Clinical Report: Model Predicts Death, Not Just Cancer

Overview

A novel prognostic model for predicting prostate cancer-specific mortality has shown superior performance compared to an established biopsy risk calculator. This model, validated in over 200,000 patients, achieved an AUC of 78% for 20-year mortality prediction, addressing a significant gap in clinical decision-making tools for prostate cancer screening.

Background

Expand on the limitations of existing calculators, particularly regarding time-to-event predictions.

Data Highlights

Key Findings

• The PLCO model achieved an AUC of 78% for 20-year mortality prediction in external validation.
• In the development cohort, the PLCO model demonstrated progressively higher discrimination with AUC values of 91% at 5 years and 80% at 10 years.
• Among patients in the highest decile of PBCG risk, the PLCO model outperformed the PBCG model in both cohorts.
• The model incorporates competing risks from other causes of mortality, enhancing its clinical utility.
• Exclusion of digital rectal examination and PSA velocity from the model reflects current guidelines and evidence on their limited efficacy.

Clinical Implications

The PLCO model provides a more accurate tool for predicting prostate cancer-specific mortality, allowing clinicians to make better-informed decisions regarding patient management. This model's ability to incorporate competing risks is particularly valuable in a population with a prolonged natural history of prostate cancer.

Conclusion

The development of this prognostic model represents a significant advancement in prostate cancer screening, offering improved risk stratification for patients. Its validation across large cohorts underscores its potential utility in clinical practice.

References

1. Patrick Lewicki, MD, MS, et al., PubMed, 2023 -- Predicting Long-Term Risk for Prostate Cancer Mortality Following a Prostate-Specific Antigen Screening Test: Prognostic Model Development and External Validation
2. American Urological Association, 2023 -- Early Detection of Prostate Cancer: AUA/SUO Guideline
3. The ASCO Post — Identifying Impending Death Helps Patients and Caregivers
4. BMC Psychiatry (Springer) — Risk prediction models for depression in older adults with cancer
5. The ASCO Post — Scientists Develop a ‘Digital Twin’ Model to Predict Cancer Treatment Responses KEY POINTS
6. asco ai in oncology — Federated Learning Enables Robust Prognostic Modeling in Anal Cancer Across International Real-World Cohorts
7. Identifying Impending Death Helps Patients and Caregivers
8. Risk prediction models for depression in older adults with cancer
9. Scientists Develop a ‘Digital Twin’ Model to Predict Cancer Treatment Responses
10. Early Detection of Prostate Cancer: AUA/SUO Guideline (2023) - American Urological Association
11. “A 16-yr Follow-up of the European Randomized study of Screening for Prostate Cancer” - PMC
12. Predicting Long-Term Risk for Prostate Cancer Mortality Following a Prostate-Specific Antigen Screening Test: Prognostic Model Development and External Validation - PubMed