Increasing power in screening trials by testing control-arm specimens: Application to multicancer detection screening.

BACKGROUND: Cancer screening trials have required large sample-sizes and long time-horizons to demonstrate cancer mortality reductions, the primary goal of cancer screening. We examine assumptions and potential power gains from exploiting information from testing control-arm specimens, which we call the "Intended Effect" (IE) analysis that we explain in detail herein. The IE analysis is particularly suited to tests that can be conducted on stored specimens in the control-arm, such as stored blood for multicancer detection (MCD) tests. METHODS: We simulated hypothetical MCD screening trials to compare power and sample-size for the standard vs IE analysis. Under two assumptions that we detail herein, we projected the IE analysis for 3 existing screening trials (National Lung Screening Trial (NLST), Minnesota Colon Cancer Control Study (MINN-FOBT-A), and Prostate, Lung, Colorectal, Ovarian Cancer Screening Trial-colorectal component (PLCO-CRC)). RESULTS: Compared to the standard analysis for the 3 existing trials, the IE design could have reduced cancer-specific mortality p-values 5-fold (NLST), 33-fold (MINN-FOBT-A), or 14,160-fold (PLCO-CRC), or alternately, reduced sample-size (90% power) by 26% (NLST), 48% (MINN-FOBT-A), or 59% (PLCO-CRC). For potential MCD trial designs requiring 100,000 subjects per-arm to achieve 90% power for multi-cancer mortality for the standard analysis, the IE analysis achieves 90% power for only 37,500-50,000 per arm, depending on assumptions concerning control-arm test-positives. CONCLUSIONS: Testing stored specimens in the control arm of screening trials to conduct the IE analysis could substantially increase power to reduce sample-size or accelerate trials, and provide particularly strong power gains for MCD tests.

Cancer screening with multicancer detection tests: A translational science review.

Multicancer detection (MCD) tests use a single, easily obtainable biospecimen, such as blood, to screen for more than one cancer concurrently. MCD tests can potentially be used to improve early cancer detection, including cancers that currently lack effective screening methods. However, these tests have unknown and unquantified benefits and harms. MCD tests differ from conventional cancer screening tests in that the organ responsible for a positive test is unknown, and a broad diagnostic workup may be necessary to confirm the location and type of underlying cancer. Among two prospective studies involving greater than 16,000 individuals, MCD tests identified those who had some cancers without currently recommended screening tests, including pancreas, ovary, liver, uterus, small intestine, oropharyngeal, bone, thyroid, and hematologic malignancies, at early stages. Reported MCD test sensitivities range from 27% to 95% but differ by organ and are lower for early stage cancers, for which treatment toxicity would be lowest and the potential for cure might be highest. False reassurance from a negative MCD result may reduce screening adherence, risking a loss in proven public health benefits from standard-of-care screening. Prospective clinical trials are needed to address uncertainties about MCD accuracy to detect different cancers in asymptomatic individuals, whether these tests can detect cancer sufficiently early for effective treatment and mortality reduction, the degree to which these tests may contribute to cancer overdiagnosis and overtreatment, whether MCD tests work equally well across all populations, and the appropriate diagnostic evaluation and follow-up for patients with a positive test.

Effectiveness of Colonoscopy Screening vs Sigmoidoscopy Screening in Colorectal Cancer.

