Estimating the incremental population health impact of a multi-cancer early detection (MCED) test to complement existing screening among elevated risk populations with multiple cancer risk factors: a mathematical modeling study Authors: Ashley Kim, Ze Cong, Abdul-Rahman Jazieh, Timothy R. Church, Heidi Reichert, Gina Nicholson, Jon Fryzek, Sarah S. Cohen Published: 2024 ABSTRACT BackgroundThe added benefits of a multi-cancer early detection (MCED) test among individuals with multiple risk factors will help policy decision-makers allocate limited healthcare resources. This study sought to estimate the population health implications of adding an MCED test to standard-of-care (SOC) cancer screening tests among individuals aged 50–79 years with additional cancer risk factors (i.e., tobacco use, family history of cancer, and/or obesity).MethodsA mathematical model was developed to assess the potential screening efficiency of an MCED test in addition to current guideline-recommended screenings. Among the US population of 107 million adults aged 50–79 years, the size of, and cancer risk among specific subgroups (i.e., smokers, obese individuals, those with a family history of cancer) as well as the general population were estimated from the literature. Published estimates of screening uptake and/or performance were used to model the number of cancers detected by SOC screening alone, and the number of incremental cancers that could be detected by an MCED test. Screening efficiency outcomes included the true-positive:false-positive (TP:FP) ratio, diagnostic yield (DY), and cancer detection rate (CDR). Sensitivity analyses were conducted by varying the values of key parameters.ResultsAmong all subgroups, the TP:FP ratios were higher with an MCED test than with SOC screening alone, and higher than in the general population, suggesting improved screening efficiency with an MCED test. The estimated TP:FP ratios were 1:43.3 (SOC)/1:1.1 (MCED), 1:40.4/1:0.8, 1:36.9/1:0.5 among former, ever, and current smokers, respectively, 1:38.3/1:0.9 (those with a family history of cancer), and 1:39.3/1:1.1 (obese individuals). Among the general population, the TP:FP ratios were 1:43.5/1:1.1. Across all subpopulations, the DY and CDR increased by up to threefold with an MCED test, when compared to SOC screening alone, with up to 75% of cancers detected with an MCED test lacking a screening paradigm. These results were robust in sensitivity analyses.ConclusionsAdding an MCED test could improve screening efficiency among individuals with multiple risk factors, as well as the general population.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12913-024-12037-1. Background The United States (US) has recently seen a rapid population growth among older adults, and this trend is expected to continue as a larger proportion of the Baby Boom and Generation X cohorts live to older ages compared to previous generations [1]. The growth in the older population may foretell a rise in the annual number of cancer cases as advanced age is the most important risk factor for cancer overall. Between 2015 and 2050, the total number of cancer cases is predicted to increase by almost 50% [2]. In addition to age, however, there are numerous other risk factors for cancer including tobacco use, excess body weight, and chronic diseases [3]. Nearly 42% of cancer cases and 45% of cancer deaths in the US are attributed to potentially modifiable risk factors [4]. Non-modifiable cancer risk factors, including family history of cancer, have also been known to be associated with multiple cancer types [5]. Reduction in cancer morbidity and mortality may be achieved through multiple avenues, including therapeutic pharmacologic (e.g., immunotherapy), preventive pharmacologic (e.g., human papillomavirus [HPV] vaccines), and behavioral (e.g., tobacco cessation) interventions that either treat cancer or reduce cancer risk, as well as cancer screening programs that may detect cancers early when they may be able to reduce cancer mortality. Currently, guideline-recommended cancer-screening technologies are specific to 4–5 single cancers, meaning that populations are screened for one cancer type at a time. The US Preventive Services Task Force (USPSTF) recommends population screening for asymptomatic individuals who may have an elevated risk of developing cancer due to individual risk factors, such as age, gender, and/or smoking history, across breast [6], cervical [7], colorectal [8], and lung cancer [9]; prostate cancer screening is recommended on an individual basis, following consultation with a healthcare provider [10]. While current single-cancer screening modalities are effective, most cancers do not fall under the current guidelines, with approximately 70% of cancers diagnosed and 70% of cancer-related deaths having no guideline-recommended screening test [11, 12], and current real-world cancer screening adherence has been shown to be suboptimal [13]. In fact, individuals with risk factors for cancer have been shown to have an increased risk of more than one cancer type, with more than 10 + cancer types associated with tobacco use, for instance [14]. Recent innovations in cancer screening, including multi-cancer early detection (MCED) tests based on targeted methylation and next-generation sequencing have created the possibility to improve upon current standard-of-care (SOC) screening practices and advance cancer screening efficiency by detecting cancer types that would