

23-03-10

Adeno-Ca

8255/3

pT2pN0cM0/G2

IB

UL le

15

07-09-09

02-11-09

Adeno-Ca

8255/3

pT2pN0cM0/G2

IB

UL ri

16

03-05-10

11-01-11

Typical carcinoid

8240/3

pT1apN0cM0

IA

UL le

17

28-07-10

12-08-10

Squamous cell

8078/3

T3N2M0/G2

IIIA

LL ri

18

02-08-10

22-10-10

Adeno-Ca

8260/3

pT1apN0cM0/G2

IA

LL le

19

07-12-10

19-08-11

Adeno-Ca

8550/3

pT1apN0cM0/G1

IA

ML ri

20

14-12-10

18-01-11

Typical carcinoid

8240/3

pT1apN0cM0

IA

LL le

21

10-02-11

28-02-11

Combined Small cell and Adeno-Ca

8045/3

T3N2M0/G3

IIIA

LL ri

22

17-02-11

18-05-11

Squamous cell

8072/3

pT2apN0cM0/G2

IB

LL ri

Interval Ca

17-11-08

01-07-09

Adeno-Ca

8550/3

pT2N0pM1/G2

IV

UL le

In the interval to the second screening round, one interval cancer was detected. The relevant nodule was detected in screening, but classified according to the study protocol as insignificant. Surgery outside the protocol confirmed a stage IV adenocarcinoma (Table 6). Although the nodule was less than 1 cm in diameter, the carcinoma was stage IV, because it had invaded the visceral pleura causing a local deposit on the corresponding parietal pleura, that is, pleural carcinosis.

Six radiologists evaluated the initial MSCTs performing 894, 81, 804, 49, 89, and 111 evaluations, respectively, from which all but the second were also involved in the subsequent rounds (data not shown).

Discussion

In terms of key characteristics of screening, detection rate, and false-positive rate, the profile of the German lung cancer screening trial LUSI appears consistent to that of comparable studies.

The detection rate of 1.1 % lies within the range of the other European trials, which reported rates between 0.8 % (DLCST and MILD) and 2.2 % (DANTE).

The false-positive rate indicates potential harm to screenees when they are confronted with the finding of a suspicious nodule and the suggestion of an early recall some months later implying waiting time and anxiety. It derives from the rate of suspicious findings minus the detection rate of true lung cancers. The rate of suspicious findings reported here corresponds to that found in previous studies (Diederich et al. 2004) and some other European trials (e.g., Lopez Pegna et al. 2009). It was however much higher than in the DLCST (7.9 % Pedersen et al. 2009) or MILD trial (5.9 %, Pastorino 2006). Discrepancies in this quantity may occur by deviating definitions of when nodules appear “suspicious.” If this definition is confined to conditions in which the screenee was told that a suspicious nodule may be cancer and must be worked up appropriately, it corresponds to our rate of immediate recalls of 3.4 %. In our view, all MSCT findings leading to additional diagnostic procedures may cause harm to screenees by anxiety, additional radiation, or even biopsy or surgery, and should be included into the false-positive rate. In our study, it was about 25 % in this first screening round, which appears much too high for a routine application of this screening modality.

Another important characteristic of a screening tool is the histological pattern of the cancers detected by screening. In our study, the histological composition of the tumors detected by screening is substantially different from that found in symptomatic cases (Fig. 3) and agrees also to the reports from other CT lung screening studies. In LUSI, the proportion of adenocarcinoma was about 80 %, similar to ELCAP, where it was about 76 % (I-ELCAP Investigators 2006), but more than in the DANTE trial, where it was 36 % (Infante et al. 2008). Thus, the cancers identified by MSCT screening were different from those causing high mortality, which are the actual target of screening. This shift in histology may be a limiting factor for efficacy of MSCT screening in terms of mortality reduction.
https://static-content.springer.com/image/art%3A10.1007%2Fs00432-012-1228-9/MediaObjects/432_2012_1228_Fig3_HTML.gif
Fig. 3

Distribution of histology of a clinically and b screen-detected lung cancers. Source of data: a Federal State Cancer Registry of the Saarland; b LUSI, first screening round (see Table 6)

As desired in screening, most of the detected cancers were in an early stage (more than 80 %) and only one in stage IV. This is also in line with the other MSCT screening trials, but in strong contrast to the clinical setting showing a large majority of symptomatic stage IV cancers. The first screening round is a prevalence round collecting all cancers including those which would cause clinical symptoms very soon and may therefore include quite a number of advanced stages, in contrast to what we found. It may be argued that the sojourn time of lung cancer in a preclinical phase may be so short that only few advanced cases may be detected in screening. Alternatively, the high proportion of early-stage cancers may be interpreted as the indicative of overdiagnosis of cancers, which would never have become clinically apparent, had they not been found by screening.

A major issue for the present study as for all other European studies is whether they be continued after the first results of the US trial have been published showing a 25 % lung cancer mortality reduction. In a trialists conference held in 2011 in Pisa, Italy, the European initiative launched a statement (“PISA position statement,” see, e.g., van Klaveren 2011) in favor of continuation of the trials for several reasons.

As addressed in the introduction section and above, previous investigations and first own figures give rise to concern about overdiagnosis and overtreatment introduced by CT screening (see, e.g., Bach et al. 2007). Overdiagnosis can be quantified in randomized trials in which the control arm is based to “usual care”, that is, no specific screening intervention in the present situation. The control arm provides then the lung cancer incidence in an unscreened population, while the incidence screening rounds (screening after the first initial round) provide the rate of lung cancer under CT screening whose ratio indicates the level of overdiagnosis. Since the US trial compared CT screening with chest X-ray, in principle, both study arms could induce overdiagnosis and hence prevent a proper quantification of the overall effect (for overdiagnosis by chest X-ray see Bach et al. 2003). In contrast, the European trials do compare CT screening with no screening and allow thus for a comprehensive assessment of overdiagnosis by the screening intervention.

Further unresolved questions regard, for example appropriate screening intervals and how to characterize high risk groups. In the European studies, variations exist in screening intervals ranging from annual screening (DANTE, DLCST, ITALUNG, and LUSI) to 2 (partially MILD) and 2½ years (partially NELSON). It will be highly important to evaluate the effect on detection rate and false-positive rate in order to specify adequate screening intervals. Correspondingly, the range of different smoking habits, that is, lung cancer risks, within the studies will allow for quantifying how lung cancer risk of screenees affects detection rates and false-positive rates. Such results will help to optimize screening modalities in terms of efficacy and economy. These latter issues can be evaluated in collaboration with the NLST, provided the European trials have been developed to a level ready for analysis. In the Pisa meeting, a first interim analysis was envisaged for the year 2012.

Acknowledgments

Numerous colleagues contributed to the success of the study. The MSCTs were carried out by Jessica Engelhart and Martina Jochim. The database was set up and managed (2007–2011) by Simon Roether. Data entry, invitation of participants, and time schedule coordination by Angelika Bari and Andrea Albrecht, quit-smoking counseling by Monika Bade and Vera Bähr, processing and storage of blood specimen by Kirsten Lenner-Fertig and Ulrike von Seydlitz-Kurzbach. The study was supported in 2007–2010 by the German Research Foundation (BE 2486/2-1) and the Dietmar-Hopp-Stiftung, and is currently (2010–2013) supported by the German Research Foundation (BE 2486/2-2), and the members of the German Center for Lung Research by the German Research Ministry (BMBF-DZL).

Conflict of interest

None.

Copyright information

© Springer-Verlag 2012