Importance: Randomized clinical screening trials have shown that sigmoidoscopy screening reduces colorectal cancer (CRC) incidence and mortality. Colonoscopy has largely replaced sigmoidoscopy for CRC screening, but long-term results from randomized trials on colonoscopy screening are still lacking. Objective: To estimate the additional screening benefit of colonoscopy compared with sigmoidoscopy. Design, Setting, and Participants: This comparative effectiveness simulation study pooled data on 358 204 men and women randomly assigned to sigmoidoscopy screening or usual care in 4 randomized sigmoidoscopy screening trials conducted in Norway, Italy, the US, and UK with inclusion periods in the years 1993 to 2001. The primary analysis of the study was conducted from January 19 to December 30, 2021. Intervention: Invitation to endoscopic screening. Main Outcomes and Measures: Primary outcomes were CRC incidence and mortality. Using pooled 15-year follow-up data, colonoscopy screening effectiveness was estimated assuming that the efficacy of colonoscopy in the proximal colon was similar to that observed in the distal colon in the sigmoidoscopy screening trials. The simulation model was validated using data from Norwegian participants in a colonoscopy screening trial. Results: This analysis included 358 204 individuals (181 971 women [51%]) aged 55 to 64 years at inclusion with a median follow-up time ranging from 15 to 17 years. Compared with usual care, colonoscopy prevented an estimated 50 (95% CI, 42-58) CRC cases per 100 000 person-years, corresponding to 30% incidence reduction (rate ratio, 0.70 [95% CI, 0.66-0.75]), and prevented an estimated 15 (95% CI, 11-19) CRC deaths per 100 000 person-years, corresponding to 32% mortality reduction (rate ratio, 0.68 [95% CI, 0.61-0.76]). The additional benefit of colonoscopy screening compared with sigmoidoscopy was 12 (95% CI, 10-14) fewer CRC cases and 4 (95% CI, 3-5) fewer CRC deaths per 100 000 person-years, corresponding to percentage point reductions of 6.9 (95% CI, 6.0-7.9) for CRC incidence and 7.6 (95% CI, 5.7-9.6) for CRC mortality. The number needed to switch from sigmoidoscopy to colonoscopy screening was 560 (95% CI, 486-661) to prevent 1 CRC case and 1611 (95% CI, 1275-2188) to prevent 1 CRC death. Conclusions and Relevance: The findings of this comparative effectiveness study assessing long-term follow-up after CRC screening suggest that there was an additional preventive effect on CRC incidence and mortality associated with colonoscopy screening compared with sigmoidoscopy screening, but the additional preventive effect was less than what was achieved by introducing sigmoidoscopy screening where no screening existed. The results probably represent the upper limit of what may be achieved with colonoscopy screening compared with sigmoidoscopy screening.

Incidentally Detected Lung Cancer in Persons Too Young or Too Old for Lung Cancer Screening in a Mississippi Delta Cohort.

INTRODUCTION: Lung cancer risk in screening age-ineligible persons with incidentally detected lung nodules is poorly characterized. We evaluated lung cancer risk in two age-ineligible Lung Nodule Program (LNP) cohorts. METHODS: Prospective observational study comparing 2-year cumulative lung cancer diagnosis risk, lung cancer characteristics, and overall survival between low-dose computed tomography (LDCT) screening participants aged 50 to 80 years and LNP participants aged 35 to younger than 50 years (young) and older than 80 years (elderly). RESULTS: From 2015 to 2022, lung cancer was diagnosed in 329 (3.43%), 39 (1.07%), and 172 (6.87%) LDCT, young, and elderly LNP patients, respectively. The 2-year cumulative incidence was 3.0% (95% confidence intervals [CI]: 2.6%-3.4%) versus 0.79% (CI: 0.54%-1.1%) versus 6.5% (CI: 5.5%-7.6%), respectively, but lung cancer diagnosis risk was similar between young LNP and Lung CT Screening Reporting and Data System (Lung-RADS) 1 (adjusted hazard ratio [aHR] = 0.88 [CI: 0.50-1.56]) and Lung-RADS 2 (aHR = 1.0 [0.58-1.72]). Elderly LNP risk was greater than Lung-RADS 3 (aHR = 2.34 [CI: 1.50-3.65]), but less than 4 (aHR = 0.28 [CI: 0.22-0.35]). Lung cancer was stage I or II in 62.92% of LDCT versus 33.33% of young (p = 0.0003) and 48.26% of elderly (p = 0.0004) LNP cohorts; 16.72%, 41.03%, and 29.65%, respectively, were diagnosed at stage IV. The aggregate 5-year overall survival rates were 57% (CI: 48-67), 55% (CI: 39-79), and 24% (CI: 15-40) (log-rank p < 0.0001). Results were similar after excluding persons with any history of cancer. CONCLUSIONS: LNP modestly benefited persons too young or old for screening. Differences in clinical characteristics and outcomes suggest differences in biological characteristics of lung cancer in these three patient cohorts.

Changes in uptake of stool-based colorectal cancer screening during the Covid-19 pandemic.

