Prognosis of Gleason Score 9-10 Prostatic Adenocarcinoma in Needle Biopsies: A Nationwide Population-based Study.

BACKGROUND: Since 2014, prostate cancer is reported using five-tier grouping of Gleason scores. Studies have suggested prognostic heterogeneity within the groups. OBJECTIVE: We assessed the risk of prostate cancer death for men diagnosed with Gleason scores 4 + 5, 5 + 4, and 5 + 5 on needle biopsy in a population-based cohort. DESIGN, SETTING, AND PARTICIPANTS: We used the data from Prostate Cancer data Base Sweden (PCBaSe) 4.0 for a survival analysis. Among 199 620 men reported to have prostate cancer in 2000-2020, 172 112 were diagnosed on needle biopsy. The primary treatment was classified as androgen deprivation therapy (66%), deferred treatment (5%), radical prostatectomy (7%), or radical radiotherapy (21%). OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The risks of death from prostate cancer in men with Gleason score 9-10 at 5 and 10 yr were used as endpoints. Multivariable Cox regression models controlling for socioeconomic factors and primary treatment were used for time-to-event analyses of death from prostate cancer and death from any causes. RESULTS AND LIMITATIONS: A total of 20 419 (12%) men had a Gleason score of 9-10, including Gleason scores of 4 + 5, 5 + 4, and 5 + 5 in 14 333 (70%), 4223 (21%), and 1863 (9%) men, respectively. The risks of prostate cancer death for men with Gleason scores 4 + 5, 5 + 4, and 5 + 5 at 10 yr of follow-up were 0.45 (confidence interval [CI] 0.44-0.46), 0.56 (0.55-0.58), and 0.66 (0.63-0.68), respectively. The risks of death of any cause for men with Gleason scores 4 + 5, 5 + 4, and 5 + 5 at 10 yr were 0.73 (CI 0.72-0.74), 0.81 (0.80-0.83), and 0.87 (0.85-0.89), respectively. CONCLUSIONS: We demonstrate in the largest and most complete cohort analyzed to date that collapsing the Gleason scores by grouping results in loss of prognostic information in men with Gleason score 9-10 cancer. PATIENT SUMMARY: Survival of prostate cancer patients with the highest tumor grades varies depending on grade composition.

Development, multi-institutional external validation, and algorithmic audit of an artificial intelligence-based Side-specific Extra-Prostatic Extension Risk Assessment tool (SEPERA) for patients undergoing radical prostatectomy: a retrospective cohort study.

BACKGROUND: Accurate prediction of side-specific extraprostatic extension (ssEPE) is essential for performing nerve-sparing surgery to mitigate treatment-related side-effects such as impotence and incontinence in patients with localised prostate cancer. Artificial intelligence (AI) might provide robust and personalised ssEPE predictions to better inform nerve-sparing strategy during radical prostatectomy. We aimed to develop, externally validate, and perform an algorithmic audit of an AI-based Side-specific Extra-Prostatic Extension Risk Assessment tool (SEPERA). METHODS: Each prostatic lobe was treated as an individual case such that each patient contributed two cases to the overall cohort. SEPERA was trained on 1022 cases from a community hospital network (Trillium Health Partners; Mississauga, ON, Canada) between 2010 and 2020. Subsequently, SEPERA was externally validated on 3914 cases across three academic centres: Princess Margaret Cancer Centre (Toronto, ON, Canada) from 2008 to 2020; L'Institut Mutualiste Montsouris (Paris, France) from 2010 to 2020; and Jules Bordet Institute (Brussels, Belgium) from 2015 to 2020. Model performance was characterised by area under the receiver operating characteristic curve (AUROC), area under the precision recall curve (AUPRC), calibration, and net benefit. SEPERA was compared against contemporary nomograms (ie, Sayyid nomogram, Soeterik nomogram [non-MRI and MRI]), as well as a separate logistic regression model using the same variables included in SEPERA. An algorithmic audit was performed to assess model bias and identify common patient characteristics among predictive errors. FINDINGS: Overall, 2468 patients comprising 4936 cases (ie, prostatic lobes) were included in this study. SEPERA was well calibrated and had the best performance across all validation cohorts (pooled AUROC of 0·77 [95% CI 0·75-0·78] and pooled AUPRC of 0·61 [0·58-0·63]). In patients with pathological ssEPE despite benign ipsilateral biopsies, SEPERA correctly predicted ssEPE in 72 (68%) of 106 cases compared with the other models (47 [44%] in the logistic regression model, none in the Sayyid model, 13 [12%] in the Soeterik non-MRI model, and five [5%] in the Soeterik MRI model). SEPERA had higher net benefit than the other models to predict ssEPE, enabling more patients to safely undergo nerve-sparing. In the algorithmic audit, no evidence of model bias was observed, with no significant difference in AUROC when stratified by race, biopsy year, age, biopsy type (systematic only vs systematic and MRI-targeted biopsy), biopsy location (academic vs community), and D'Amico risk group. According to the audit, the most common errors were false positives, particularly for older patients with high-risk disease. No aggressive tumours (ie, grade >2 or high-risk disease) were found among false negatives. INTERPRETATION: We demonstrated the accuracy, safety, and generalisability of using SEPERA to personalise nerve-sparing approaches during radical prostatectomy. FUNDING: None.