otherwise not be found with the current single cancer screening paradigm, while minimizing the false-positive rate [12] and burden on individuals eligible for screening. To date, a mathematical model evaluating the inclusion of an MCED test as an enhancement to current guideline-recommended screening tests demonstrated that adding an MCED test may provide improved efficiencies to screening in the general population aged 50 + years beyond those provided by SOC screening in the US and United Kingdom (UK) [15]. However, the population health impact of an MCED test among individuals with cancer risk factors (i.e., tobacco use, family history of cancer, obesity) in addition to age has not yet been evaluated. As such, this study will provide a conceptual framework for estimating the impact of adding an MCED test to current guideline-recommended screening among higher risk populations in general. Additionally, our study will provide public health policy and decision makers with a better understanding around resource allocation when facing resource constraints with cancer screening, and potentially limiting unnecessary diagnostic workups associated with false-positive results and anxiety among patients. Methods Study design An existing mathematical model [15] assessed the potential impact of one year of screening within an ongoing screening program in the US among individuals aged 50–79 years. The previous model estimated the efficiency (i.e., true-positive:false-positive [TP:FP] ratio, and the cost of diagnostic investigations among screen-positives, per cancer detected) of current guideline-recommended SOC screening alone, and when adding an MCED test to SOC screening similar-aged adults without cancer detected by recommended screening, or who were ineligible for recommended screening. Input parameters for SOC screening included cancer incidence rates, test uptake, adherence, and performance (i.e., test-specific sensitivity and false-positive rates for current screening), and for the incremental MCED test included the number of cancers that were not detected by screening, the screening performance of an MCED test (i.e., cancer-specific sensitivities by stage), and adherence. Cancers not detected by current recommended screening included all cancers without a screening test, and only interval cancers for those with a recommended screening test [15]. Of note, in addition to the cancer types that can be screened with SOC screening, an MCED test can screen for a wider range of cancers by predicting a common cancer signal across more than 50 cancer types (see Fig. 1) [16].Fig. 1Model schematic. aCalculated by multiplying the number of cancers covered by each screening type, adherence to that screening guideline, and the sensitivity of the screening modality. bCalculated by multiplying the cancers not detected by current SOC by the sensitivity of an MCED test for that cancer type, and adherence to MCED testing; additional cancer types detected by MCED testing included anus, anal canal and anorectum, bladder, breast, cervix uteri, colon and rectum, corpus and uterus, esophagus, head and neck, kidney and renal pelvis, liver and intrahepatic bile duct, lung and bronchus, lymphoid, melanoma, myeloid, myeloma, ovary, pancreas, prostate, sarcoma, stomach, thyroid, etc. Model schematic. aCalculated by multiplying the number of cancers covered by each screening type, adherence to that screening guideline, and the sensitivity of the screening modality. bCalculated by multiplying the cancers not detected by current SOC by the sensitivity of an MCED test for that cancer type, and adherence to MCED testing; additional cancer types detected by MCED testing included anus, anal canal and anorectum, bladder, breast, cervix uteri, colon and rectum, corpus and uterus, esophagus, head and neck, kidney and renal pelvis, liver and intrahepatic bile duct, lung and bronchus, lymphoid, melanoma, myeloid, myeloma, ovary, pancreas, prostate, sarcoma, stomach, thyroid, etc. In the current base-case analysis, the existing model was adapted to assess cancer screenings among individuals aged 50–79 years with additional cancer risk factors [17], including previous and/or current tobacco use, family history of cancer (i.e., having a history of cancer in first-degree relatives), and obesity (i.e., body mass index of ≥ 30), as well as the general population aged 50–79 years (used as the reference group), in the US. A schematic of the current model is described in Fig. 1. Model inputs Inputs into the mathematical model included screening characteristics and metrics for both SOC screening alone and an incremental MCED test (Table 1). For SOC screening, the specific screening modality, screening interval, and screening-eligible population for each of the four cancer types were based on the most recent guideline-recommended criteria from the USPSTF and the American Cancer Society (ACS) [6–9]. In the base case, the screening modalities assessed included mammography for breast cancer [6], cytology and high-risk human papillomavirus (HPV) testing for cervical cancer [7], fecal immunochemical test (FIT)-DNA test (e.g., Cologuard) for colorectal cancer [8] given its intermediate sensitivity and specificity relative to other recommended colorectal cancer screening tests, and low-dose computed tomography (LDCT) for eligible individuals with a history of smoking [9]. Estimates of screening adherence and test performance, which included test sensitivity and