PURPOSE: Colorectal cancer (CRC) screening is underutilized and endoscopic colon screening includes a number of barriers that were exacerbated by the Covid-19 pandemic. At-home stool-based screening (SBS) increased during the pandemic and potentially reached eligible adults hesitant to be screened by endoscopy. The purpose of this analysis was to examine the change in uptake of SBS during the pandemic among adults not screened within guidelines by endoscopy. METHODS: We used data from the 2019 and 2021 National Health Interview Surveys to estimate uptake of SBS among adults aged 50-75 years, without a previous diagnosis of CRC and without guideline-concordant endoscopic screening. We also examined provider recommendations for screening tests. To examine if changes in uptake differed during the pandemic by demographic and health characteristics, we combined survey years and ran logistic regression models with an interaction term for each factor and survey year. RESULTS: In our study population, SBS increased 74% overall from 2019 to 2021 (8.7% to 15.1%; p < 0.001), with the largest percent increase among those aged 50-52 years (3.5% to 9.9%; p < 0.001). Among those aged 50-52 years, the ratio of endoscopy to SBS changed from 83%/17% in 2019 to 55%/45% in 2021. Cologuard was the only screening test where recommendations by healthcare providers significantly increased from 2019 (10.6% to 16.1%; p = 0.002). CONCLUSIONS: Use and recommendations for SBS increased substantially during the pandemic. Increased awareness among patients could potentially improve future CRC screening rates if uptake of SBS occurs among those unable or unwilling to be screened by endoscopy.

Oncologists' perceived confidence and attitudes toward managing pre-existing chronic comorbidities during patients' active cancer treatment.

PURPOSE: To examine practicing oncologists' perceived confidence and attitudes toward management of pre-existing chronic conditions(PECC) during active cancer treatment(ACT). METHODS: In December 2018, oncologists in the National Cancer Institute's Community Oncology Research Program (NCORP) were invited to complete a was pilot-tested, IRB-approved online survey about their perceived confidence in managing PECC. Pearson chi-square test was used to identify oncologists' differences in perceived confidence to manage PECC and attitudes toward co-management of patients' PECC with non-oncologic care providers. Perceived confidence and attitudes were analyzed as a function of medical specialty while controlling demographic and medical practice variables. RESULTS: A total of 391 oncologists responded to the survey, 45.8% stated medical oncology as their primary specialty, 15.1% hematology oncology, 15.1% radiation oncology, 6.9% surgical oncology, and 17.1% other specialties such as gynecology oncology. Overall, 68.3% agreed (agree/strongly agree) that they were confident to manage PECC in the context of standard of care. However, only 46.6% and 19.7% remained confident when managing PECC previously managed by a primary care physician (PCP) and by a non-oncology subspecialist, respectively. Most oncologists (58.3%) agreed that patients' overall care was well coordinated, and 63.7% agreed that patients had optimal cancer and non-cancer care when PECC was co-managed with a non-oncology care provider. CONCLUSION: Most oncologists felt confident to manage all PECC during patients' ACT, but their perceived confidence decreased for PECC previously managed by PCPs or by non-oncology subspecialists. Additionally, they had positive attitudes toward co-management of PECC with non-oncologic care providers. These results indicate opportunities for greater collaboration between oncologists and non-oncology care providers to ensure comprehensive and coordinated care for cancer patients with PECC.

Diagnostic follow-up of indeterminate pulmonary nodules in the Medicare population.

BACKGROUND: Management of indeterminate pulmonary nodules (IPNs) is associated with redistribution of lung cancer to earlier stages, but most subjects with IPNs do not have lung cancer. The burden of IPN management in Medicare recipients was assessed. METHODS: Surveillance, Epidemiology, and End Results-Medicare data were analyzed for IPNs, diagnostic procedures, and lung cancer status. IPNs were defined as chest computed tomography (CT) scans with accompanying International Classification of Diseases (ICD) codes of 793.11 (ICD-9) or R91.1 (ICD-10). Two cohorts were defined: persons with IPNs during 2014-2017 comprised the IPN cohort, whereas those with chest CT scans without IPNs during 2014-2017 comprised the control cohort. Excess rates of various procedures due to reported IPNs over 2 years of follow-up (chest CT, positron emission tomography [PET]/PET-CT, bronchoscopy, needle biopsy, and surgical procedures) were estimated using multivariable Poisson regression models comparing the cohorts adjusted for covariates. Prior data on stage redistribution associated with IPN management were then used to define a metric of excess procedures per late-stage case avoided. RESULTS: Totals of 19,009 and 60,985 subjects were included in the IPN and control cohorts, respectively; 3.6% and 0.8% had lung cancer during follow-up. Excess procedures per 100 persons with IPNs over a 2-year follow-up were 63, 8.2, 1.4, 1.9, and 0.9 for chest CT, PET/PET-CT, bronchoscopy, needle biopsy, and surgery, respectively. Corresponding excess procedures per late-stage case avoided were 48, 6.3, 1.1, 1.5, and 0.7 based on an estimated 1.3 late-stage cases avoided per 100 IPN cohort subjects. CONCLUSIONS: The metric of excess procedures per late-stage case avoided can be used to measure the benefits-to-harms tradeoff of IPN management.

