By Billy Candra 
Understanding Interval Breast Cancers: A Critical Challenge in Early Detection
Breast cancer remains one of the most common cancers among women worldwide, and early detection is paramount for successful treatment and improved survival rates. Routine mammography screenings have significantly contributed to this progress, allowing for the identification of cancerous lesions often before they are palpable or symptomatic. However, a persistent challenge in breast cancer screening is the phenomenon of "interval breast cancers." These are cancers that are diagnosed between scheduled screening mammograms, typically within 12 to 24 months of a prior negative mammogram. They represent a significant clinical concern because they are often more aggressive, tend to be diagnosed at a later stage, and carry a poorer prognosis compared to cancers detected during routine screening.
Interval cancers account for a notable proportion of all breast cancer diagnoses, with estimates varying but often ranging from 20% to 30% of all breast cancers detected in screened populations. The reasons for their emergence are multifactorial, encompassing rapidly growing tumors, subtle imaging signs missed by radiologists, or cancers that are truly invisible on mammograms at the time of screening. The UCLA Health Jonsson Comprehensive Cancer Center’s recent study delves into this critical area, proposing that advanced AI algorithms could serve as a powerful tool to identify these elusive cancers earlier, thereby bridging a crucial gap in current screening protocols.
The UCLA Health Study: Unpacking the Methodology and Findings
Published in the prestigious Journal of the National Cancer Institute, the study offers compelling evidence that AI possesses the capability to identify "mammographically-visible" types of interval cancers at an earlier stage. This identification occurs by flagging suspicious areas during the initial screening, even when human radiologists might have overlooked them. These include tumors that, while visible on mammograms, were not detected by human interpretation due to their subtle nature, faint signs, or characteristics arguably below the human eye’s level of detection.
The researchers estimate that a widespread incorporation of AI into existing screening practices could lead to a substantial 30% reduction in the incidence of interval breast cancers. This figure underscores the transformative potential of AI in enhancing the efficacy of population-level breast cancer screening programs.
Dr. Tiffany Yu, an assistant professor of Radiology at the David Geffen School of Medicine at UCLA and the study’s first author, emphasized the clinical significance 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." This sentiment resonates deeply within the medical community, where the axiom "early detection saves lives" is a guiding principle. Early-stage cancers often require less invasive surgeries, may avoid chemotherapy or radiation entirely, and are associated with significantly higher five-year survival rates.
Navigating the Transatlantic Divide: U.S. vs. European Screening Paradigms
While similar research exploring AI’s role in breast cancer detection has been conducted in Europe, the UCLA study stands out as one of the first to specifically investigate its application for interval breast cancers within the distinct screening landscape of the United States. This geographical distinction is crucial, as there are fundamental differences in screening practices between the U.S. and European healthcare systems.
In the U.S., the dominant screening modality is digital breast tomosynthesis (DBT), commonly known as 3D mammography. Patients typically undergo annual screenings, a more frequent schedule than many European programs. In contrast, European screening initiatives predominantly utilize digital mammography (DM), or 2D mammography, and often recommend screenings every two to three years. These variations in technology and frequency mean that AI algorithms developed and validated in one context may not be directly transferable or equally effective in another without specific adaptation and testing. The UCLA study’s focus on DBT and annual screenings makes its findings particularly relevant for informing U.S. clinical practice and policy.
The retrospective study analyzed a vast dataset, encompassing nearly 185,000 past mammograms performed between 2010 and 2019. This comprehensive review included both DM and DBT images, providing a robust foundation for analysis. From this extensive collection, the research team meticulously focused on 148 cases where a woman had been subsequently diagnosed with interval breast cancer.
To understand why these cancers were not initially detected, radiologists meticulously reviewed each case. The study adapted a classification system, originally developed in Europe, to categorize the types of interval cancers. This system includes:
- Missed reading error: The cancer was visible on the initial mammogram but overlooked by the interpreting radiologist.
- Minimal signs-actionable: Subtle signs of cancer were present on the mammogram, which, in retrospect, should have prompted further investigation.
- Minimal signs-non-actionable: Very subtle signs were present, but their significance was so marginal that they would not typically trigger a recall for further workup.
- True interval cancer: The cancer developed rapidly after a genuinely negative screening, meaning it was not present or detectable at the time of the prior mammogram.
- Occult: The cancer was truly invisible on the mammogram, even in retrospect, and was detected by other means (e.g., clinical exam, ultrasound, MRI).
- Missed due to a technical error: Issues with image acquisition or processing contributed to the cancer not being detected.
