Validation of Microsimulation Models against Alternative Model Predictions and Long-Term Colorectal Cancer Incidence and Mortality Outcomes of Randomized Controlled Trials

Benefit-Harm Analysis for Informed Decision Making on Participating in Colorectal Cancer Screening: A Modeling Study

OBJECTIVES

To facilitate informed decision making on participating in colorectal cancer (CRC) screening, we assessed the benefit-harm balance of CRC screening for a wide range of subgroups over different time horizons.

METHODS

The study combined incidence proportions of benefits and harms of (not) participating in CRC screening estimated by the Adenoma and Serrated pathway to CAncer microsimulation model, a preference eliciting survey, and benefit-harm balance modeling combining all outcomes to determine the net health benefit of CRC screening over 10, 20, and 30 years. Probability of net health benefit was estimated for 210 different subgroups based on age, sex, previous participation in CRC screening, and lifestyle.

RESULTS

CRC screening was net beneficial in 183 of 210 subgroups over 30 years (median probability [MP] of 0.79, interquartile range [IQR] of 0.69-0.85) across subgroups. Net health benefit was greater for men (MP 0.82; IQR 0.69-0.89) than women (MP 0.76; IQR 0.67-0.83) and for those without history of participation in previous screenings (MP 0.84; IQR 0.80-0.89) compared with those with (MP 0.69; IQR 0.59-0.75). Net health benefit decreased with increasing age, from MP of 0.84 (IQR 0.80-0.86) at age 55 to 0.61 (IQR 0.56-0.71) at age 75. Shorter time horizons led to lower benefit, with MP of 0.70 (IQR 0.62-0.80) over 20 years and 0.54 (IQR 0.48-0.67) over 10 years.

CONCLUSIONS

Our benefit-harm analysis provides information about net health benefit of screening participation, based on important characteristics and preferences of individuals, which could assist screening invitees in making informed decisions on screening participation.

The multitarget faecal immunochemical test for improving stool-based colorectal cancer screening programmes: a Dutch population-based, paired-design, intervention study

BACKGROUND

The faecal immunochemical test (FIT) is widely employed for colorectal cancer screening. However, its sensitivity for advanced precursor lesions remains suboptimal. The multitarget FIT (mtFIT), measuring haemoglobin, calprotectin, and serpin family F member 2, has demonstrated enhanced sensitivity for advanced neoplasia, especially advanced adenomas, at equal specificity to FIT. This study aimed to prospectively validate and investigate the clinical utlitity of mtFIT versus FIT in a setting of population-based colorectal cancer screening.

METHODS

Individuals aged 55-75 years and who were eligible for the Dutch national FIT-based colorectal cancer screening programme were invited to submit both a FIT and mtFIT sample collected from the same bowel movement. Positive FIT (47 μg/g haemoglobin cutoff) or mtFIT (based on decision-tree algorithm) led to a colonoscopy referral. The primary outcome was the relative detection rate of mtFIT versus FIT for all advanced neoplasia. Secondary outcomes were the relative detection rates of colorectal cancer, advanced adenoma, and advanced serrated polyps individually and the long-term effect of mtFIT-based versus FIT-based programmatic screening on colorectal cancer incidence, mortality, and cost, determined with microsimulation modelling. The study has been registered in ClinicalTrials.gov, NCT05314309, and is complete.

FINDINGS

Between March 25 and Dec 7, 2022, 35 786 individuals were invited to participate in the study, of whom 15 283 (42·7%) consented, and 13 187 (86·3%) of 15 283 provided both mtFIT and FIT samples with valid results. Of the 13 187 participants, 6637 (50·3%) were male and 6550 (49·7%) were female. mtFIT showed a 9·11% (95% CI 8·62-9·61) positivity rate and 2·27% (95% CI 2·02-2·54) detection rate for advanced neoplasia, compared with a positivity rate of 4·08% (3·75-4·43) and a detection rate of 1·21% (1·03-1·41) for FIT. Detection rates of mtFIT versus FIT were 0·20% (95% CI 0·13-0·29) versus 0·17% (0·11-0·27) for colorectal cancer; 1·64% (1·43-1·87) versus 0·86% (0·72-1·04) for advanced adenoma, and 0·43% (0·33-0·56) versus 0·17% (0·11-0·26) for advanced serrated polyps. Modelling demonstrated that mtFIT-based screening could reduce colorectal cancer incidence by 21% and associated mortality by 18% compared with the current Dutch colorectal cancer screening programme, at feasible costs. Furthermore, at equal positivity rates, mtFIT outperformed FIT in terms of diagnostic yield. At an equally low positivity rate, mtFIT-based screening was predicted to further decrease colorectal cancer incidence by 5% and associated mortality by 4% compared with FIT-based screening.