Recalibration of a Deep Learning Model for Low-Dose Computed Tomographic Images to Inform Lung Cancer Screening Intervals.

Importance: Annual low-dose computed tomographic (LDCT) screening reduces lung cancer mortality, but harms could be reduced and cost-effectiveness improved by reusing the LDCT image in conjunction with deep learning or statistical models to identify low-risk individuals for biennial screening. Objective: To identify low-risk individuals in the National Lung Screening Trial (NLST) and estimate, had they been assigned a biennial screening, how many lung cancers would have been delayed 1 year in diagnosis. Design, Setting, and Participants: This diagnostic study included participants with a presumed nonmalignant lung nodule in the NLST between January 1, 2002, and December 31, 2004, with follow-up completed on December 31, 2009. Data were analyzed for this study from September 11, 2019, to March 15, 2022. Exposures: An externally validated deep learning algorithm that predicts malignancy in current lung nodules using LDCT images (Lung Cancer Prediction Convolutional Neural Network [LCP-CNN]; Optellum Ltd) was recalibrated to predict 1-year lung cancer detection by LDCT for presumed nonmalignant nodules. Individuals with presumed nonmalignant lung nodules were hypothetically assigned annual vs biennial screening based on the recalibrated LCP-CNN model, Lung Cancer Risk Assessment Tool (LCRAT + CT [a statistical model combining individual risk factors and LDCT image features]), and the American College of Radiology recommendations for lung nodules, version 1.1 (Lung-RADS). Main Outcomes and Measures: Primary outcomes included model prediction performance, the absolute risk of a 1-year delay in cancer diagnosis, and the proportion of people without lung cancer assigned a biennial screening interval vs the proportion of cancer diagnoses delayed. Results: The study included 10 831 LDCT images from patients with presumed nonmalignant lung nodules (58.7% men; mean [SD] age, 61.9 [5.0] years), of whom 195 were diagnosed with lung cancer from the subsequent screen. The recalibrated LCP-CNN had substantially higher area under the curve (0.87) than LCRAT + CT (0.79) or Lung-RADS (0.69) to predict 1-year lung cancer risk (P < .001). If 66% of screens with nodules were assigned to biennial screening, the absolute risk of a 1-year delay in cancer diagnosis would have been lower for recalibrated LCP-CNN (0.28%) than LCRAT + CT (0.60%; P = .001) or Lung-RADS (0.97%; P < .001). To delay only 10% of cancer diagnoses at 1 year, more people would have been safely assigned biennial screening under LCP-CNN than LCRAT + CT (66.4% vs 40.3%; P < .001). Conclusions and Relevance: In this diagnostic study evaluating models of lung cancer risk, a recalibrated deep learning algorithm was most predictive of 1-year lung cancer risk and had least risk of 1-year delay in cancer diagnosis among people assigned biennial screening. Deep learning algorithms could prioritize people for workup of suspicious nodules and decrease screening intensity for people with low-risk nodules, which may be vital for implementation in health care systems.

Evaluation of Lung Cancer Risk Among Persons Undergoing Screening or Guideline-Concordant Monitoring of Lung Nodules in the Mississippi Delta.

