By Billy Candra 
A groundbreaking new study led by investigators at the UCLA Health Jonsson Comprehensive Cancer Center suggests that artificial intelligence (AI) could significantly enhance the detection of interval breast cancers – those aggressive tumors that manifest between scheduled routine mammography screenings. This pioneering research indicates that AI has the potential to identify these cancers at an earlier, more treatable stage, thereby promising better screening practices, swifter intervention, and ultimately, vastly improved patient outcomes. Published in the esteemed Journal of the National Cancer Institute, the findings underscore a pivotal advancement in the ongoing battle against breast cancer.
The Silent Threat: Understanding Interval Breast Cancers
Breast cancer remains one of the most prevalent and devastating diseases globally, affecting millions of women each year. In the United States alone, approximately one in eight women will develop invasive breast cancer in their lifetime, with an estimated 287,850 new cases and 43,250 deaths projected for 2022. While routine mammography screenings have been instrumental in reducing mortality rates by enabling early detection, a significant challenge persists in the form of "interval breast cancers."
Interval cancers are defined as cancers detected after a negative mammogram but before the next scheduled screening. These cancers represent a critical concern because they often exhibit more aggressive biological characteristics, grow faster, and are typically diagnosed at a more advanced stage than screen-detected cancers, leading to a poorer prognosis. Epidemiological data suggest that interval cancers account for roughly 20-30% of all breast cancers diagnosed, making their early identification a high-priority area for medical research. The very nature of these cancers — their rapid progression or their subtle presentation — often means they are harder for radiologists to spot during initial screenings, even when retrospectively visible. This difficulty highlights a critical gap in current screening protocols that AI now promises to address.
A New Frontier: AI’s Role in Early Detection
The integration of artificial intelligence into medical diagnostics represents one of the most transformative advancements in modern healthcare. AI algorithms, particularly those based on machine learning and deep learning, possess the capacity to analyze vast datasets and identify intricate patterns that might be imperceptible or easily overlooked by the human eye. In the realm of medical imaging, AI has shown immense promise in assisting radiologists by enhancing diagnostic accuracy, reducing interpretation times, and flagging suspicious areas for further review. The market for AI in healthcare is experiencing rapid growth, with significant investment channeled into developing sophisticated tools that can support clinicians across various specialties.
For breast cancer screening, AI is emerging as a powerful adjunct to human expertise. While radiologists are highly skilled, factors such as fatigue, high workload, and the inherent subtlety of early cancer signs can contribute to missed detections. AI, with its tireless analytical capability, offers a consistent "second opinion" that can augment human performance without succumbing to such human limitations. The journey of AI in medical imaging has seen it evolve from basic pattern recognition to complex deep learning models capable of interpreting nuanced radiological features, making it an ideal candidate for tackling the challenge of interval breast cancers.
UCLA Health’s Landmark Study: Methodology and Distinctive Approach
The retrospective study conducted by UCLA Health investigators marks a significant stride in validating AI’s potential within the specific context of U.S. breast cancer screening practices. The research analyzed a substantial dataset comprising nearly 185,000 past mammograms performed between 2010 and 2019, which included both digital mammography (DM) and digital breast tomosynthesis (DBT). From this extensive pool, the research team meticulously focused on 148 cases where a woman had subsequently been diagnosed with interval breast cancer.
A critical aspect of this study was its focus on the distinct differences between U.S. and European screening methodologies. In the United States, digital breast tomosynthesis (DBT), often referred to as 3D mammography, has become the predominant screening modality, characterized by its ability to create a three-dimensional image of the breast, reducing tissue overlap and improving cancer detection rates compared to traditional 2D digital mammography. U.S. patients typically undergo annual screenings. Conversely, European screening programs have historically relied more heavily on 2D digital mammography (DM) and operate on longer screening intervals, usually every two to three years. These variations underscore the importance of U.S.-specific research, as findings from European studies, while valuable, may not directly translate due to differences in technology, screening frequency, and population demographics. This UCLA study is among the first to explore AI’s utility for interval cancer detection within the unique framework of U.S. screening protocols.
