Natural History Of Lung Cancer Research Articles

Cost Utility Analysis of a Pilot Study for the Korean Lung Cancer Screening Project.

PurposeThe aim of this study was to evaluate the cost utility of a pilot study of Korean Lung Cancer Screening Project.Materials and MethodsWe constructed a Markov model consisting of 26 states based on the natural history of lung cancer according to the Surveillance, Epidemiology, and End Results summary stage (localized, regional, distant). In the base case, people aged 55–74 years were under consideration for annual screening. Costs and quality-adjusted life years were simulated to calculate the incremental cost utility ratio. Sensitivity analyses were performed on the uncertainty associated with screening target ages, stage distribution, cost, utility, mortality, screening duration, and discount rate.ResultsThe base case (US$25,383 per quality-adjusted life year gained) was cost-effective compared to the scenario of no screening and acceptable considering a willingness-to-pay threshold of US$27,000 per quality-adjusted life years gained. In terms of the target age of screening, the age between 60 and 74 years was the most cost-effective. Lung cancer screening was still cost-effective in the sensitivity analyses on the cost for treatment, utility, mortality, screening duration, and less than 5% discount rates, although the result was sensitive to a rise in positive rates or variation of stage distribution.ConclusionOur results showed the cost-effectiveness of annual low-dose computed tomography screening for lung cancer in high-risk populations.

Health and economic impact at a population level of both primary and secondary preventive lung cancer interventions: A model-based cost-effectiveness analysis

ObjectivesRobust economic evaluations are needed to identify efficient strategies for lung cancer prevention that combine brief and intensive smoking cessation intervention programmes with screening using low-dose computed tomography (LDCT) at different ages, frequencies, and coverages. We aimed to assess the cost-effectiveness of smoking cessation approaches combined with lung cancer screening in the European context at a population level from a societal perspective. Materials and MethodsA microsimulation model that describes the natural history of lung cancer and incorporates several prevention strategies was developed. Discounted lifetime QALYs and costs at a rate of 3% were used to calculate incremental cost-effectiveness ratios, defined as additional costs in 2017 Euros per QALY gained. ResultsSmoking cessation interventions reduce the incidence of lung cancer by 8%-46% and are consistently more effective and cost-effective when starting at younger ages. Screening reduces lung cancer mortality by 1%-24% and is generally less effective and more costly than smoking cessation interventions. The most cost-effective strategy would be to implement intensive smoking cessation interventions at ages 35, 40 and 45, combined with screening every three years between the ages of 55 and 65. ConclusionsCombining smoking cessation interventions with LDCT screening is a very attractive prevention strategy that substantially diminishes the burden of lung cancer. These combined prevention strategies, especially when providing several intensive interventions for smoking cessation at early ages, are more cost-effective than both approaches separately and allow for a more intensified LDCT without losing efficiency.

High incidence of concurrent disease states detected during mobile lung cancer screening in an underserved population.