Cost-effectiveness of Prostate Cancer Screening Using Magnetic Resonance Imaging or Standard Biopsy Based on the STHLM3-MRI Study.

Importance: The combination of prostate-specific antigen (PSA) testing with magnetic resonance imaging (MRI) for prostate cancer detection has rarely been evaluated in a screening context. The STHLM3-MRI screening-by-invitation study (NCT03377881) has reported the benefits of using MRI with subsequent combined targeted and standard biopsies compared with using standard biopsies alone. Objective: To investigate the cost-effectiveness of prostate cancer screening using MRI with combined targeted and standard biopsies compared with standard biopsies alone among men aged 55 to 69 years in Sweden, based on evidence from the STHLM3-MRI study. Design, Setting, and Participants: This economic evaluation study was conducted from a lifetime health care perspective using a microsimulation model to evaluate no screening and screening strategies among adult men in Sweden. Men aged 55 to 69 years in Sweden were simulated for no screening and screening strategies. Input parameters were obtained from the STHLM3-MRI study and recent reviews. One-way and probabilistic sensitivity analyses were performed in May 2022. Interventions: No screening, quadrennial PSA screening using standard biopsies alone, and MRI-based screening using combined targeted and standard biopsies. Main Outcomes and Measures: The number of tests, incidence, deaths, costs, quality-adjusted life-years (QALY), and incremental cost-effectiveness ratios (ICERs) were estimated. Results: A total 603 men were randomized to the standard arm, 165 of these participants (27.4%) did not undergo standard biopsy; 929 men were randomized to the experimental arm, 111 (11.9%) of whom did undergo MRI or any biopsy. Compared with no screening, the screening strategies were associated with reduced lifetime prostate cancer-related deaths by 6% to 9%. Screening with MRI and the combined biopsies resulted in an ICER of US $53 736, which is classified as a moderate cost per QALY gained in Sweden. Relative to screening with standard biopsies alone, MRI-based screening reduced the number of both lifetime biopsies and overdiagnosis by approximately 50% and had a high probability of being cost-effective than the traditional PSA screening. Conclusions and Relevance: For prostate cancer screening, this economic evaluation study found that PSA testing followed by MRI with subsequent combined targeted and standard biopsies had a high probability to be more cost-effective compared with the traditional screening pathway using PSA and standard biopsy. MRI-based screening may be considered for early detection of prostate cancer in Sweden.

Cost-Effectiveness of the Stockholm3 Test and Magnetic Resonance Imaging in Prostate Cancer Screening: A Microsimulation Study.

BACKGROUND: Stockholm3 is a risk model that combines the prostate-specific antigen (PSA) test, other plasma protein biomarkers, single nucleotide polymorphisms, and clinical variables. The STHLM3-MRI study (NCT03377881) found that the Stockholm3 test with magnetic resonance imaging (MRI) and combined targeted and systematic biopsies maintained the sensitivity for clinically significant cancers, and reduced the number of benign biopsies and clinically insignificant cancers. OBJECTIVE: To assess the cost-effectiveness of MRI-based screening for prostate cancer using either Stockholm3 as a reflex test or PSA alone. DESIGN, SETTING, AND PARTICIPANTS: A cost-utility analysis was performed from a lifetime societal perspective using a microsimulation model for men aged 55-69 yr in Sweden. Test characteristics were estimated from the STHLM3-MRI study. INTERVENTION: No screening and three quadrennial screening strategies, including either PSA ≥3 ng/ml or Stockholm3 with reflex test thresholds of PSA ≥1.5 or 2 ng/ml as criteria for referral to MRI, were performed, and those who were MRI positive had combined targeted and systematic biopsies. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Predictions included the number of tests, cancer incidence and mortality, costs, and quality-adjusted life-years. Uncertainties in key parameters were assessed using sensitivity analyses. RESULTS AND LIMITATIONS: Compared with no screening, the screening strategies were predicted to reduce prostate cancer deaths by 7-9% across a lifetime. The use of Stockholm3 with PSA ≥2 ng/ml resulted in a 60% reduction in MRI compared with screening using PSA. This Stockholm3 strategy was cost-effective with a probability of 70% at a cost-effectiveness threshold of €47 218 (500 000 Swedish Kronor). As a potential limitation, the economic perspective was specific to Sweden. CONCLUSIONS: Screening with the Stockholm3 test at a reflex threshold of PSA ≥2 ng/ml and MRI was predicted to be cost-effective in Sweden. PATIENT SUMMARY: The Stockholm3 test with image-based screening may reduce screening-related harms and costs, while maintaining the health benefits from early detection of prostate cancer.