AI as a "Second Set of Eyes": Key Findings and Potential
After the human review and classification, researchers applied a commercially available AI software, Transpara, to the initial screening mammograms that preceded the interval cancer diagnoses. The objective was to determine if the AI could detect the subtle signs of cancer that had been missed by human radiologists during their initial interpretation, or at the very least, flag them as suspicious. The Transpara tool operates by scoring each mammogram on a scale of 1 to 10 for cancer risk, with a score of 8 or higher prompting a flag as potentially concerning.
The results were illuminating, demonstrating AI’s capacity to act as a valuable "second set of eyes." The software successfully identified a significant proportion of the interval cancers that had subtle or missed signs on the original mammograms. This ability to flag even faint anomalies suggests that AI could augment human radiologists’ performance, particularly in high-volume screening environments where fatigue or cognitive biases can sometimes lead to oversights.
The Nuances of AI: Promises and Practicalities
Despite the exciting potential, the study also provided a realistic appraisal of current AI limitations. Dr. Hannah Milch, an assistant professor of Radiology at the David Geffen School of Medicine and senior author of the study, highlighted these complexities. "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 commented. She pointed to specific challenges, such as the AI’s performance with occult cancers – those genuinely invisible on mammography. "For example, despite being invisible on mammography, the AI tool still flagged 69% of the screening mammograms that had occult cancers. However, when we looked at the specific areas on the images that the AI marked as suspicious, the AI did not do as good of a job and only marked the actual cancer 22% of the time."
This finding is critical. While AI’s ability to identify a high percentage of cases that later proved to be occult cancers is intriguing, its lack of precision in pinpointing the exact location of these invisible lesions presents a practical hurdle. If an AI flags a mammogram as suspicious but cannot accurately localize the anomaly, it could lead to unnecessary further investigations, increased patient anxiety, and a potential drain on healthcare resources. This underscores the current understanding that AI in diagnostics is best utilized as an assistive tool, not a replacement for human expertise.
The need for larger prospective studies is evident to fully comprehend how radiologists would integrate AI into their daily practice. Key questions remain, such as how clinicians should interpret and act upon AI flags, especially when the AI identifies suspicious areas not visible to the human eye, and when the AI’s localization accuracy is imperfect. These are not merely technical challenges but also involve complex workflow adjustments, medico-legal considerations, and the development of clear clinical guidelines.
Broader Implications for Breast Cancer Screening and Patient Outcomes
The implications of this research extend far beyond the technical capabilities of AI. If successfully integrated, AI-powered screening could usher in a new era of proactive cancer detection. For patients, this translates to the profound benefit of earlier diagnoses, which often correlate with less aggressive treatment regimens, improved quality of life during and after treatment, and significantly higher survival rates. The potential reduction in interval cancers by 30% could save countless lives and mitigate the physical and emotional toll of advanced-stage disease.
For radiologists, AI could serve as an invaluable decision-support tool, potentially reducing diagnostic fatigue and enhancing accuracy. In a world where radiologist workloads are increasing and the complexity of medical imaging continues to grow, AI offers a promising avenue for augmentation, allowing human experts to focus their attention more effectively on genuinely complex cases. However, it also necessitates new training paradigms for radiologists to effectively interact with and interpret AI-generated insights.
The Road Ahead: Integration, Regulation, and Future Research
Dr. Yu summarized the nuanced perspective on AI’s role: "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 vision suggests a future where AI helps eliminate the "missed reading" and "minimal signs" categories, leaving primarily those cancers that truly emerge 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 journey from promising research to widespread clinical adoption is complex. It involves rigorous prospective studies to validate these retrospective findings in real-world clinical settings, addressing issues of generalizability across diverse patient populations and imaging equipment. Regulatory bodies, such as the U.S. Food and Drug Administration (FDA), play a crucial role in evaluating the safety and efficacy of AI algorithms, a process that is still evolving for medical AI. Furthermore, ethical considerations surrounding data privacy, algorithmic bias, and accountability for AI-assisted diagnoses must be carefully addressed to build trust among patients and clinicians.
The development and integration of AI in breast cancer screening also necessitate a collaborative effort among researchers, clinicians, technology developers, and policymakers. This study, supported in part by the National Institutes of Health, the National Cancer Institute, the Agency for Healthcare Research and Quality, and Early Diagnostics Inc., represents a significant step forward in this collaborative endeavor.
The co-authors, all 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. Their collective expertise underscores the multidisciplinary nature of this groundbreaking research, bringing together radiology, oncology, and computer science to tackle one of the most persistent challenges in cancer diagnostics. As AI continues to mature, its role in transforming healthcare, particularly in early disease detection, promises to become increasingly central, offering hope for a future with fewer advanced cancers and improved patient outcomes globally.