Importance: Guideline-concordant management of lung nodules promotes early lung cancer diagnosis, but the lung cancer risk profile of persons with incidentally detected lung nodules differs from that of screening-eligible persons. Objective: To compare lung cancer diagnosis hazard between participants receiving low-dose computed tomography screening (LDCT cohort) and those in a lung nodule program (LNP cohort). Design, Setting, and Participants: This prospective cohort study included LDCT vs LNP enrollees from January 1, 2015, to December 31, 2021, who were seen in a community health care system. Participants were prospectively identified, data were abstracted from clinical records, and survival was updated at 6-month intervals. The LDCT cohort was stratified by Lung CT Screening Reporting and Data System as having no potentially malignant lesions (Lung-RADS 1-2 cohort) vs those with potentially malignant lesions (Lung-RADS 3-4 cohort), and the LNP cohort was stratified by smoking history into screening-eligible vs screening-ineligible groups. Participants with prior lung cancer, younger than 50 years or older than 80 years, and lacking a baseline Lung-RADS score (LDCT cohort only) were excluded. Participants were followed up to January 1, 2022. Main Outcomes and Measures: Comparative cumulative rates of lung cancer diagnosis and patient, nodule, and lung cancer characteristics between programs, using LDCT as a reference. Results: There were 6684 participants in the LDCT cohort (mean [SD] age, 65.05 [6.11] years; 3375 men [50.49%]; 5774 [86.39%] in the Lung-RADS 1-2 and 910 [13.61%] in the Lung-RADS 3-4 cohorts) and 12 645 in the LNP cohort (mean [SD] age, 65.42 [8.33] years; 6856 women [54.22%]; 2497 [19.75%] screening eligible and 10 148 [80.25%] screening ineligible). Black participants constituted 1244 (18.61%) of the LDCT cohort, 492 (19.70%) of the screening-eligible LNP cohort, and 2914 (28.72%) of the screening-ineligible LNP cohort (P < .001). The median lesion size was 4 (IQR, 2-6) mm for the LDCT cohort (3 [IQR, 2-4] mm for Lung-RADS 1-2 and 9 [IQR, 6-15] mm for Lung-RADS 3-4 cohorts), 9 (IQR, 6-16) mm for the screening-eligible LNP cohort, and 7 (IQR, 5-11) mm for the screening-ineligible LNP cohort. In the LDCT cohort, lung cancer was diagnosed in 80 participants (1.44%) in the Lung-RADS 1-2 cohort and 162 (17.80%) in the Lung-RADS 3-4 cohort; in the LNP cohort, it was diagnosed in 531 (21.27%) in the screening-eligible cohort and 447 (4.40%) in the screening-ineligible cohort. Compared with Lung-RADS 1-2, the fully adjusted hazard ratios (aHRs) were 16.2 (95% CI, 12.7-20.6) for the screening-eligible cohort and 3.8 (95% CI, 3.0-5.0) for the screening-ineligible cohort; compared with Lung-RADS 3-4, the aHRs were 1.2 (95% CI, 1.0-1.5) and 0.3 (95% CI, 0.2-0.4), respectively. The stage of lung cancer was I to II in 156 of 242 patients (64.46%) in the LDCT cohort, 276 of 531 (52.00%) in the screening-eligible LNP cohort, and 253 of 447 (56.60%) in the screening-ineligible LNP cohort. Conclusions and Relevance: In this cohort study, the cumulative lung cancer diagnosis hazard of screening-age persons enrolled in the LNP was higher than that in a screening cohort, irrespective of smoking history. The LNP provided access to early detection for a higher proportion of Black persons.

Cervical cancer screening and predictors of screening by diabetes status.

PURPOSE: Women with diabetes have lower survival rates after a cervical cancer diagnosis compared to women without diabetes. Pap smears and human papilloma virus (HPV) testing are highly effective screening tests for cervical cancer, therefore, it is important to know the prevalence of guideline-concordant screening among women with diabetes and understand if their predictors of screening differ. The purpose of this analysis was to assess guideline-concordant cervical cancer screening and predictors by diabetes status. METHODS: We used the 2019 National Health Interview Survey data, limited to women aged 21-65 years without a previous diagnosis of cancer, a hysterectomy, or diagnosed with diabetes in the year prior to the survey. We considered the Pap and HPV tests together and concordance as being tested within the past 3 years as part of a routine exam. We calculated weighted, adjusted prevalence, and prevalence ratios (PRs) of screening concordance comparing women with diabetes to those without. RESULTS: The unadjusted prevalence of concordant screening was 66.5% for women with diabetes compared to 73.3% for women without diabetes (PR = 0.91 95% CI 0.84-0.98). In the fully adjusted model adjusting for factors known to be associated with diabetes and access to healthcare, the association was attenuated and no longer statistically significant (PR = 0.96 95% CI 0.89-1.04). CONCLUSION: Cervical cancer screening concordance was lower in women with diabetes compared to those without overall but the deficit appears to be due primarily to underlying differences in sociodemographic characteristics and access to healthcare and not diabetes independently.

Use and Outcomes of Low-Dose CT Scan Lung Cancer Screening in the Medicare Population.

