Jump to comment:

• Response to “Prospective randomised controlled trial: fixed 1-year screening interval group versus a tailored intervals group,” letter response by Silva et al
  Anton Schreuder
  Published on: 21 June 2018
• Prospective randomised controlled trial: fixed 1-year screening interval group versus a tailored intervals group
  Mario Silva
  Published on: 21 June 2018

• Published on: 21 June 2018

  Response to “Prospective randomised controlled trial: fixed 1-year screening interval group versus a tailored intervals group,” letter response by Silva et al

  • Anton Schreuder, PhD student Radboudumc

  We thank the authors of the letter in response to our paper for their interest and positive appraisal of our model. Likewise, we appreciate the design of the Multicenter Italian Lung Detection (MILD) trial which, despite its small sample size, demonstrates that annual intervals are unnecessary for the majority of screenees. Once more European data is available to perform cost-effectiveness analyses, we hypothesize that personalised screening intervals will prove to be the preferred design. Furthermore, it is estimated that most inclusion criteria used to select high-risk participants encompass only 70% of all lung cancer cases in the population; reassessing risk and tailoring interval groups after the baseline scan may enable the inclusion of persons of lower risk. As Silva et al mentioned, there is no reason to set the upper limit of follow-up intervals at 2-years. We also agree that volumetric nodule measurements are better suited for determining follow-up procedures than (perpendicular) diameter, and hope to be able to implement this into a future model. Moreover, risk scores may be calculated autonomously by computers in the future, with only a select few dubious cases requiring radiologist attention.

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 21 June 2018

  Prospective randomised controlled trial: fixed 1-year screening interval group versus a tailored intervals group

  • Mario Silva, Radiologist University of Parma

  In 2011, the National Lung Cancer Screening Trial (NLST) showed that annual low-dose computed tomography (LDCT) improved overall survival (1). More recently, longer interval between LDCT rounds was advocated to improve screening efficiency after baseline (2).
  Schreuder et al reported a comprehensive model for optimization of LDCT by biennial rounds for subjects at lower 2-year risk of lung cancer (3). They built a promising polynomial model including both patient characteristics and nodule descriptors. The retrospective simulation on NLST data provided enough power to test Schreuder’s model (3) as well as other models for selection of subjects to be forwarded to biennial screening (2, 4). We appreciate this approach to parsimonious LDCT administration as we are strongly convinced that annual screening should be tailored to subjects with remarkably high risk of lung cancer. The authors refer that prospective randomized controlled trial with tailored screening intervals would be hardly feasible, however we would like to remind that some experience was already reported in the literature.
  Since 2005, the Multicenter Italian Lung Detection (MILD) trial conducted a prospective comparison between annual (LDCT1 = 1,152 screenees) and biennial LDCT (LDCT2 = 1,151 screenees) (5). The LDCT2 screenees were shifted to annual screening in case of solid nodule > 60 mm^3 and/or subsolid nodules. In other words, the MILD trial prospectively tested a risk model for tailored s...