INTERPRETATION

The higher detection rate of mtFIT for advanced adenoma compared with FIT holds the potential to translate into additional and clinically meaningful long-term colorectal cancer incidence and associated mortality reductions in programmatic colorectal cancer screening.

FUNDING

Stand Up to Cancer, Dutch Cancer Society, Dutch Digestive Foundation, and Health~Holland.

SMARTERscreen protocol: a three-arm cluster randomised controlled trial of patient SMS messaging in general practice to increase participation in the Australian National Bowel Cancer Screening Program

BACKGROUND

Australia persistently has one of the highest rates of colorectal cancer (CRC) in the world. Australia's National Bowel Cancer Screening Program (NBCSP) sends a biennial Faecal Immunochemical Test (FIT)-the 'NBCSP kit'-to everyone eligible for the programme between 50 and 74 years old; however, participation in the programme is low, especially in the 50- to 60-year-old age group. Our previous efficacy trial ('SMARTscreen') demonstrated an absolute increase in uptake of 16.5% (95% confidence interval = 2.02-30.9%) for people sent an SMS with motivational and instructional videos, from their general practice prior to receiving their NBCSP kit, compared to those receiving usual care. Building on the strengths of the SMARTscreen trial and addressing limitations, the 'SMARTERscreen' trial will test the effect on participation in the NBCSP of sending either an SMS only or an SMS with online video material to general practice patients due to receive their NBCSP compared to 'usual care'.

METHODS

SMARTERscreen is a three-arm stratified cluster randomised controlled trial involving 63 general practices in two states in Australia. Eligible patients are patients who are aged 49-60 years and due to receive their NBCSP kit within the next 2 weeks during the intervention period. General practices will be equally randomised to three trial arms (21:21:21, estimated average 260 patients/practice). The two interventions include (i) an SMS with an encouraging message from their general practice or (ii) the same SMS with weblinks to additional motivational and instructional videos. The control arm will receive 'usual care'. Using the intention-to-treat approach, primary analysis will estimate the three pair-wise between-arm differences in the proportion of eligible patients who participate in the NBCSP within 6 months of when their kit is sent, utilising screening data from the Australian National Cancer Screening Register (NCSR). Patient intervention adherence to the interventions will also be evaluated. Findings will be incorporated into the Policy1-Bowel microsimulation model to estimate the long-term health benefits and cost-effectiveness of the interventions.

DISCUSSION

SMARTERscreen will provide high-level evidence determining whether an SMS or an SMS with web-based material sent to general practice patients prior to receiving their NBCSP kit increases participation in bowel cancer screening.

TRIAL REGISTRATION

Australian New Zealand Clinical Trials Registry ACTRN12623000036617. Registered on 13 January 2023. Trial URL: https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=385119&isClinicalTrial=False.

Calibration and Validation of the Colorectal Cancer and Adenoma Incidence and Mortality (CRC-AIM) Microsimulation Model Using Deep Neural Networks

OBJECTIVES

Machine learning (ML)-based emulators improve the calibration of decision-analytical models, but their performance in complex microsimulation models is yet to be determined.