6540 Background: Studies such as the National Lung Screening Trial (NLST; N Engl J Med 2011;365:395-409) have shown a survival benefit to low-dose Lung CT screening in high-risk smokers. Levine Cancer Institute (LCI) initiated the first mobile low dose computerized tomographic (LDCT) lung screening program for underserved populations in 2017. In addition to being able to intervene early in the natural history of lung cancer, the project has also shown a previously unreported high incidence of incidental diseases in this population. We characterize these findings in 1198 patients. Methods: From May 2017, subjects with criteria eligible for NLST screening were identified and underwent LDCT for lung cancer detection. Patients screened in the program were all uninsured or underinsured, mean age 60.8 years, 18% were African American, 3% Latin-x and 78% were rural with an overall 47.1 mean pack-year smoking history. These patients were screened using a novel mobile LDCT (J Clin Oncol 37, 2019 suppl; abstr 6567) created for this program. By December 2020, 1198 patients completed their first screening. All CT scans were reviewed by two separate radiologists and were reviewed for quality assurance by a separate expert multidisciplinary team. Results: Of the 1198 subjects, 84% (1006 subjects) were found by LDCT to have a least one incidental disease. More than half of the subjects (645, 53.8%) had coronary atherosclerosis. Of those, 25% (183) were described to have at least moderate disease with 8% (96) described as severe. Overall, 42% (504) were found to have emphysema and 25% (299) had vascular atherosclerotic disease; 1.8% (22) of those screened had a detected aortic aneurysm. In total, thirty separate disease findings were found (listed from fourth to tenth most common finding: degenerative spine changes [205], cholelithiasis [59], hiatal hernia [52], pericardial effusions [38], fatty liver [32], kidney stone [3]), and cardiomegaly [30]). 3.5% (42) were found to have an undiscovered breast, adrenal, liver or kidney mass that required further workup. Conclusions: The number of incidental findings in our mainly rural underserved subject group was very high (84%). 35.5% of patients in the National Lung Screening Trial died from heart disease or respiratory disease. These numbers have not been overtly discussed and our study confirms the number of concerning incidental diseases that can lead to morbidity or mortality. In this high-risk, underserved population of heavy smokers, the opportunity for positive impact on other disease states can be increased by a mobile lung cancer screening program by increasing access to care.

Bayesian versus Empirical Calibration of Microsimulation Models: A Comparative Analysis.

Calibration of a microsimulation model (MSM) is a challenging but crucial step for the development of a valid model. Numerous calibration methods for MSMs have been suggested in the literature, most of which are usually adjusted to the specific needs of the model and based on subjective criteria for the selection of optimal parameter values. This article compares 2 general approaches for calibrating MSMs used in medical decision making, a Bayesian and an empirical approach. We use as a tool the MIcrosimulation Lung Cancer (MILC) model, a streamlined, continuous-time, dynamic MSM that describes the natural history of lung cancer and predicts individual trajectories accounting for age, sex, and smoking habits. We apply both methods to calibrate MILC to observed lung cancer incidence rates from the Surveillance, Epidemiology and End Results (SEER) database. We compare the results from the 2 methods in terms of the resulting parameter distributions, model predictions, and efficiency. Although the empirical method proves more practical, producing similar results with smaller computational effort, the Bayesian method resulted in a calibrated model that produced more accurate outputs for rare events and is based on a well-defined theoretical framework for the evaluation and interpretation of the calibration outcomes. A combination of the 2 approaches is an alternative worth considering for calibrating complex predictive models, such as microsimulation models.

PCN248 LUNG CANCER SCREENING BY LOW-DOSE COMPUTERISED TOMOGRAPHY: A COST EFFECTIVENESS ANALYSIS OF ALTERNATIVE PROGRAMMES IN THE UNITED KINGDOM USING A NEWLY DEVELOPED NATURAL HISTORY BASED ECONOMIC MODEL

In a systematic review of economic evaluations for lung cancer screening there were no economic models of the UK setting based on disease natural history. This is a well-recognised modelling approach with advantages in modelling population screening, able to explore the cost-effectiveness of multiple alternative potential programme formats. ENaBL is an individual patient simulation model constructed to test lung cancer screening programme designs. It was based on the observed natural history of lung cancer and calibrated against data from the US National Lung Screening Trial. Costs were taken from the UK Lung Cancer Screening Trial. Participants were simulated current or former smokers aged between 55 and 80 years at a higher risk of lung cancer relative to the general population, from which twelve distinct population subgroups were identified. Each were examined in single, triple, annual and biennial arrangements of screening number and interval length, creating 48 distinct screening strategies. These were compared to no screening. The outcome was incremental cost-effectiveness of strategies. Low-dose CT screening is predicted to be effective in bringing forward stage distribution at diagnosis and at reducing lung cancer mortality with decreases ranging from 4.2% to 7.7% relative to no screening, depending on screen frequency. Savings in the cost of treatment from earlier detection are outweighed by costs associated with over-diagnosis, and overall healthcare costs are predicted to increase. Single screen programmes targeted at people with ≥3% predicted lung cancer risk in the age ranges 55 to 75 years, and 60 to 75 years produced ICERs of £28,784 and £28,169 per QALY gained, respectively. Lung cancer screening may not be cost-effective, depending on the cost-effectiveness threshold used. Thresholds of £20,000 to £30,000 per QALY are commonly used in the UK, and screening may be cost-effective with the higher threshold.

