BC20240716

Image by freepik

Introduction

Breast cancer remains a significant health concern worldwide, with early detection being crucial for improving patient outcomes. Traditional mammography, while effective, has limitations, particularly in women with dense breast tissue. Recent advancements in artificial intelligence (AI) have introduced new methods for enhancing breast cancer screening. This review focuses on the use of an AI tool, AISmartDensity, in supplemental MRI screening, as investigated in the ScreenTrustMRI trial. The trial's findings suggest that AI-based selection for MRI can significantly improve cancer detection rates and cost-effectiveness compared to traditional methods.

Background

Screening mammography has been the cornerstone of breast cancer detection, reducing mortality rates. However, interval cancers—those diagnosed between regular screenings—pose a challenge, especially in women with dense breasts. Dense breast tissue can mask tumors, making them harder to detect with mammography alone. Supplemental screening methods, such as MRI, have been proposed to address this issue. MRI is more sensitive than mammography but is also more expensive and resource-intensive. Therefore, an effective triage method is needed to identify individuals who would benefit most from supplemental MRI.

AISmartDensity and the ScreenTrustMRI Trial

The ScreenTrustMRI trial aimed to evaluate the effectiveness of an AI tool, AISmartDensity, in selecting individuals for supplemental MRI screening. The trial included 59,354 women whose mammograms were assessed using AISmartDensity. Of these, 4,103 women with a 'very high' AISmartDensity score (top 6.9%) were eligible for the study. Participants were randomly assigned to receive either supplemental MRI or no additional screening.

Cancer Detection Rates

The trial reported a cancer detection rate of 64.4 cancers per 1,000 MRI examinations, significantly higher than traditional methods. For comparison, the DENSE trial, which used traditional mammographic density for selection, reported a detection rate of 16.5 cancers per 1,000 MRI exams. The positive predictive value (PPV) for individuals recalled after MRI was 38%, and for those who underwent biopsy, it was 50.7%. These results indicate that AISmartDensity is more effective in identifying individuals at high risk of breast cancer, leading to earlier and more accurate detection.

Cost-Effectiveness

Cost-effectiveness is a critical consideration in implementing supplemental MRI screening. The DENSE trial estimated the cost per quality-adjusted life year (QALY) gained for supplemental MRI every three years at €37,181. Given the higher cancer detection rate with AISmartDensity, the cost per QALY is likely to be significantly lower. This suggests that AI-based selection could make supplemental MRI screening more economically viable, potentially leading to broader implementation in population-wide screening programs.

Diagnostic Outcomes and Cancer Characteristics

The study also provided detailed insights into the diagnostic outcomes and characteristics of detected cancers. Among the 36 participants diagnosed with cancer, the median size of invasive cancers was 13 mm, smaller than the average sizes reported for mammography-detected and interval cancers. This smaller size indicates earlier detection, which is crucial for improving treatment outcomes. Most detected cancers had invasive components, with a significant proportion being ductal breast cancer. Only a small percentage of cases had lymph node metastases, further highlighting the potential for early intervention.

Comparison with Other Studies

The effectiveness of AISmartDensity was compared with other supplementary screening methods, including digital breast tomosynthesis, handheld ultrasound, and automated breast ultrasound. MRI consistently showed superior cancer detection rates. The meta-analysis of 22 studies, including the DENSE trial, reported a pooled cancer detection rate of 25.7 per 1,000 MRI exams. In contrast, the ScreenTrustMRI trial's detection rate was more than double, underscoring the potential of AI-based selection in improving screening outcomes.

Limitations and Future Directions

While the results are promising, the study has some limitations. The AISmartDensity tool has not yet received regulatory approval, and its effectiveness needs further validation in diverse populations. Additionally, the trial's follow-up period is ongoing, and long-term outcomes are yet to be fully assessed. Future research should focus on refining AI algorithms, exploring their applicability in different demographic groups, and evaluating the long-term benefits and cost-effectiveness of AI-based supplemental MRI screening.

Conclusion

The ScreenTrustMRI trial demonstrates that AI-based selection using AISmartDensity significantly enhances the detection of breast cancer in women with dense breast tissue. The higher cancer detection rates and potential cost-effectiveness make it a promising tool for improving breast cancer screening programs. As AI technology continues to evolve, its integration into clinical practice could revolutionize cancer detection, leading to earlier diagnosis and better patient outcomes. Further research and regulatory approval are essential to fully realize the benefits of this innovative approach.

Read PaperNature Cancer