METHODS

We demonstrated the use of an ML-based emulator with the Colorectal Cancer (CRC)-Adenoma Incidence and Mortality (CRC-AIM) model, which includes 23 unknown natural history input parameters to replicate the CRC epidemiology in the United States. We first generated 15,000 input combinations and ran the CRC-AIM model to evaluate CRC incidence, adenoma size distribution, and the percentage of small adenoma detected by colonoscopy. We then used this data set to train several ML algorithms, including deep neural network (DNN), random forest, and several gradient boosting variants (i.e., XGBoost, LightGBM, CatBoost) and compared their performance. We evaluated 10 million potential input combinations using the selected emulator and examined input combinations that best estimated observed calibration targets. Furthermore, we cross-validated outcomes generated by the CRC-AIM model with those made by CISNET models. The calibrated CRC-AIM model was externally validated using the United Kingdom Flexible Sigmoidoscopy Screening Trial (UKFSST).

RESULTS

The DNN with proper preprocessing outperformed other tested ML algorithms and successfully predicted all 8 outcomes for different input combinations. It took 473 s for the trained DNN to predict outcomes for 10 million inputs, which would have required 190 CPU-years without our DNN. The overall calibration process took 104 CPU-days, which included building the data set, training, selecting, and hyperparameter tuning of the ML algorithms. While 7 input combinations had acceptable fit to the targets, a combination that best fits all outcomes was selected as the best vector. Almost all of the predictions made by the best vector laid within those from the CISNET models, demonstrating CRC-AIM's cross-model validity. Similarly, CRC-AIM accurately predicted the hazard ratios of CRC incidence and mortality as reported by UKFSST, demonstrating its external validity. Examination of the impact of calibration targets suggested that the selection of the calibration target had a substantial impact on model outcomes in terms of life-year gains with screening.

CONCLUSIONS

Emulators such as a DNN that is meticulously selected and trained can substantially reduce the computational burden of calibrating complex microsimulation models.

HIGHLIGHTS

Calibrating a microsimulation model, a process to find unobservable parameters so that the model fits observed data, is computationally complex.We used a deep neural network model, a popular machine learning algorithm, to calibrate the Colorectal Cancer Adenoma Incidence and Mortality (CRC-AIM) model.We demonstrated that our approach provides an efficient and accurate method to significantly speed up calibration in microsimulation models.The calibration process successfully provided cross-model validation of CRC-AIM against 3 established CISNET models and also externally validated against a randomized controlled trial.

Updated cost-effectiveness analysis of lung cancer screening for Australia, capturing differences in the health economic impact of NELSON and NLST outcomes

BACKGROUND

A national, lung cancer screening programme is under consideration in Australia, and we assessed cost-effectiveness using updated data and assumptions.

METHODS

We estimated the cost-effectiveness of lung screening by applying screening parameters and outcomes from either the National Lung Screening Trial (NLST) or the NEderlands-Leuvens Longkanker Screenings ONderzoek (NELSON) to Australian data on lung cancer risk, mortality, health-system costs, and smoking trends using a deterministic, multi-cohort model. Incremental cost-effectiveness ratios (ICERs) were calculated for a lifetime horizon.

RESULTS

The ICER for lung screening compared to usual care in the NELSON-based scenario was AU$39,250 (95% CI $18,150-108,300) per quality-adjusted life year (QALY); lower than the NLST-based estimate (ICER = $76,300, 95% CI $41,750-236,500). In probabilistic sensitivity analyses, lung screening was cost-effective in 15%/60% of NELSON-like simulations, assuming a willingness-to-pay threshold of $30,000/$50,000 per QALY, respectively, compared to 0.5%/6.7% for the NLST. ICERs were most sensitive to assumptions regarding the screening-related lung cancer mortality benefit and duration of benefit over time. The cost of screening had a larger impact on ICERs than the cost of treatment, even after quadrupling the 2006-2016 healthcare costs of stage IV lung cancer.

DISCUSSION

Lung screening could be cost-effective in Australia, contingent on translating trial-like lung cancer mortality benefits to the clinic.