Smoking and Lung Cancer Mortality in the United States From 2015 to 2065: A Comparative Modeling Approach.

Tobacco control efforts implemented in the United States since the 1960s have led to considerable reductions in smoking and smoking-related diseases, including lung cancer. To project reductions in tobacco use and lung cancer mortality from 2015 to 2065 due to existing tobacco control efforts. Comparative modeling approach using 4 simulation models of the natural history of lung cancer that explicitly relate temporal smoking patterns to lung cancer rates. U.S. population, 1964 to 2065. Adults aged 30 to 84 years. Models were developed using U.S. data on smoking (1964 to 2015) and lung cancer mortality (1969 to 2010). Each model projected lung cancer mortality by smoking status under the assumption that current decreases in smoking would continue into the future (status quo trends). Sensitivity analyses examined optimistic and pessimistic scenarios. Under the assumption of continued decreases in smoking, age-adjusted lung cancer mortality was projected to decrease by 79% between 2015 and 2065. Concomitantly, and despite the expected growth, aging, and longer life expectancy of the U.S. population, the annual number of lung cancer deaths was projected to decrease from 135000 to 50000 (63% reduction). However, 4.4 million deaths from lung cancer are still projected to occur in the United States from 2015 to 2065, with about 20 million adults aged 30 to 84 years continuing to smoke in 2065. Projections assumed no changes to tobacco control efforts in the future and did not explicitly consider the potential effect of lung cancer screening. Tobacco control efforts implemented since the 1960s will continue to reduce lung cancer rates well into the next half-century. Additional prevention and cessation efforts will be required to sustain and expand these gains to further reduce the lung cancer burden in the United States. National Cancer Institute.

Abstract 4228: Incidence of pericardial and pleural effusions in patients with small cell lung cancer treated with third-line or later-line therapy: An analysis of electronic health records data