To understand why these cancers were not detected earlier, radiologists involved in the study meticulously reviewed each of the 148 interval cancer cases. They adapted a well-established European classification system to categorize the reasons for non-detection, providing a granular understanding of the challenges. This classification system included:
- Missed reading error: The cancer was visible on the initial mammogram but overlooked by the radiologist.
- Minimal signs-actionable: Subtle signs were present that, in retrospect, could have prompted further investigation.
- Minimal signs-non-actionable: Very faint signs were present, arguably below the threshold for human detection or action.
- True interval cancer: The cancer developed rapidly after a truly negative screening and was not visible retrospectively.
- Occult: The cancer was genuinely invisible on mammography, regardless of review.
- Missed due to a technical error: Issues with image acquisition or processing contributed to the missed diagnosis.
Unveiling Subtle Signs: How AI Performed
Following the human review, the research team applied a commercially available AI software, Transpara, to the initial screening mammograms that preceded the interval cancer diagnoses. The objective was to assess if the AI tool could identify subtle indicators of cancer that had been missed by human radiologists during the initial screenings or at least flag them as suspicious. Transpara, like many AI diagnostic tools, evaluates each mammogram and assigns a score, in this case, from 1 to 10, indicating the level of cancer risk. A score of 8 or higher was predetermined as a flag, signaling a potentially concerning finding requiring further scrutiny.
The results were compelling. The AI demonstrated a significant ability to identify "mammographically-visible" types of interval cancers earlier by flagging them at the time of screening. These are the tumors that, while present on the mammogram, were not initially detected by radiologists due to their subtle nature or the faintness of their signs. Researchers conservatively estimate that integrating this AI capability into standard screening practices could lead to a substantial reduction – approximately 30% – in the number of interval breast cancers. This reduction could translate directly into lives saved and improved quality of life for countless patients.
Dr. Tiffany Yu, assistant professor of Radiology at the David Geffen School of Medicine at UCLA and the study’s first author, emphasized the profound implications of these findings. "This finding is important because these interval cancer types could be caught earlier when the cancer is easier to treat," Dr. Yu stated. "For patients, catching cancer early can make all the difference. It can lead to less aggressive treatment and improve the chances of a better outcome." Early detection is consistently linked to higher survival rates, smaller tumors, and often, less invasive therapeutic interventions such as lumpectomy over mastectomy, and potentially avoiding chemotherapy.
Navigating Nuance: AI’s Promise and Current Limitations
While the study presents a powerful case for AI integration, it also provides a crucial, balanced perspective on the technology’s current limitations. Dr. Hannah Milch, assistant professor of Radiology at the David Geffen School of Medicine and senior author of the study, highlighted the need for further exploration of AI accuracy in real-world settings. "While we had some exciting results, we also uncovered a lot of AI inaccuracy and issues that need to be further explored in real-world settings," Dr. Milch noted.
One particularly insightful finding pertained to occult cancers – those genuinely invisible on mammography. Intriguingly, despite their invisibility to the human eye, the AI tool still flagged 69% of the screening mammograms associated with occult cancers as suspicious. This suggests that AI might be picking up on extremely subtle, perhaps even non-visual, patterns or textures that correlate with later cancer development. However, a significant caveat emerged: when the researchers examined the specific areas on the images that the AI marked as suspicious, the tool was accurate in pinpointing the actual cancer location only 22% of the time. This discrepancy indicates that while AI can detect a "signal" indicating heightened risk, its ability to precisely localize truly occult lesions still requires refinement.
This nuance is critical for the practical implementation of AI. If an AI system flags a mammogram as high-risk but cannot pinpoint the exact location of the suspicious area, it presents a challenge for radiologists. It could lead to increased anxiety for patients, potentially more follow-up tests (like ultrasound or MRI) that may ultimately be negative, and a higher burden on healthcare resources. As Dr. Yu wisely added, "While AI isn’t perfect and shouldn’t be used on its own, these findings support the idea that AI could help shift interval breast cancers toward mostly true interval cancers." This means AI could help reduce the proportion of interval cancers that were technically "missed" by human readers, allowing radiologists to focus on cases where cancer truly develops between screenings. "It shows potential to serve as a valuable second set of eyes, especially for the types of cancers that are the hardest to catch early. This is about giving radiologists better tools and giving patients the best chance at catching cancer early, which could lead to more lives saved."