Guidance for setting international standards on reporting longitudinal adherence to stool-based colorectal cancer screening

Longitudinal adherence to colorectal cancer (CRC) screening is reported using different summarizing measures, which hampers international comparison. We provide evidence to guide recommendations on which longitudinal adherence measure to report. Using adherence data over four stool-based CRC screening rounds in three countries, we calculated six summarizing adherence measures; adherence over all rounds, adherence per round, rescreening, full programme adherence (yes/no), regularity (never/inconsistent/consistent screenees) and number of times participated. For each measure, we calculated the accuracy in capturing the observed adherence patterns. Using the ASCCA model, we predicted screening effectiveness when using summarizing measures as model input versus the observed adherence patterns. Adherence over all rounds in the Italian, Spanish and Dutch cohorts was 64.9%, 42.8% and 61.5%, respectively, and the proportion of consistent screenees was 50.9%, 26.3% and 45.7%. Number of times participated and regularity were most accurate and resulted in similar model-predicted screening effectiveness as simulating the observed adherence patterns of Italy, Spain and the Netherlands (mortality reductions: 24.4%, 16.9% and 23.5%). Adherence over all rounds and adherence per round were least accurate. Screening effectiveness was overestimated when using adherence over all rounds (mortality reductions: 26.8%, 19.4% and 25.7%) and adherence per round (mortality reductions: 26.8%, 19.5% and 25.9%). To conclude, number of times participated and regularity were most accurate and resulted in similar model-predicted screening effectiveness as using the observed adherence patterns. However they require longitudinal data. To facilitate international comparison of CRC screening programme performance, consensus on an accurate adherence measure to report should be reached.

Coupé VM. A progressive three-state model to estimate time to cancer: a likelihood-based approach

Background

To optimize colorectal cancer (CRC) screening and surveillance, information regarding the time-dependent risk of advanced adenomas (AA) to develop into CRC is crucial. However, since AA are removed after diagnosis, the time from AA to CRC cannot be observed in an ethically acceptable manner. We propose a statistical method to indirectly infer this time in a progressive three-state disease model using surveillance data.

Methods

Sixteen models were specified, with and without covariates. Parameters of the parametric time-to-event distributions from the adenoma-free state (AF) to AA and from AA to CRC were estimated simultaneously, by maximizing the likelihood function. Model performance was assessed via simulation. The methodology was applied to a random sample of 878 individuals from a Norwegian adenoma cohort.

Results

Estimates of the parameters of the time distributions are consistent and the 95% confidence intervals (CIs) have good coverage. For the Norwegian sample (AF: 78%, AA: 20%, CRC: 2%), a Weibull model for both transition times was selected as the final model based on information criteria. The mean time among those who have made the transition to CRC since AA onset within 50 years was estimated to be 4.80 years (95% CI: 0; 7.61). The 5-year and 10-year cumulative incidence of CRC from AA was 13.8% (95% CI: 7.8%;23.8%) and 15.4% (95% CI: 8.2%;34.0%), respectively.

Conclusions

The time-dependent risk from AA to CRC is crucial to explain differences in the outcomes of microsimulation models used for the optimization of CRC prevention. Our method allows for improving models by the inclusion of data-driven time distributions.

Supplementary Information

The online version contains supplementary material available at (10.1186/s12874-022-01645-2).

Colorectal Cancer Screening within Colonoscopy Capacity Constraints: Can FIT-Based Programs Save More Lives by Trading off More Sensitive Test Cutoffs against Longer Screening Intervals?

Introduction. Colorectal cancer (CRC) prevention programs using fecal immunochemical testing (FIT) in screening rely on colonoscopy for secondary and surveillance testing. Colonoscopy capacity is an important constraint. Some European programs lack sufficient capacity to provide optimal screening intensity regarding age ranges, intervals, and FIT cutoffs. It is currently unclear how to optimize programs within colonoscopy capacity constraints. Design. Microsimulation modeling, using the MISCAN-Colon model, was used to determine if more effective CRC screening programs can be identified within constrained colonoscopy capacity. A total of 525 strategies were modeled and compared, varying 3 key screening parameters: screening intervals, age ranges, and FIT cutoffs, including previously unevaluated 4- and 5-year screening intervals (using a lifetime horizon and 100% adherence). Results were compared with the policy decisions taken in Ireland to provide CRC screening within available colonoscopy capacity. Outcomes estimated net costs, quality-adjusted life-years (QALYs), and required colonoscopies. The optimal strategies within finite colonoscopy capacity constraints were identified. Results. Combining a reduced FIT cutoff of 10 µg Hb/g, an extended screening interval of 4 y and an age range of 60–72 y requires 6% fewer colonoscopies, reduces net costs by 23% while preventing 15% more CRC deaths and saving 16% more QALYs relative to a strategy (FIT 40 µg Hb/g, 2-yearly, 60–70 year) approximating current policy. Conclusion. Previously overlooked longer screening intervals may optimize cancer prevention with finite colonoscopy capacity constraints. Changes could save lives, reduce costs, and relieve colonoscopy capacity pressures. These findings are relevant to CRC screening programs across Europe that employ FIT-based testing, which face colonoscopy capacity constraints.