Abstract Small cell lung cancer (SCLC) is an aggressive, rapidly fatal disease. Pericardial and pleural effusions are, to a large extent, part of the natural history of lung cancer and the occurrences in SCLC may have differed due to novel therapies. We estimated the incidence of treatment-emergent pericardial and pleural effusions in SCLC patients treated with third-line or later-line therapy using Optum's electronic health records (EHR) data. The database consists of de-identified EHR data from a network of healthcare provider organizations in the United States. Natural language processing (NLP) was used to identify SCLC. Of 8,291 patients with newly diagnosed SCLC between 01/01/2008 and 09/30/2016 who met our inclusion criteria, 428 patients received 3rd line of therapy (LOT). 206 LOTs were 4th or later lines. Pericardial and pleural effusions were identified by a combination of ICD-9/ICD-10, procedures, or terms from NLP structured data. At start of 3rd LOT, median age was 65, 46% were male, 97% had extensive disease (excluding n=52 with missing staging). The incidence rate was 1.58 (95% CI, 1.17-2.07) per 1,000 person-days for pericardial effusion and 4.40 (95% CI, 3.67-5.25) for pleural effusion in patients with 3rd LOT while on-treatment (Table 1). The incidence rate was 1.15 (95% CI, 0.88-1.49) for pericardial effusion and 3.68 (95% CI, 3.12-4.32) for pleural effusion in patients with 3rd LOT while on-treatment and off-treatment. Patients treated with 4th or later lines had higher incidence of pericardial and pleural effusions compared to patients treated with 3rd LOT. The incidence proportions of the effusions (Table 1) should be interpreted with care, as the patients had varying length of follow-up. In conclusion, this analysis is the first, to our knowledge, to provide population-based estimates of incidence for pericardial and pleural effusions in SCLC patients treated with third- or later-line therapy. Table 1. Incidence of pericardial and pleural effusionsIncidence Proportion % (95% CI)Incidence Rate per 1,000 Person-Days (95% CI)On-treatment* 3rd LOT (n=428)Pericardial (n=51)11.9% (9.00-15.4%)1.58 (1.17-2.07)Pleural (n=125)29.2% (24.9-33.8%)4.40 (3.67-5.25)4th+ LOT(n=206)Pericardial (n=32)15.5% (10.9-21.2%)2.14 (1.46-3.01)Pleural (n=66)32.0% (25.7-38.9%)5.49 (4.25-6.99)On-treatment and off-treatment**3rd LOT(n=428)Pericardial (n=59)13.8% (10.7-17.4%)1.15 (0.88-1.49)Pleural (n=152)35.5% (31.0-40.3%)3.68 (3.12-4.32)4th+ LOT(n=206)Pericardial (n=37)18.0% (13.0-23.9%)1.49 (1.05-2.05)Pleural (n=80)38.8% (32.1-45.9%)4.14 (3.28-5.15)*from start of LOT to earlier of (30 days after end of LOT or start of next LOT); **from start of LOT to the day prior to start of next LOT or earlier of (365 days from end of LOT or loss to follow-up date), if there is no subsequent LOT; for those with the event, time to first occurrence was used Citation Format: Rui Jiang, Scott H. Gulbranson, Mondira Bhattacharya, Fabio A. Lievano, Philip Komarnitsky, Martina M. Koch, Ravi Potluri, Jerzy E. Tyczynski. Incidence of pericardial and pleural effusions in patients with small cell lung cancer treated with third-line or later-line therapy: An analysis of electronic health records data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4228.

PC02.02 Pro Chemotherapy

PC02.02 Pro Chemotherapy

Prise en charge des métastases cérébrales des carcinomes bronchiques

Brain metastases are common in the natural history of lung cancer and are correlated with poor prognosis. Over the past ten years, the treatment of brain metastases has evolved considerably ; a better understanding of the evolution and molecular biology of these processes has enabled the development of new treatments specifically targeting the Epidermal Growth Factor Receptor (EGFR) and the Anaplastic Lymphoma Kinase (ALK). Meanwhile, advances in imaging techniques have enabled early diagnosis of these tumors and improved assessment of the tumoral response. The repeated use of radiosurgery (SRS) or hippocampal-avoidance whole brain radiotherapy have allowed high tumor control rates and reduced radiation-induced cerebral injury. We summarized the treatment options and perspectives in the management of cerebral secondary lesions of lung cancer.

MILC: A Microsimulation Model of the Natural History of Lung Cancer

The Microsimulation Lung Cancer (MILC) model was developed to simulate individual trajectories and predict outcomes of lung cancer for populations. The model describes the natural history of lung cancer from a disease-free state to death. Predictions of individual trajectories depend on a set of covariates including age, sex, and smoking behaviors. The module presented here is designed as part of a comprehensive decision-making toolkit for evaluating lung cancer prevention, screening and treatment policies. The MILC package implements the model in the open-source statistical software R. This paper introduces the main components, simulation algorithm, and specifics of the MILC model, validates it by reproducing observed lung cancer incidence trends in the US population, and uses it to make plausible predictions for 50-year-old men and women with a range of smoking histories.

Lung cancer detectability by test, histology, stage, and gender: estimates from the NLST and the PLCO trials.