BACKGROUND: Relatively little is known about various aspects of low-dose CT (LDCT) scan lung cancer screening in US clinical practice, including characteristics of cases diagnosed after screening. We assessed this using the Surveillance, Epidemiology, and End Results (SEER)-Medicare database. RESEARCH QUESTION: What were the characteristics of patients with lung cancer, including stage and survival, whose disease was diagnosed after LDCT scan screenings? STUDY DESIGN AND METHODS: We created an LDCT scan use cohort consisting of everyone in the 5% SEER-Medicare sample with ≥ 12 months of non-health maintenance organization (HMO) Part A and B coverage while 65 to 77 years of age from 2015 through 2019. LDCT scan use and lung cancer diagnosis rates were assessed in this cohort. Additionally, we created a lung cancer cohort consisting of patients who received a diagnosis between 2015 and 2017 at 65 to 78 years of age with complete (non-HMO Part A and B) coverage the year before diagnosis. The cases cohort comprised those screened or unscreened based on undergoing screening during that period; lung cancer characteristics and survival were compared between these groups. RESULTS: In the LDCT scan use cohort (n = 414,358), use rates increased from 0.10 (per 100 person-years) in 2015 to 1.3 in 2019. Among those with first screenings, 39.2% underwent a subsequent screen within 18 months. The 1-year cumulative lung cancer diagnosis rate after initial screenings was 2.4%. Claims for prescreen counseling were infrequent (about 10%). Of 48,891 patients in the lung cancer cohort, 1,150 (2.4%) underwent screening. Among screened patients, 52.3%, 11.0%, 20.7%, and 16.0% received diagnoses of stages I, II, III, and IV disease, respectively. Lung cancer-specific survival through 3 years was significantly greater in screened versus unscreened patients overall and for all stages except stage II; 3-year lung cancer-specific survival was 89.0% in screened patients with stage I disease. INTERPRETATION: LDCT scan use was low but increased over time. The lung cancer yield was substantial; cases among those who underwent screening primarily were in the early stage with high survival rates. Although screening rates were unacceptably low, screening outcomes in those Medicare recipients undergoing screening were favorable.

Multi-Cancer Early Detection Tests: Current Progress and Future Perspectives.

Advances in cancer screening and early detection methodologies may lead to the detection of precancerous lesions or early-stage cancer. The development of blood-based multi-cancer early detection (MCED) tests may aid in this challenge. Furthermore, MCED tests have the potential to address early detection gaps for cancers with and without screening modalities and lessen cancer disparities, but many unknowns remain. In this issue, Clarke and colleagues describe stage- and cancer-specific incidence and survival, derived from Surveillance, Epidemiology and End Results Program Data, stratified by race/ethnicity and sex. The investigators discuss the potential to identify earlier-stage cancers (stage shift) that could improve overall patient outcomes. In a simulation model, the authors found fewer cancer-related deaths when cancers were down-staged at the time of diagnosis. In this commentary, we discuss some unanswered questions in using MCED tests for screening, as well as what stage shifting may actually mean for patient outcomes. See related article by Clarke et al., p. 521.

Incidental Findings on Low-Dose CT Scan Lung Cancer Screenings and Deaths From Respiratory Diseases.

BACKGROUND: Incidental respiratory disease-related findings are frequently observed on low-dose CT (LDCT) lung cancer screenings. This study analyzed data from the National Lung Screening Trial (NLST) to assess the relationship between such findings and respiratory disease mortality (RDM), excluding lung cancer. RESEARCH QUESTION: Are incidental respiratory findings on LDCT scanning associated with increased RDM? STUDY DESIGN AND METHODS: Subjects in the NLST LDCT arm received three annual screens. Trial radiologists noted findings related to possible lung cancer, as well as respiratory-related incidental findings. Demographic characteristics, smoking history, and medical history were captured in a baseline questionnaire. Kaplan-Meier curves were used to assess cumulative RDM. Multivariate proportional hazards models were used to assess risk factors for RDM; in addition to incidental CT scan findings, variables included respiratory disease history (COPD/emphysema, and asthma), smoking history, and demographic factors (age, race, sex, and BMI). RESULTS: Of 26,722 subjects in the NLST LDCT arm, 25,002 received the baseline screen and a subsequent LDCT screen. Overall, 59% were male, 26.5% were aged ≥ 65 years at baseline, and 10.6% reported a history of COPD/emphysema. Emphysema on LDCT scanning was reported in 30.7% of subjects at baseline and in 44.2% at any screen. Of those with emphysema on baseline LDCT scanning, 18% reported a history of COPD/emphysema. Median mortality follow-up was 10.3 years. There were 3,639 deaths, and 708 were from respiratory diseases. Among subjects with no history of COPD/emphysema, 10-year cumulative RDM ranged from 3.9% for subjects with emphysema and reticular opacities to 1.1% for those with neither condition; the corresponding range among subjects with a COPD/emphysema history was 17.3% (both) to 3.7% (neither). Emphysema on LDCT imaging was associated with a significantly elevated RDM hazard ratio (2.27; 95% CI, 1.92-2.7) in the multivariate model. Reticular opacities (including honeycombing/fibrosis/scar) also had a significantly elevated hazard ratio (1.39; 95% CI, 1.19-1.62). INTERPRETATION: Incidental respiratory disease-related findings observed on NLST LDCT screens were frequent and associated with increased mortality from respiratory diseases.