specificity, were also obtained from the literature [15, 18, 19]. Table 1Model inputs of screening characteristics and metrics for standard-of-care (SOC) screening and an MCED testCancer TypeScreening Modality% adults aged 50–79 years eligible for current recommended screeningAdherenceScreening IntervalSensitivitySpecificityBreastMammography90% of women78.3%Annuallyd87%89%ColorectalCologuard93%69.6%Every 3 years92%87%CervicalCotestinga63% of women79.8%Every 5 years95%85%LungLDCTb22% of general population;100% of current smokers;77.2% of ever smokers;66.8% of former smokersc5%Annually85%87%MultipleeMCED test100% of individuals who are not eligible for, or missed by, current SOC screening100%Annually51.5%99.5%aCotesting = Cervical cytology and hrHPV testing together. There is uncertainty in the estimate of uptake of cervical cancer screening as the BRFSS survey focuses on PAP testing rather than cotesting (e.g. women may not be aware of whether they had cotesting or not)bLDCT Low-dose computed tomographycFrom published literature estimates (Pinsky et al. 2015) of the lung cancer risk among individuals with varying years since quitting (YSQ) among former smokers were used to inform the proportion of the lung cancer incidence that was due to former smokers (89.0%) and ever smokers with a YSQ < 15 (94.1%) among all former and ever smokers, and the proportion of former smokers (66.8%) and ever smokers with a YSQ < 15 (77.2%) among all individuals eligible for LDCTdFollowing recommendations from the American Cancer Society and NCCN to screen annually (for some age groups) in order to maximize the benefits associated with current screeningeMultiple cancer types detected by MCED testing included anus, anal canal and anorectum, bladder, breast, cervix uteri, colon and rectum, corpus and uterus, esophagus, head and neck, kidney and renal pelvis, liver and intrahepatic bile duct, lung and bronchus, lymphoid, melanoma, myeloid, myeloma, ovary, pancreas, prostate, sarcoma, stomach, thyroid Table 1Model inputs of screening characteristics and metrics for standard-of-care (SOC) screening and an MCED testCancer TypeScreening Modality% adults aged 50–79 years eligible for current recommended screeningAdherenceScreening IntervalSensitivitySpecificityBreastMammography90% of women78.3%Annuallyd87%89%ColorectalCologuard93%69.6%Every 3 years92%87%CervicalCotestinga63% of women79.8%Every 5 years95%85%LungLDCTb22% of general population;100% of current smokers;77.2% of ever smokers;66.8% of former smokersc5%Annually85%87%MultipleeMCED test100% of individuals who are not eligible for, or missed by, current SOC screening100%Annually51.5%99.5%aCotesting = Cervical cytology and hrHPV testing together. There is uncertainty in the estimate of uptake of cervical cancer screening as the BRFSS survey focuses on PAP testing rather than cotesting (e.g. women may not be aware of whether they had cotesting or not)bLDCT Low-dose computed tomographycFrom published literature estimates (Pinsky et al. 2015) of the lung cancer risk among individuals with varying years since quitting (YSQ) among former smokers were used to inform the proportion of the lung cancer incidence that was due to former smokers (89.0%) and ever smokers with a YSQ < 15 (94.1%) among all former and ever smokers, and the proportion of former smokers (66.8%) and ever smokers with a YSQ < 15 (77.2%) among all individuals eligible for LDCTdFollowing recommendations from the American Cancer Society and NCCN to screen annually (for some age groups) in order to maximize the benefits associated with current screeningeMultiple cancer types detected by MCED testing included anus, anal canal and anorectum, bladder, breast, cervix uteri, colon and rectum, corpus and uterus, esophagus, head and neck, kidney and renal pelvis, liver and intrahepatic bile duct, lung and bronchus, lymphoid, melanoma, myeloid, myeloma, ovary, pancreas, prostate, sarcoma, stomach, thyroid Model inputs of screening characteristics and metrics for standard-of-care (SOC) screening and an MCED test 22% of general population; 100% of current smokers; 77.2% of ever smokers; 66.8% of former smokersc aCotesting = Cervical cytology and hrHPV testing together. There is uncertainty in the estimate of uptake of cervical cancer screening as the BRFSS survey focuses on PAP testing rather than cotesting (e.g. women may not be aware of whether they had cotesting or not) bLDCT Low-dose computed tomography cFrom published literature estimates (Pinsky et al. 2015) of the lung cancer risk among individuals with varying years since quitting (YSQ) among former smokers were used to inform the proportion of the lung cancer incidence that was due to former smokers (89.0%) and ever smokers with a YSQ < 15 (94.1%) among all former and ever smokers, and the proportion of former smokers (66.8%) and ever smokers with a YSQ < 15 (77.2%) among all individuals eligible for LDCT dFollowing recommendations from the American Cancer Society and NCCN to screen annually (for some age groups) in order to maximize the benefits associated with current screening eMultiple cancer types detected by MCED testing included anus, anal canal and anorectum, bladder, breast, cervix uteri, colon and rectum, corpus and uterus, esophagus, head and neck, kidney and renal pelvis, liver and intrahepatic bile duct, lung and bronchus, lymphoid, melanoma, myeloid, myeloma, ovary, pancreas, prostate, sarcoma, stomach, thyroid The estimated size of each subpopulation was scaled down