Implementing optimal lung cancer screening programs requires knowledge of the natural history and detectability of lung cancer. This information can be derived from the results of clinical trials with the aid of microsimulation models. Data from the Surveillance, Epidemiology, and End Results (SEER) program and individual-level data from the National Lung Screening Trial (NLST) and the Prostate, Lung, Colon, and Ovarian Cancer Screening trial (PLCO) were used to investigate the sensitivity (by histology and stage) of CT and chest radiography (CXR) and the mean preclinical sojourn time (MPST) of lung cancer (by gender, histology, and stage). The MISCAN-Lung model was used to reproduce the lung cancer incidence by method of detection (clinically or screen-detected), gender, histology, and stage in both trials and SEER, by calibrating CT and CXR sensitivity and natural history parameters. CT sensitivity ranges from 8.83% to 99.35% and CXR sensitivity from 2.51% to 97.31%, depending on histology and stage. CT sensitivity for stage IA is more than 3-fold higher compared with CXR, for all histologies. The total MPST estimates for lung cancer progressing through preclinical stages IA to IV range from 3.09 to 5.32 years for men and 3.35 to 6.01 years for women. The largest difference in total MPST between genders was estimated for adenocarcinoma. We estimate longer MPSTs for lung cancer compared with previous research, suggesting a greater window of opportunity for lung cancer screening. This study provides detailed insights into the natural history of lung cancer and CT screening effectiveness.

Detection of skeletal muscle metastases on initial staging of lung cancer: a retrospective case series

Estimation of skeletal muscle metastases (SMMs) at the time of diagnosis and/or initial staging of lung cancer. Retrospective evaluation of clinical charts and imaging data suggestive of SMMs of patients with histology-proved lung cancer over a 5-year period. SMMs were identified in 46 out of 1,754 patients. Single and multiple (62.9% of cases) SMMs were detected by total body multi-detector computed tomography (MDCT). They were associated with poorly differentiated (43%) and advanced adenocarcinomas (52%) without clinically relevant symptoms and/or signs. Psoas and buttock muscles were most frequently involved (33.3%). MDCT findings consisted of well-defined homogeneously hyperdense oval masses (31%), lesions with ring-like enhancement and central hypoattenuation (68%), or large abscess-like necrotic lesions (24%). Sonography revealed well-defined hypoechoic masses (41.6%), ill-defined hypoechoic lesions (33.3%), or anechoic areas with a necrotic centre (25%). Positron emission tomography revealed that all SMMs were metabolically active. SMMs are uncommon but not negligible in lung cancer, with an estimated prevalence of 2.62% in our series. Although histology remains the recommended method, use of high-performance imaging techniques and increased clinical suspicion may improve their early detection. Efforts addressing their effect on the natural history of lung cancer are needed.

Chapter 13: CISNET lung models: comparison of model assumptions and model structures.

Sophisticated modeling techniques can be powerful tools to help us understand the effects of cancer control interventions on population trends in cancer incidence and mortality. Readers of journal articles are, however, rarely supplied with modeling details. Six modeling groups collaborated as part of the National Cancer Institute's Cancer Intervention and Surveillance Modeling Network (CISNET) to investigate the contribution of U.S. tobacco-control efforts toward reducing lung cancer deaths over the period 1975-2000. The six models included in this monograph were developed independently and use distinct, complementary approaches toward modeling the natural history of lung cancer. The models used the same data for inputs, and agreed on the design of the analysis and the outcome measures. This article highlights aspects of the models that are most relevant to similarities of or differences between the results. Structured comparisons can increase the transparency of these complex models.