Prioritisation of colonoscopy services in colorectal cancer screening programmes to minimise impact of COVID-19 pandemic on predicted cancer burden: A comparative modelling study

Objectives

Colorectal cancer (CRC) screening with a faecal immunochemical test (FIT) has been disrupted in many countries during the COVID-19 pandemic. Performing catch-up of missed screens while maintaining regular screening services requires additional colonoscopy capacity that may not be available. This study aimed to compare strategies that clear the screening backlog using limited colonoscopy resources.

Methods

A range of strategies were simulated using four country-specific CRC natural-history models: Adenoma and Serrated pathway to Colorectal CAncer (ASCCA) and MIcrosimulation SCreening ANalysis for CRC (MISCAN-Colon) (both in the Netherlands), Policy1-Bowel (Australia) and OncoSim (Canada). Strategies assumed a 3-month screening disruption with varying recovery period lengths (6, 12, and 24 months) and varying FIT thresholds for diagnostic colonoscopy. Increasing the FIT threshold reduces the number of referrals to diagnostic colonoscopy. Outcomes for each strategy were colonoscopy demand and excess CRC-related deaths due to the disruption.

Results

Performing catch-up using the regular FIT threshold in 6, 12 and 24 months could prevent most excess CRC-related deaths, but required 50%, 25% and 12.5% additional colonoscopy demand, respectively. Without exceeding usual colonoscopy demand, up to 60% of excess CRC-related deaths can be prevented by increasing the FIT threshold for 12 or 24 months. Large increases in FIT threshold could lead to additional deaths rather than preventing them.

Conclusions

Clearing the screening backlog in 24 months could avert most excess CRC-related deaths due to a 3-month disruption but would require a small increase in colonoscopy demand. Increasing the FIT threshold slightly over 24 months could ease the pressure on colonoscopy resources.

Clinical Validation of a Multitarget Fecal Immunochemical Test for Colorectal Cancer Screening : A Diagnostic Test Accuracy Study

BACKGROUND

The fecal immunochemical test (FIT) is used in colorectal cancer (CRC) screening, yet it leaves room for improvement.

OBJECTIVE

To develop a multitarget FIT (mtFIT) with better diagnostic performance than FIT.

DESIGN

Diagnostic test accuracy study.

SETTING

Colonoscopy-controlled series.

PARTICIPANTS

Persons (n = 1284) from a screening (n = 1038) and referral (n = 246) population were classified by their most advanced lesion (CRC [n = 47], advanced adenoma [n = 135], advanced serrated polyp [n = 30], nonadvanced adenoma [n = 250], and nonadvanced serrated polyp [n = 53]), along with control participants (n = 769).

MEASUREMENTS

Antibody-based assays were developed and applied to leftover FIT material. Classification and regression tree (CART) analysis was applied to biomarker concentrations to identify the optimal combination for detecting advanced neoplasia. Performance of this combination, the mtFIT, was cross-validated using a leave-one-out approach and compared with FIT at equal specificity.

RESULTS

The CART analysis showed a combination of hemoglobin, calprotectin, and serpin family F member 2-the mtFIT-to have a cross-validated sensitivity for advanced neoplasia of 42.9% (95% CI, 36.2% to 49.9%) versus 37.3% (CI, 30.7% to 44.2%) for FIT (P = 0.025), with equal specificity of 96.6%. In particular, cross-validated sensitivity for advanced adenomas increased from 28.1% (CI, 20.8% to 36.5%) to 37.8% (CI, 29.6% to 46.5%) (P = 0.006). On the basis of these results, early health technology assessment indicated that mtFIT-based screening could be cost-effective compared with FIT.