from the general US population of approximately 107,000,000 individuals aged 50–79 years (for 2020) [15], by applying the prevalence of each risk factor in the general population as identified from the literature (Table 2) [14, 20]. For each risk factor, a standardized incidence ratio (SIR) was derived from the literature and 2020 US Census data and calculated as an estimate of the increased risk of cancer relative to the general population aged 50 + years (Table 2). Given the USPSTF recommendations, published literature estimates of the lung cancer risk among individuals with varying years since quitting (YSQ) among former smokers [21] were used to inform the proportion of the lung cancer incidence that was due to former smokers (89.0%) and ever smokers with a YSQ < 15 (94.1%) among all former and ever smokers, and the proportion of former smokers (66.8%) and ever smokers with a YSQ < 15 (77.2%) among all individuals eligible for LDCT. Ever smokers were assumed to be comprised of 60% former and 40% current smokers based on the prevalence of former and current smokers in the general population [14]. For an MCED test, the screening performance was based on the Galleri® test (GRAIL, Inc; Menlo Park, CA), which utilizes targeted methylation and a next-generation sequencing-based assay to analyze circulating cell-free DNA to detect multiple cancer types, all of which are specified in Fig. 1 [16]. Published sensitivities for each cancer type and stage were used in the model based on a real-world observational study with an overall sensitivity of 51.5% and a false-positive rate of 0.5% [16] across all cancer types included in the study (Additional file 1). Table 2Prevalence and risk of cancer among elevated-risk subpopulationsPopulationaSize of Population (n)Prevalence of risk factor (%)Standardized incidence ratioGeneral population107,000,000N/A1.00 (reference)Former smokers27,606,00025.8%b1.03Ever smokers45,796,00042.8%b1.24Current smokers18,083,00016.9%b1.57Individuals with a family history of cancer38,092,00035.6%c1.23Obese individuals40,981,00038.3%b1.09aEstimates reflect the prevalence and risk of cancer among individuals aged 50–79 years with an elevated risk of cancer, except for those with a family history of cancer, which includes all adult individualsbFrom published literature estimates (Islami et al. 2018) of the prevalence of individuals with a history of smoking and obesitycFrom published literature estimates (Teerlink et al. 2012) of the prevalence of individuals with a family history of cancer Table 2Prevalence and risk of cancer among elevated-risk subpopulationsPopulationaSize of Population (n)Prevalence of risk factor (%)Standardized incidence ratioGeneral population107,000,000N/A1.00 (reference)Former smokers27,606,00025.8%b1.03Ever smokers45,796,00042.8%b1.24Current smokers18,083,00016.9%b1.57Individuals with a family history of cancer38,092,00035.6%c1.23Obese individuals40,981,00038.3%b1.09aEstimates reflect the prevalence and risk of cancer among individuals aged 50–79 years with an elevated risk of cancer, except for those with a family history of cancer, which includes all adult individualsbFrom published literature estimates (Islami et al. 2018) of the prevalence of individuals with a history of smoking and obesitycFrom published literature estimates (Teerlink et al. 2012) of the prevalence of individuals with a family history of cancer Prevalence and risk of cancer among elevated-risk subpopulations aEstimates reflect the prevalence and risk of cancer among individuals aged 50–79 years with an elevated risk of cancer, except for those with a family history of cancer, which includes all adult individuals bFrom published literature estimates (Islami et al. 2018) of the prevalence of individuals with a history of smoking and obesity cFrom published literature estimates (Teerlink et al. 2012) of the prevalence of individuals with a family history of cancer Outcomes Several outcome measures of screening performance were captured in the model, including the number of cancers detected by current screening, which was calculated by multiplying the number of cancers covered by each SOC screening modality, adherence to that screening modality, and its test sensitivity. The number of incremental cancers detected by an MCED test was calculated by multiplying the number of cancers not detected by current SOC screening, by the stage-specific sensitivity of and adherence to an MCED test, assuming 100% uptake of an MCED test. Additional outcomes included the TP:FP ratio, given positive screening test results will warrant a series of clinical and diagnostic investigations (e.g., imaging, biopsies), with a higher TP:FP ratio (i.e., fewer false positives per true-positive cancer detected) indicating higher screening efficiency, the cancer detection rate (CDR; i.e., the number of cancers detected divided by the number of cancers expected, based on the SIR], and the diagnostic yield (DY; i.e., the number of cancers detected among the total population of screened individuals). Scenario and sensitivity analyses Several scenario analyses were conducted to account for variabilities in the model inputs. In the first scenario, in order to reflect a healthy population bias in which individuals who agree to be screened may be healthier than the general population, cancer incidence was decreased by 20% in the SOC screening population. In the second scenario, given prostate cancer screening is recommended per the USPSTF