Revolution in Lung Cancer: New Challenges for the Surgical Pathologist

Traditionally, lung cancer has been viewed as an aggressive, relentlessly progressive disease with few treatment options and poor survival. The traditional role of the pathologist has been primarily to differentiate small cell carcinoma from non-small cell carcinoma on biopsy and cytology specimens and to stage non-small cell carcinomas that underwent resection. In recent years, our concepts of lung cancer have undergone a revolution, including (1) the advent of successful, new, molecular-targeted therapies for lung cancer, many of which are associated with specific histologic cell types and subtypes; (2) new observations on the natural history of lung cancer derived from ongoing high-resolution computed tomography screening studies and recent histologic findings; and (3) proposals to revise the classification of lung cancers, particularly adenocarcinomas, in part because of the first 2 developments. To summarize the important, new developments in lung cancer, emphasizing the role of the surgical pathologist in personalized care for patients with lung cancer. Information about the new developments in lung cancer was obtained from the peer-review medical literature and the authors' experiences. For decades, we have perceived lung cancer as a relentlessly aggressive and mostly incurable disease for which the surgical pathologist had a limited role. Today, surgical pathologists have an important and expanding role in the diagnosis and treatment of lung cancer, and it is essential to keep informed of new advances.

Revolution in Lung Cancer: New Challenges for the Surgical Pathologist

Abstract Context—Traditionally, lung cancer has been viewed as an aggressive, relentlessly progressive disease with few treatment options and poor survival. The traditional role of the pathologist has been primarily to differentiate small cell carcinoma from non–small cell carcinoma on biopsy and cytology specimens and to stage non–small cell carcinomas that underwent resection. In recent years, our concepts of lung cancer have undergone a revolution, including (1) the advent of successful, new, molecular-targeted therapies for lung cancer, many of which are associated with specific histologic cell types and subtypes; (2) new observations on the natural history of lung cancer derived from ongoing high-resolution computed tomography screening studies and recent histologic findings; and (3) proposals to revise the classification of lung cancers, particularly adenocarcinomas, in part because of the first 2 developments. Objective—To summarize the important, new developments in lung cancer, emphasizing the ro...

Abstract CN12-03: The inflammation-EMT-cancer initiating cell axis in the pathogenesis of lung cancer: Breaking ground in the pursuit of targeted chemoprevention

Abstract CN12-03: The inflammation-EMT-cancer initiating cell axis in the pathogenesis of lung cancer: Breaking ground in the pursuit of targeted chemoprevention

Lung cancer growth dynamics

Mathematical modeling furnishes valuable and otherwise unattainable insights, some counterintuitive, into the natural history of lung cancer. We modeled lung cancer growth dynamics to show that: (1) early diagnosis of lethal lung cancer by means of radiographic or CT screening is an unattainable goal. (2) At a given dimension and constant tumor volume doubling time, the rate of diameter increase is an exponential function (base 1.26) of the number of tumor volume doublings, and the rate of increase in volume, a quadratic function of the radius. (3) This methodology delineates the magnitude of diameter increase in small nodules required to discern volume growth. (4) Under the assumption of a high degree of compliance with sequential screens in large-scale trials, mean tumor volume doubling time can be estimated from the prevalence:incidence ratio. For example, in the Mayo Clinic computerized tomography trial, stage IA tumor volume doubling time was 230 days.

Abstract 857: A preliminary study of the correlation of tumor size and epidemiological characteristics in patients with non-small cell lung cancer