External Evaluation of 3 Commercial Artificial Intelligence Algorithms for Independent Assessment of Screening Mammograms.

Importance: A computer algorithm that performs at or above the level of radiologists in mammography screening assessment could improve the effectiveness of breast cancer screening. Objective: To perform an external evaluation of 3 commercially available artificial intelligence (AI) computer-aided detection algorithms as independent mammography readers and to assess the screening performance when combined with radiologists. Design, Setting, and Participants: This retrospective case-control study was based on a double-reader population-based mammography screening cohort of women screened at an academic hospital in Stockholm, Sweden, from 2008 to 2015. The study included 8805 women aged 40 to 74 years who underwent mammography screening and who did not have implants or prior breast cancer. The study sample included 739 women who were diagnosed as having breast cancer (positive) and a random sample of 8066 healthy controls (negative for breast cancer). Main Outcomes and Measures: Positive follow-up findings were determined by pathology-verified diagnosis at screening or within 12 months thereafter. Negative follow-up findings were determined by a 2-year cancer-free follow-up. Three AI computer-aided detection algorithms (AI-1, AI-2, and AI-3), sourced from different vendors, yielded a continuous score for the suspicion of cancer in each mammography examination. For a decision of normal or abnormal, the cut point was defined by the mean specificity of the first-reader radiologists (96.6%). Results: The median age of study participants was 60 years (interquartile range, 50-66 years) for 739 women who received a diagnosis of breast cancer and 54 years (interquartile range, 47-63 years) for 8066 healthy controls. The cases positive for cancer comprised 618 (84%) screen detected and 121 (16%) clinically detected within 12 months of the screening examination. The area under the receiver operating curve for cancer detection was 0.956 (95% CI, 0.948-0.965) for AI-1, 0.922 (95% CI, 0.910-0.934) for AI-2, and 0.920 (95% CI, 0.909-0.931) for AI-3. At the specificity of the radiologists, the sensitivities were 81.9% for AI-1, 67.0% for AI-2, 67.4% for AI-3, 77.4% for first-reader radiologist, and 80.1% for second-reader radiologist. Combining AI-1 with first-reader radiologists achieved 88.6% sensitivity at 93.0% specificity (abnormal defined by either of the 2 making an abnormal assessment). No other examined combination of AI algorithms and radiologists surpassed this sensitivity level. Conclusions and Relevance: To our knowledge, this study is the first independent evaluation of several AI computer-aided detection algorithms for screening mammography. The results of this study indicated that a commercially available AI computer-aided detection algorithm can assess screening mammograms with a sufficient diagnostic performance to be further evaluated as an independent reader in prospective clinical trials. Combining the first readers with the best algorithm identified more cases positive for cancer than combining the first readers with second readers.

The economic burden of prostate cancer - a Swedish prevalence-based register study.

BACKGROUND: Incidence and prevalence of prostate cancer in Sweden have increased markedly due to prostate-specific antigen (PSA) testing. Moreover, new diagnostic tests and treatment technologies are expected to further increase the overall costs. Our aims were (i) to estimate the societal costs for existing testing, diagnosis, management and treatment of prostate cancer, and (ii) to provide reference values for future cost-effectiveness analyses of prostate cancer screening and treatment. METHODS: Taking a societal perspective, this study aimed to investigate the annual cost of prostate cancer in Sweden using a prevalence-based cost-of-illness approach. Resource utilisation and related costs within Stockholm Region during 2016 were quantified using data from the Stockholm PSA and Biopsy Register and other health and population registers. Costs included: (i) direct medical costs for health care utilisation at primary care, hospitals, palliative care and prescribed drugs; (ii) informal care; and (iii) indirect costs due to morbidity and premature mortality. The resource utilisation was valued using unit costs for direct medical costs and the human capital method for informal care and indirect costs. Costs for the Stockholm region were extrapolated to Sweden based on cancer prevalence and the average costs by age and resource type. RESULTS: The societal costs due to prostate cancer in Stockholm in 2016 were estimated to be €64 million Euro (€Mn), of which the direct medical costs, informal care and productivity losses represented 62, 28 and 10% of the total costs, respectively. The total annual costs extrapolated to Sweden were calculated to be €281 Mn. The average direct medical cost, average costs for informal care and productivity losses per prevalent case were €1510, €828 and €271, respectively. These estimates were sensitive to assumptions related to the proportion of primary care visits associated with PSA testing and the valuation method for informal care. CONCLUSION: The societal costs due to prostate cancer were substantial and constitute a considerable burden to Swedish society. Data from this study are relevant for future cost-effectiveness evaluations of prostate cancer screening and treatment.