Deep Learning for Lung Cancer Detection on Screening CT Scans: Results of a Large-Scale Public Competition and an Observer Study with 11 Radiologists.

PURPOSE: To determine whether deep learning algorithms developed in a public competition could identify lung cancer on low-dose CT scans with a performance similar to that of radiologists. MATERIALS AND METHODS: In this retrospective study, a dataset consisting of 300 patient scans was used for model assessment; 150 patient scans were from the competition set and 150 were from an independent dataset. Both test datasets contained 50 cancer-positive scans and 100 cancer-negative scans. The reference standard was set by histopathologic examination for cancer-positive scans and imaging follow-up for at least 2 years for cancer-negative scans. The test datasets were applied to the three top-performing algorithms from the Kaggle Data Science Bowl 2017 public competition: grt123, Julian de Wit and Daniel Hammack (JWDH), and Aidence. Model outputs were compared with an observer study of 11 radiologists that assessed the same test datasets. Each scan was scored on a continuous scale by both the deep learning algorithms and the radiologists. Performance was measured using multireader, multicase receiver operating characteristic analysis. RESULTS: The area under the receiver operating characteristic curve (AUC) was 0.877 (95% CI: 0.842, 0.910) for grt123, 0.902 (95% CI: 0.871, 0.932) for JWDH, and 0.900 (95% CI: 0.870, 0.928) for Aidence. The average AUC of the radiologists was 0.917 (95% CI: 0.889, 0.945), which was significantly higher than grt123 (P = .02); however, no significant difference was found between the radiologists and JWDH (P = .29) or Aidence (P = .26). CONCLUSION: Deep learning algorithms developed in a public competition for lung cancer detection in low-dose CT scans reached performance close to that of radiologists.Keywords: Lung, CT, Thorax, Screening, Oncology Supplemental material is available for this article. © RSNA, 2021.

Differences in the relationship between diabetes and prostate cancer among Black and White non-Hispanic men.

PURPOSE: Studies finding lower incidence rates of prostate cancer among men with diabetes have been primarily conducted in White non-Hispanic (WNH) populations. The purpose of this analysis is to compare the relationship between diabetes and prostate cancer among Black (BNH) and White non-Hispanic men. METHODS: We used Surveillance, Epidemiology, and End Results (SEER)-Medicare data from 2011 to 2015 to compare incidence rates and tumor characteristics between BNH and WNH men by diabetes status. Age-adjusted incidence rates and corresponding rate ratios (RR) by diabetes status were calculated overall and by tumor grade, stage, and PSA level separately for BNH and WNH men. We used multivariable logistic regression to compare tumor characteristics among men with prostate cancer in the numerator, both within and across race/ethnic groups. RESULTS: Overall age-adjusted incidence rates were significantly lower in men with diabetes compared to those without among WNH men [RR = 0.88 95% Confidence Interval (CI) 0.86-0.90] but there was no difference in rates by diabetes status among BNH men (RR = 1.01 95% CI 0.96-1.07). Men with diabetes were less likely to be diagnosed with distant-staged tumors compared to those without diabetes in both race/ethnic groups but the magnitude of difference by diabetes status was greater in BNH [Odds Ratio (OR) = 0.52 95% CI 0.42-0.64] than WNH (OR = 0.88 95% CI 0.81-0.95) men (p-value for interaction < 0.001). CONCLUSION: The relationship between diabetes and prostate cancer differed between BNH and WNH men. The differences could have implications in evaluating the effectiveness of prostate cancer screening in men with diabetes across racial/ethnic subgroups.