Abstract Introduction: Tumor size at diagnosis is an indicator for prognosis and the type of treatment of lung cancer. It reflects tumor growth rate and predicts metastasis status and survival in patients. It is therefore of great importance to evaluate what factors are associated with tumor size at diagnosis. Experimental procedures: We reviewed the electronic medical records of 1190 lung cancer patients from M.D. Anderson Cancer Center to collect information on tumor characteristics (lung cancer radiologically or surgically assessed tumor size) and patient characteristics (including age, gender, ethnicity, smoking history and vital status at the last contact date). We used t-test, Kruskal-Wallis test and linear regression to explore the relationship between the tumor size at diagnosis and a number of epidemiological factors derived from questionnaire data. Results: Smoking status was associated with tumor size at the time of diagnosis. We observed that never smokers presented with significantly smaller tumors (2.83 cm) than smokers, whether former smokers (3.50cm; P=0.001), recent quitters (4.00cm; P<0.001) or current smokers (3.55cm; P=0.004). There was no significant difference (global P=0.061 from Kruskal-Wallis test) by tumor size in the three smoker groups. No significant difference was observed between younger (age≤65; 3.70 cm) and older (age>65; 3.30 cm; P=0.071) patients. Men had larger tumors (4.00 cm) than women (3.00 cm; P<0.001) at presentation. The gender-stratified analysis did not show evidence (P>0.05) of relationship between either height or body mass index (BMI) and tumor size. Patients reporting dust or asbestos exposure presented with larger tumors while exposure to X-ray examination showed an opposite result. Hormone replacement therapy use in women was not associated with lung tumor size. Patients self-reporting with emphysema had smaller tumors (3.00 cm) at presentation than those without emphysema (3.70 cm; P=0.001). There was a difference in tumor size by cell type (adenocarcinoma, 3.42 cm vs squamous cell carcinoma, 4.52 cm, P<0.01) and degree of differentiation (not differentiated to poorly, 4.33 cm vs moderately to well, 3.50 cm, P<0.01). African Americans (4.10 cm) had significantly larger tumors (P<0.001) at diagnosis compared to Caucasian patients (3.50 cm), although the sample size for African Americans was limited to 127. Health care system utilization patterns did not explain the observed associations. Conclusions: In this study, lung tumor size at presentation was significantly associated with gender, ethnicity, smoking status, emphysema, cell type, degree of differentiation, X-ray examinations, and exposures to dust and asbestos. Mathematical modeling that allows for different growth and metastatic rate of tumors may help further understand the impact of various factors on the natural history of lung cancer. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 857.

Modeling the transition of lung cancer from early to advanced stage

We present a stochastic parametric model of the natural history of lung cancer that predicts the primary tumor volume at the moment the disease transits from early to advanced stage. Our model also produces estimates for the probability of symptomatic detection as a function of tumor volume and clinical stage. We estimate model parameters by likelihood maximization using data from the Mayo Lung Project (MLP), which was a clinical trial that evaluated screening for lung cancer in the 1970s. Mayo Lung Project cancer cases reported in Stage III or greater, according to the 1979 AJCC staging for lung cancer, were considered advanced stage. Our estimator distinguishes between the cases detected because of clinical symptoms and cases detected by screening. For nonsmall cell lung cancer cases detected in MLP, we estimate that the median primary tumor diameter at the onset of advanced stage disease was 4.1 cm. In addition, we estimate that the rate of patients symptomatically detected with their disease increases as their primary tumor increases in size, and for patients with a primary tumor of a given size, the rate of symptomatic detection is 12.8 times greater among patients with advanced stage disease compared to patients with early stage disease.

Calibration of Disease Simulation Model Using an Engineering Approach

ObjectivesCalibrating a disease simulation model's outputs to existing clinical data is vital to generate confidence in the model's predictive ability. Calibration involves two challenges: 1) defining a total goodness-of-fit (GOF) score for multiple targets if simultaneous fitting is required, and 2) searching for the optimal parameter set that minimizes the total GOF score (i.e., yields the best fit). To address these two prominent challenges, we have applied an engineering approach to calibrate a microsimulation model, the Lung Cancer Policy Model (LCPM). MethodsFirst, 11 targets derived from clinical and epidemiologic data were combined into a total GOF score by a weighted-sum approach, accounting for the user-defined relative importance of the calibration targets. Second, two automated parameter search algorithms, simulated annealing (SA) and genetic algorithm (GA), were independently applied to a simultaneous search of 28 natural history parameters to minimize the total GOF score. Algorithm performance metrics were defined for speed and model fit. ResultsBoth search algorithms obtained total GOF scores below 95 within 1000 search iterations. Our results show that SA outperformed GA in locating a lower GOF. After calibrating our LCPM, the predicted natural history of lung cancer was consistent with other mathematical models of lung cancer development. ConclusionAn engineering-based calibration method was able to simultaneously fit LCPM output to multiple calibration targets, with the benefits of fast computational speed and reduced the need for human input and its potential bias.