and ACS guidelines on an individual basis, it was added to SOC screening, using adherence estimates of 39.5% from Behavioral Risk Factor Surveillance System (BRFSS) survey data [22] and sensitivity (21%) and specificity (91%) estimates from the literature [23]. In the third scenario, colonoscopy, which has a higher specificity than Cologuard and is commonly performed in the US, was assumed to be the SOC screening modality for colorectal cancer, as opposed to Cologuard in the base case, by assuming a 100% specificity for colonoscopy. In the fourth scenario, to project the impact of HPV vaccination on cervical cancer incidence rate among women who may have previously received the vaccine, the reduction in cervical cancer incidence associated with HPV vaccination from the literature (9.5%) [24] was projected forward in time until the women who were first eligible for vaccination in 2006 at age 26 reached age 50. In the final scenario, due to varying estimates of LDCT uptake among eligible individuals in the literature, a lower and upper bound of LDCT uptake at 1.2% [25] and 5.7% [26], respectively, were used. In addition, a probabilistic sensitivity analysis (PSA) was conducted to account for uncertainty in the model parameters by sampling parameters from their respective distributions, as opposed to simply using the mean/median parameter values, at the same time and to test the robustness of the model for the general population and subpopulations. A Monte Carlo simulation was conducted with 1,000 iterations, and varied base-case model inputs simultaneously using beta distributions for SOC screening test performance and gamma distributions for cancer incidence rates and MCED test sensitivity values. The MCED test sensitivity values of different stages were randomly sampled from 4,000 sensitivity draws based on existing data [16] for each cancer type, and a 10–15% variation around the base case value for adherence to SOC test was assumed and sampled from given the lack of data. Results Base case results for subpopulations with an elevated risk of cancer Among former, ever, and current smokers, SOC screening detected an estimated 49,958 (TP:FP ratio, 1:43.3), 89,391 (1:40.4), and 39,397 (1:36.9) cancers, respectively. An incremental MCED test with 100% uptake detected an additional 122,756 (1:1.1), 285,284 (1:0.8), and 161,730 (1:0.5) cancers among former, ever, and current smokers, respectively (Table 3). For former smokers, for instance, the TP:FP ratio of 1:43.3 with SOC screening assumes that to detect one person with any of the four cancers with SOC screening, approximately 43 people without these cancers may have diagnostic investigations following a screen-positive result. The TP:FP ratio of 1:1.1 with the addition of an MCED test for former smokers, however, suggests that to detect one person with cancer, only 1.1 people without cancer may undergo diagnostic investigations following a positive MCED test, thereby indicating greater screening efficiency. For those with a family history of cancer, an estimated 74,956 (1:38.3) cancers were detected with SOC screening, and an additional 216,914 (1:0.9) cancers with an incremental MCED test. For obese individuals, approximately 78,545 (1:39.3) cancers were detected by SOC screening, with an additional 192,303 (1:1.1) by an incremental MCED test. Of note, in the general population, an estimated 185,666 (1:43.5) cancers were detected by SOC screening, and an additional 461,427 (1:1.1) using an incremental MCED test. Table 3Base case results for the general and elevated-risk populationsPopulationSOC screening onlyaIncremental MCED testTotal positives (n)TP (yield per 1,000)TP:FP ratioDYb(n)CDRc (%)Total positives (n)TP (yield per 1,000)TP:FP ratioDYb(n)CDRc(%)General population8,255,903185,6661:43.52.7115%990,233461,4271:1.14.3136%Former smokers2,211,04349,9581:43.32.7315%259,142122,7561:1.14.4536%Ever smokers3,704,34089,3911:40.42.9313%510,970285,2841:0.86.2342%Current smokers1,492,69839,3971:36.93.2612%250,502161,7301:0.58.9448%Individuals with a family history of cancer2,947,55774,9561:38.33.0813%404,669216,9141:0.95.6939%Obese individuals3,169,00178,5451:39.33.0015%394,629192,3031:1.14.6936%aIncludes screening for breast, cervical, lung, and colorectal cancerbDiagnostic yield was computed as the number of cancers detected/total population of screened individuals, expressed per 1,000 peoplecCancer detection rate was computed as the number of cancers detected/the number of cancers expected Table 3Base case results for the general and elevated-risk populationsPopulationSOC screening onlyaIncremental MCED testTotal positives (n)TP (yield per 1,000)TP:FP ratioDYb(n)CDRc (%)Total positives (n)TP (yield per 1,000)TP:FP ratioDYb(n)CDRc(%)General population8,255,903185,6661:43.52.7115%990,233461,4271:1.14.3136%Former smokers2,211,04349,9581:43.32.7315%259,142122,7561:1.14.4536%Ever smokers3,704,34089,3911:40.42.9313%510,970285,2841:0.86.2342%Current smokers1,492,69839,3971:36.93.2612%250,502161,7301:0.58.9448%Individuals with a family history of cancer2,947,55774,9561:38.33.0813%404,669216,9141:0.95.6939%Obese individuals3,169,00178,5451:39.33.0015%394,629192,3031:1.14.6936%aIncludes screening for breast, cervical, lung, and colorectal cancerbDiagnostic yield was computed as the number of cancers detected/total population of screened individuals, expressed per 1,000 peoplecCancer detection rate was computed as the number of cancers detected/the number of cancers expected