  Show More

  In 2011, the National Lung Cancer Screening Trial (NLST) showed that annual low-dose computed tomography (LDCT) improved overall survival (1). More recently, longer interval between LDCT rounds was advocated to improve screening efficiency after baseline (2).
  Schreuder et al reported a comprehensive model for optimization of LDCT by biennial rounds for subjects at lower 2-year risk of lung cancer (3). They built a promising polynomial model including both patient characteristics and nodule descriptors. The retrospective simulation on NLST data provided enough power to test Schreuder’s model (3) as well as other models for selection of subjects to be forwarded to biennial screening (2, 4). We appreciate this approach to parsimonious LDCT administration as we are strongly convinced that annual screening should be tailored to subjects with remarkably high risk of lung cancer. The authors refer that prospective randomized controlled trial with tailored screening intervals would be hardly feasible, however we would like to remind that some experience was already reported in the literature.
  Since 2005, the Multicenter Italian Lung Detection (MILD) trial conducted a prospective comparison between annual (LDCT1 = 1,152 screenees) and biennial LDCT (LDCT2 = 1,151 screenees) (5). The LDCT2 screenees were shifted to annual screening in case of solid nodule > 60 mm^3 and/or subsolid nodules. In other words, the MILD trial prospectively tested a risk model for tailored screening intervals that was based on baseline LDCT findings. Given the small sample of this trial, we underline that the absence of outcome difference between LDCT1 and LDCT2 should be carefully interpreted (5).
  In summary, baseline LDCT2 prospectively selected 147/1,151 (12.8%) screenees for annual screening, whilst 1,004/1,151 (87.2%) were forwarded to biennial round (6). The LDCT2 risk model allowed a sharp reduction by almost 90% LDCTs at the first incidence round. This is a considerable proportion compared to the retrospective analysis presented by Schreuder et al (simulations led to potential LDCT reduction ranging from 10.4% to 81.6%) (3). A similar figure was seen every other year throughout 7.3 years, with overall cuts of 32% LDCTs in LDCT2 compared with LDCT1 (6).
  We would like to emphasize that the volumetric risk model allowed higher detection rate in LDCT2 screenees selected for annual screening (1.36% namely 2 lung cancers in 147 screenees) compared to LDCT1 screenees (0.45% namely 5 lung cancers in 1,111 screenees), at the first annual LDCT after baseline. Under this prospective condition, the number of scans needed to diagnose one lung cancer case (NND) was 74 in LDCT2, which is quite close to the 60 NND proposed by Schreuder by accepting a delay of diagnosis in 25% of screenees. On the other hand, NND was 222.2 in LDCT1, namely within the NND range reported in the literature (from 125 to 385) at the first annual LDCT by fixed annual screening algorithms (1, 7, 8). Hence, the LDCT2 algorithm prompted the lowest prospective NND reported in the literature. Through the median 7.3 years of LDCT screening in MILD, the cumulative NND during annual screenings was 63 for selected LDCT2 screenees and 250 for LDCT1 (p=0.003). Selective 3-year screening interval is now being tested in the bioMILD trial with higher volumetric threshold for solid nodules and including circulating biomarkers (ClinicalTrials.gov: NCT02247453; >4,000 baseline LDCTs acquired, results expected by 2020).
  We hypothesize that the prospective application of Schreuder’s model might even improve the efficiency of MILD LDCT2. Furthermore, enrichment by clinical variables is encouraged, including circulating biomarkers. More models like Schreuder’s are fostered for optimization of radiologists’ workload for the seemingly approaching practice of population-based lung cancer screening by LDCT.

  References
  1. National Lung Screening Trial Research T, Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. The New England journal of medicine. 2011;365(5):395-409.
  2. Patz EF, Jr., Greco E, Gatsonis C, Pinsky P, Kramer BS, Aberle DR. Lung cancer incidence and mortality in National Lung Screening Trial participants who underwent low-dose CT prevalence screening: a retrospective cohort analysis of a randomised, multicentre, diagnostic screening trial. The Lancet Oncology. 2016;17(5):590-9.
  3. Schreuder A, Schaefer-Prokop CM, Scholten ET, Jacobs C, Prokop M, van Ginneken B. Lung cancer risk to personalise annual and biennial follow-up computed tomography screening. Thorax. 2018.
  4. White CS, Dharaiya E, Campbell E, Boroczky L. The Vancouver Lung Cancer Risk Prediction Model: Assessment by Using a Subset of the National Lung Screening Trial Cohort. Radiology. 2017;283(1):264-72.
  5. Pastorino U, Rossi M, Rosato V, Marchiano A, Sverzellati N, Morosi C, et al. Annual or biennial CT screening versus observation in heavy smokers: 5-year results of the MILD trial. European journal of cancer prevention : the official journal of the European Cancer Prevention Organisation. 2012;21(3):308-15.
  6. Sverzellati N, Silva M, Calareso G, Galeone C, Marchiano A, Sestini S, et al. Low-dose computed tomography for lung cancer screening: comparison of performance between annual and biennial screen. European radiology. 2016;26(11):3821-9.
  7. Yousaf-Khan U, van der Aalst C, de Jong PA, Heuvelmans M, Scholten E, Walter J, et al. Risk stratification based on screening history: the NELSON lung cancer screening study. Thorax. 2017.
  8. Tammemagi MC, Schmidt H, Martel S, McWilliams A, Goffin JR, Johnston MR, et al. Participant selection for lung cancer screening by risk modelling (the Pan-Canadian Early Detection of Lung Cancer [PanCan] study): a single-arm, prospective study. The Lancet Oncology. 2017.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top