LIMITATION

Study population is enriched with persons from a referral population.

CONCLUSION

Compared with FIT, the mtFIT showed better diagnostic accuracy in detecting advanced neoplasia because of an increased detection of advanced adenomas. Moreover, early health technology assessment indicated that these results provide a sound basis to pursue further development of mtFIT as a future test for population-based CRC screening. A prospective screening trial is in preparation.

PRIMARY FUNDING SOURCE

Stand Up to Cancer/Dutch Cancer Society, Dutch Digestive Foundation, and HealthHolland.

Impact of the COVID-19 pandemic on faecal immunochemical test-based colorectal cancer screening programmes in Australia, Canada, and the Netherlands: a comparative modelling study

BACKGROUND

Colorectal cancer screening programmes worldwide have been disrupted during the COVID-19 pandemic. We aimed to estimate the impact of hypothetical disruptions to organised faecal immunochemical test-based colorectal cancer screening programmes on short-term and long-term colorectal cancer incidence and mortality in three countries using microsimulation modelling.

METHODS

In this modelling study, we used four country-specific colorectal cancer microsimulation models-Policy1-Bowel (Australia), OncoSim (Canada), and ASCCA and MISCAN-Colon (the Netherlands)-to estimate the potential impact of COVID-19-related disruptions to screening on colorectal cancer incidence and mortality in Australia, Canada, and the Netherlands annually for the period 2020-24 and cumulatively for the period 2020-50. Modelled scenarios varied by duration of disruption (3, 6, and 12 months), decreases in screening participation after the period of disruption (0%, 25%, or 50% reduction), and catch-up screening strategies (within 6 months after the disruption period or all screening delayed by 6 months).

FINDINGS

Without catch-up screening, our analysis predicted that colorectal cancer deaths among individuals aged 50 years and older, a 3-month disruption would result in 414-902 additional new colorectal cancer diagnoses (relative increase 0·1-0·2%) and 324-440 additional deaths (relative increase 0·2-0·3%) in the Netherlands, 1672 additional diagnoses (relative increase 0·3%) and 979 additional deaths (relative increase 0·5%) in Australia, and 1671 additional diagnoses (relative increase 0·2%) and 799 additional deaths (relative increase 0·3%) in Canada between 2020 and 2050, compared with undisrupted screening. A 6-month disruption would result in 803-1803 additional diagnoses (relative increase 0·2-0·4%) and 678-881 additional deaths (relative increase 0·4-0·6%) in the Netherlands, 3552 additional diagnoses (relative increase 0·6%) and 1961 additional deaths (relative increase 1·0%) in Australia, and 2844 additional diagnoses (relative increase 0·3%) and 1319 additional deaths (relative increase 0·4%) in Canada between 2020 and 2050, compared with undisrupted screening. A 12-month disruption would result in 1619-3615 additional diagnoses (relative increase 0·4-0·9%) and 1360-1762 additional deaths (relative increase 0·8-1·2%) in the Netherlands, 7140 additional diagnoses (relative increase 1·2%) and 3968 additional deaths (relative increase 2·0%) in Australia, and 5212 additional diagnoses (relative increase 0·6%) and 2366 additional deaths (relative increase 0·8%) in Canada between 2020 and 2050, compared with undisrupted screening. Providing immediate catch-up screening could minimise the impact of the disruption, restricting the relative increase in colorectal cancer incidence and deaths between 2020 and 2050 to less than 0·1% in all countries. A post-disruption decrease in participation could increase colorectal cancer incidence by 0·2-0·9% and deaths by 0·6-1·6% between 2020 and 2050, compared with undisrupted screening.

INTERPRETATION

Although the projected effect of short-term disruption to colorectal cancer screening is modest, such disruption will have a marked impact on colorectal cancer incidence and deaths between 2020 and 2050 attributable to missed screening. Thus, it is crucial that, if disrupted, screening programmes ensure participation rates return to previously observed rates and provide catch-up screening wherever possible, since this could mitigate the impact on colorectal cancer deaths.

FUNDING

Cancer Council New South Wales, Health Canada, and Dutch National Institute for Public Health and Environment.