Base case results for the general and elevated-risk populations aIncludes screening for breast, cervical, lung, and colorectal cancer bDiagnostic yield was computed as the number of cancers detected/total population of screened individuals, expressed per 1,000 people cCancer detection rate was computed as the number of cancers detected/the number of cancers expected Compared to the general population, the TP:FP ratio was higher (i.e., greater screening efficiency) among all subpopulations with an elevated risk of cancer with SOC screening only, particularly current smokers, followed by those with a family history of cancer and obese individuals (Table 3). Similarly, compared to the TP:FP ratio with an incremental MCED test in the general population, the TP:FP ratios were greater with the addition of an MCED test among current smokers, ever smokers, and those with a family history of cancer, and comparable with an incremental MCED test among former smokers and obese individuals (Table 3). With SOC screening alone, a lower fraction of all cancers that would be detected in the general population was likely to be detected among ever and current smokers, and those with a family history of cancer, as evidenced by the lower CDRs. The CDRs among former smokers and obese individuals were similar to that among the general population. Conversely, with an incremental MCED test, a larger fraction of all cancers that would be detected in the general population was likely to be detected among ever and current smokers, and those with a family history of cancer (Table 3). Overall, among all the high-risk groups assessed, a larger fraction of all cancers that would be detected in the general population was likely to be detected with an incremental MCED test among all subpopulations, given the CDR ranging from 36%–48%, which represented up to a threefold increase in cancers detected when compared to the CDR of 12%–15% with SOC screening alone (Table 3). The diagnostic yield was consistently higher among each of the subpopulations, compared to that among the general population, following similar trends of being the highest among current smokers, followed by those with a family history of cancer and obese individuals with SOC screening only, and the highest among current smokers, ever smokers, and those with a family history of cancer with the addition of an MCED test (Table 3). Overall, more cancers may be found per screen with an incremental MCED test versus SOC screening alone, with a 1.59- to 1.85-fold increase in diagnostic yield with the addition of an MCED test among the general population, those with a family history of cancer, obese individuals, and former smokers, and a 2.13- to 2.74-fold increase for ever and current smokers. Furthermore, of the cancer types that could be detected with an incremental MCED test, a large proportion lack a screening paradigm: 74% (general population); 76% (those with a family history of cancer, obese individuals); 60% (former smokers); 51% (ever smokers); 40% (current smokers). Results from the scenario and sensitivity analyses The results of the scenario analyses are shown in Additional file 2 for the general population and subpopulations with an elevated risk of cancer. Trends in screening efficiency outcomes for SOC screening only and an incremental MCED test were similar to those seen in the base case results for all subpopulations. First, for all scenarios, a similar trend was observed in the substantial increase in the TP:FP ratio (i.e., greater efficiency) with SOC screening alone and with an incremental MCED test among current smokers, followed by those with a family history of cancer, compared to that among the general population. With SOC screening alone, the TP:FP ratio increased by approximately 15% among current smokers and those with a family history of cancer across all scenarios, compared to that among the general population. With an incremental MCED test, the TP:FP ratio among current smokers and those with a family history of cancer was 1:0.5 and 1:0.8, respectively, compared to 1:1.1 among the general population. With SOC alone, the DY increased (i.e., improved efficiency) the most—though slightly—among current smokers, compared to the general population, and the CDR decreased slightly among current smokers, compared to the general population. Conversely, with an incremental MCED test, the DY and CDR increased substantially for all scenarios among current smokers compared to the general population. Among the same subpopulation of current smokers with an incremental MCED test, the DY was 9.17/1,000 people when considering decreased cancer incidence in the screening population, compared to 4.48/1,000 people among the general population, with a CDR of 49% among current smokers, compared to 37% among the general population. Similarly, when adding prostate cancer screening to SOC screening, the DY was 8.93/1,000 people among current smokers (vs. 4.30/1,000 people among the general population), with a CDR of 47% (vs. 36%). When replacing colorectal cancer screening with colonoscopy, the DY was 8.94/1000 people among current smokers (vs. 4.31/1,000 people among the general population), with a CDR of 47% (vs. 36%). When varying the cervical cancer incidence to account for the HPV vaccination, the DY was 9.29/1000 people among current smokers (vs. 4.54/1,000 people among the general population), with a CDR of 49% (vs. 38%). Finally, when varying the LDCT uptake from 1.2%–5.7%, the