Aggregate cost of mammography screening in the United States: comparison of current practice and advocated guidelines.

BACKGROUND: Controversy exists over how often and at what age mammography screening should be implemented. Given that evidence supports less frequent screening, the cost differences among advocated screening policies should be better understood. OBJECTIVE: To estimate the aggregate cost of mammography screening in the United States in 2010 and compare the costs of policy recommendations by professional organizations. DESIGN: A model was developed to estimate the cost of mammography screening in 2010 and 3 screening strategies: annual (ages 40 to 84 years), biennial (ages 50 to 69 years), and U.S. Preventive Services Task Force (USPSTF) guidelines (biennial for those aged 50 to 74 years and personalized based on risk for those younger than 50 years and based on comorbid conditions for those 75 years and older). SETTING: United States. PATIENTS: Women aged 40 to 85 years. INTERVENTION: Mammography annually, biennially, or following USPSTF guidelines. MEASUREMENTS: Cost of screening per year, using Medicare reimbursements. RESULTS: The estimated cost of mammography screening in the United States in 2010 was $7.8 billion, with approximately 70% of women screened. The simulated cost of screening 85% of women was $10.1 billion, $2.6 billion, and $3.5 billion for annual, biennial, and USPSTF guidelines, respectively. The largest drivers of cost (in order) were screening frequency, percentage of women screened, cost of mammography, percentage of women screened with digital mammography, and percentage of mammography recalls. LIMITATION: Cost estimates and assumptions used in the model were conservative. CONCLUSION: The cost of mammography varies by at least $8 billion per year on the basis of screening strategy. The USPSTF guidelines are based on the scientific evidence to date to maximize patient benefit and minimize harm but also result in far more effective use of resources. PRIMARY FUNDING SOURCE: University of California and the Safeway Foundation.

The C1C2: a framework for simultaneous model selection and assessment.

BACKGROUND: There has been recent concern regarding the inability of predictive modeling approaches to generalize to new data. Some of the problems can be attributed to improper methods for model selection and assessment. Here, we have addressed this issue by introducing a novel and general framework, the C1C2, for simultaneous model selection and assessment. The framework relies on a partitioning of the data in order to separate model choice from model assessment in terms of used data. Since the number of conceivable models in general is vast, it was also of interest to investigate the employment of two automatic search methods, a genetic algorithm and a brute-force method, for model choice. As a demonstration, the C1C2 was applied to simulated and real-world datasets. A penalized linear model was assumed to reasonably approximate the true relation between the dependent and independent variables, thus reducing the model choice problem to a matter of variable selection and choice of penalizing parameter. We also studied the impact of assuming prior knowledge about the number of relevant variables on model choice and generalization error estimates. The results obtained with the C1C2 were compared to those obtained by employing repeated K-fold cross-validation for choosing and assessing a model. RESULTS: The C1C2 framework performed well at finding the true model in terms of choosing the correct variable subset and producing reasonable choices for the penalizing parameter, even in situations when the independent variables were highly correlated and when the number of observations was less than the number of variables. The C1C2 framework was also found to give accurate estimates of the generalization error. Prior information about the number of important independent variables improved the variable subset choice but reduced the accuracy of generalization error estimates. Using the genetic algorithm worsened the model choice but not the generalization error estimates, compared to using the brute-force method. The results obtained with repeated K-fold cross-validation were similar to those produced by the C1C2 in terms of model choice, however a lower accuracy of the generalization error estimates was observed. CONCLUSION: The C1C2 framework was demonstrated to work well for finding the true model within a penalized linear model class and accurately assess its generalization error, even for datasets with many highly correlated independent variables, a low observation-to-variable ratio, and model assumption deviations. A complete separation of the model choice and the model assessment in terms of data used for each task improves the estimates of the generalization error.