The Road Ahead: Implications for Patients, Providers, and Policy
The implications of this UCLA Health study extend far beyond the laboratory, potentially transforming breast cancer screening paradigms for patients, healthcare providers, and policy-makers alike.
For patients, the promise of earlier detection of interval cancers offers a beacon of hope. A reduction in these aggressive, late-stage diagnoses means a greater likelihood of less invasive treatments, improved long-term survival rates, and a better quality of life post-diagnosis. It could also lead to reduced anxiety for individuals with dense breast tissue, who are at higher risk for interval cancers and for whom traditional mammography can be less effective.
For radiologists and healthcare providers, AI is poised to become an indispensable diagnostic assistant. Rather than replacing human expertise, AI tools like Transpara can act as a sophisticated "second reader," a tireless sentinel that helps identify subtle anomalies that might otherwise be missed. This could alleviate some of the immense pressure on radiologists, allowing them to focus their expertise on complex cases, integrate AI flags with clinical context, and ultimately enhance the overall accuracy and efficiency of screening programs. The challenge will be to integrate AI seamlessly into clinical workflows, ensuring that its alerts are actionable and provide meaningful insights without creating an excessive burden of false positives.
For healthcare systems and policymakers, the potential economic and logistical benefits are substantial. Earlier detection can lead to significant cost savings by reducing the need for more expensive, aggressive treatments for advanced cancers. It can also improve resource allocation by better identifying high-risk individuals who may benefit from supplementary screening modalities. However, careful consideration must be given to the regulatory landscape for AI in medicine. Robust validation studies, transparent algorithms, and clear guidelines for deployment will be essential to ensure patient safety and maintain public trust. Ethical considerations, such as potential biases in AI algorithms trained on specific demographics and data privacy concerns, must also be meticulously addressed as these technologies become more widespread.
Expert Perspectives and Future Outlook
The consensus among experts in the field is that while AI presents an exciting frontier, its full potential will only be realized through continued research and careful implementation. Oncologists emphasize that any tool that can improve early detection, especially for aggressive cancer types, is invaluable. Patient advocates often highlight the emotional and physical toll of late-stage diagnoses, underscoring the urgency for advancements that provide more definitive answers sooner. Developers of AI technology are continually refining algorithms, striving to improve both sensitivity and specificity while addressing the challenges of precise localization, especially for hard-to-visualize lesions.
The call for larger prospective studies, where AI is integrated into real-time clinical settings, is paramount. These studies will be crucial for understanding how radiologists interact with AI tools in practice, how patient management changes based on AI flags, and how the technology impacts overall screening program effectiveness. Key questions remain: How will radiologists interpret and act upon AI flags for areas not visible to the human eye? What are the optimal strategies for integrating AI into the diagnostic pathway to maximize benefits while minimizing potential drawbacks like increased false positives or unnecessary follow-ups?
The UCLA Health study provides a powerful testament to the transformative potential of AI in breast cancer screening. It offers a glimpse into a future where technology works hand-in-hand with human expertise to catch cancers earlier, improve treatment outcomes, and ultimately save more lives. This collaborative vision of AI as a ‘valuable second set of eyes’ is not merely an enhancement; it represents a fundamental shift towards more precise, proactive, and patient-centric cancer care.
Acknowledgements
The comprehensive work was supported in part by generous contributions from the National Institutes of Health, the National Cancer Institute, the Agency for Healthcare Research and Quality, and Early Diagnostics Inc.
Other contributing authors from UCLA include Dr. Anne Hoyt, Dr. Melissa Joines, Dr. Cheryce Fischer, Dr. Nazanin Yaghmai, Dr. James Chalfant, Dr. Lucy Chow, Dr. Shabnam Mortazavi, Christopher Sears, Dr. James Sayre, Dr. Joann Elmore, and Dr. William Hsu.