DY was 8.92–9.10/1,000 people among current smokers (vs. 4.31–4.33/1,000 people among the general population), with a CDR of 49% (vs. 38%). Results from the PSA demonstrated robustness in the model inputs for the general population and subpopulations. Among the general population (Fig. 2a), the median TP:FP ratio was approximately 1:41 with SOC screening alone (95% central range, 1:26–77), and 1:1.14 with an incremental MCED test (1:1.05–1.19). The median TP:FP ratio was similar among former smokers (Fig. 2b) at 1:41 with SOC screening (1:1.25– 1.78) and 1:1.11 (1:1.03–1:1.16) with an incremental MCED test. Among ever smokers (Fig. 2c), the median TP:FP ratio was 1:38 with SOC screening (1:23–70), vs 1:0.79 (1:0.75–0.82) with an incremental MCED test. Conversely, among current smokers (Fig. 2d), the median TP:FP ratio from the PSA was higher at 1:36 with SOC screening alone (1:22–67), versus 1:0.55 (1:0.52–0.57) with an incremental MCED test. Among those with a family history of cancer (Fig. 2e), the median TP:FP ratio was 1:36 with SOC screening alone (1:22–1:68), and 1:0.93 (1:0.85–0.97) with an incremental MCED test.Fig. 2Distribution of TP:FP ratios for SOC screening alone and an incremental MCED test in various populations based on probabilistic sensitivity analyses. The plots show TP:FP ratios resulting from 1,000 iterations of the model for SOC screening alone and an incremental MCED test, where values on the X-axis represent the range of TP:FP ratios. Panel a represents general population; panel b represents former smokers; panel c represents ever smokers; panel d represents current smokers; panel e represents individuals with a family history of cancer; and panel f represents obese individuals Distribution of TP:FP ratios for SOC screening alone and an incremental MCED test in various populations based on probabilistic sensitivity analyses. The plots show TP:FP ratios resulting from 1,000 iterations of the model for SOC screening alone and an incremental MCED test, where values on the X-axis represent the range of TP:FP ratios. Panel a represents general population; panel b represents former smokers; panel c represents ever smokers; panel d represents current smokers; panel e represents individuals with a family history of cancer; and panel f represents obese individuals Finally, among obese individuals (Fig. 2f), the median TP:FP ratio was 1:37 with SOC screening (1:23–1:70), and 1:0.94 (1:0.88–0.98) with an incremental MCED test. Based on the distribution of values generated from the PSA for SOC screening alone and an incremental MCED test among both the general population and subpopulations, the distribution of the TP:FP ratios of the incremental MCED test had a good separation from the distribution of that from the SOC screening only, and remained higher (i.e., greater screening efficiency). Discussion In this follow-up analysis to a previously published model focusing on the general US population aged 50 + years, there may be even greater screening efficiency when screening individuals aged 50 + years with additional cancer risk factors. In this US-based model, more than half of all cancers—both with and without currently available guideline-recommended screening—in the general population may be detected by combining SOC screening and an MCED test, with observed CDRs at or exceeding 50% for all elevated-risk populations evaluated. While implementing MCED testing in the general population aged 50 + years old with or without additional cancer risk factors has been demonstrated to represent an efficient cancer screening approach, the efficiency of screening has the potential to be even greater when adding an MCED test among more targeted elevated risk populations, most of whom have an increased risk of multiple cancer types, many of which are not currently screened for. In addition, while these subgroups are substantially smaller in size, they can identify a large proportion of additional cancers that would have been found in the overall population, despite identifying a fewer total number of additional cancers overall. For example, ever smokers constitute approximately 43% of the general population [14] but contribute nearly 53% of all expected cancers in the general population. In our modeling, we observed that if an MCED test were administered in addition to SOC screening to the ever smoker population in a given year, approximately 58% of cancers that would have otherwise been detected in the overall population would be detected among ever smokers, suggesting considerable screening efficiency. Among current smokers, the TP:FP ratio more than doubled with the addition of an MCED test at 1:0.5, compared to the TP:FP ratio of 1:1.1 among the general population, whereas the TP:FP ratio increased to a lesser extent by 20% with SOC screening alone among current smokers, compared to that among the general population, thereby implying greater efficiency with an incremental MCED test. This may be because SOC screening—particularly with the low uptake of LDCT—may not detect the majority of lung cancers that may be present among current smokers. Among individuals with a family history of cancer, both SOC screening alone and an incremental MCED test resulted in a higher TP:FP ratio and DY, compared to those in the general population. In contrast, the CDR was lower than that of the general population with SOC screening alone, and greater with an incremental MCED test, suggesting greater screening efficiency within this subpopulation. When an MCED test was added to SOC screening among obese individuals, the TP:FP ratio (1:1.1) and CDR (36%) were comparable to those among the general population, but the diagnostic yield was slightly higher (4.69 vs. 4.31). This may suggest that while more cancers may be found among the screened population of obese individuals, a similar type or subset of cancers may be detected within the obese population, many of which may be detected with an incremental MCED test. This study has several limitations inherent to the modeling approach. Several assumptions were made for the model inputs including estimates of screening uptake for breast, lung, colorectal, and cervical cancer based on national survey data [17] and likely higher than real-world screening adherence [25]. Estimates of cancer risk among, or prevalence of, individuals with additional cancer risk factors were also based on national survey data from 2012–2013 [14, 20], but were assumed to remain the same in this model (in 2020), given this was the best evidence to date. We also assumed that mammography was conducted annually, following ACS guidelines, instead of biennially, which assumes a greater number of cancers would be detected by SOC screening and less favorable outcomes for an MCED test. In addition, data on the psychosocial impact and additional harms associated with MCED testing and subsequent diagnostic regimens are limited and need continued evaluation. However, early studies [27, 28] have suggested a low negative psychological effect with MCED test results overall, with a small and transient increase in anxiety in individuals with a positive test result. In addition, a vignette-based pilot utility elicitation study suggested there was low disutility (i.e., quality of life impact) associated with false-positive results from an MCED test [29, 30]. Further, the estimated parameters and associated uncertainty intervals may not accurately reflect the characteristics of the specific subpopulations in which the MCED test might be used, due to factors like healthy screenee bias. While the direct correlation between any of these risk factors and screening utilization is unclear, individuals who smoke or are obese may be less focused on health and thus have similar or lower utilization of current screening as the general population as a result. Conversely, those with a family history of cancer may be more invested in undergoing screening. Our model did not vary adherence rates to screening for different subpopulations, and this may have resulted in either under- or over-estimates if in fact there is variation in adherence within different risk groups. Also, for the MCED test, a 100% uptake was assumed to reflect the maximum impact of an MCED test in this model, though the results from Hackshaw et al. [15] were robust to lower uptake assumptions and the TP:FP ratios were similar even with lower MCED uptake. Additionally, this model did not consider the development of cancers between cancer screening, and did not include assumptions on natural history, which could decrease test performance. This analysis also focused on several of the most common cancer risk factors, specifically those for which literature is widely available regarding prevalence and risk of multiple cancer types, but numerous other risk factors for cancer exist. Finally, our analyses used US-specific data to estimate the impact of screening among different subpopulations in the US only, whereas Hackshaw et al. [15] looked at similar outcomes but also in the general UK population. Despite differences in population size between the US and UK, the screening efficiency was higher (i.e., favorable TP:FP ratio) when adding an MCED test in both countries. This is also expected among higher risk populations with similar characteristics. Of note, given the scope of our analysis included the screening efficiency of cancer screening tests, and not a comprehensive value assessment of screening tests, cost data was not included for both the MCED test itself and the clinical follow-up required for individuals with both true- and false-positive test results, though the costs of diagnostic investigations following a positive screening result can be obtained and extrapolated from Hackshaw et al. [15]. Conclusion While new technologies for cancer screening, such as an MCED test, should be used upon careful consideration of their efficacy and cost, adding an MCED test to current guideline-recommended cancer screening may be a highly efficient strategy in select subpopulations with multiple risk factors (e.g., advanced age and tobacco use) to increase not only the number of, but also the detection of, multiple cancer types, including those that are aggressive and/or currently do not have recommended screening available. Our findings may inform public health experts and policy makers about the potential for future innovations in cancer screening to be able to screen for all cancers, instead of individual cancer types, while minimizing additional false-positive results, in specific populations with an even higher risk of multiple cancer types, compared to the general population. Future research is warranted to understand the effectiveness and economic impact of MCED tests, and the real-world implications of screening in a more enriched population. Supplementary Information Supplementary Material 1.Supplementary Material 2. Supplementary Material 1. Supplementary Material 2.