Effect of breast tissue density on cell-free orphan non-coding RNAs secreted by breast cancers

Background

• Breast cancer is the second leading cause of cancer death in women. Approximately 1 in 8 women will be impacted by breast cancer in their lifetimes¹
• Mammography is the current standard imaging screening modality for breast cancer
• Earlier detection of breast cancer is crucial for improved patient outcomes but cannot always be achieved through mammography
• Confounding issues such as breast density, which masks the appearance of tumors and affects more than 40% of women can result in false negatives and delayed diagnosis
• Nearly half of all women have dense breasts, which increases their risk for breast cancer and means that mammograms may not work as well for them
• Breast density has been identified as one of the strongest independent risk factors for the development of breast cancer aside from age and genetics². Women with dense breast tissue are at 2X higher risk for interval cancers
• A high-performing blood test could complement mammography, especially for women with dense breasts
• Orphan non-coding RNA (oncRNAs) are a novel category of small RNAs that are present in tumors and largely absent in healthy tissue
• Tumor-derived oncRNAs are present in the blood as cell-free RNA (cfRNA) that are stable, resilient and continually released from living cells

Goals

• Exai Bio has previously demonstrated high sensitivity and specificity for early detection of invasive breast cancer using a blood-based liquid biopsy assay that leverages advanced AI to detect cancer signals from oncRNAs⁵
• Exai Bio’s liquid biopsy assay using deep generative AI models to analyze circulating oncRNAs also achieves accurate detection of early-stage non-small cell lung cancer⁶
• If oncRNA presence is unaffected by breast tissue density, this assay could offer an alternative to screening mammogram for women with a finding of dense breasts on a previous screening

Methods

• We prospectively analyzed samples from 68 patients with known breast density and with newly diagnosed breast cancer of different stages`
• We first assayed the cell-free small RNA content of every sample at an average depth of 50 million 50-bp single-end reads. Reads were then annotated using our proprietary RNA database to quantify oncRNA burden (expressed as the count of oncRNA reads per million mapped,unique reads). A previously developed oncRNA-based AI model for breast cancer detection was applied to calculate an oncRNA score for every sample
• We compared both the oncRNA burden and the oncRNA score between breast density groups of fatty breast tissue (BI-RADS A/B) vs. dense breast tissue (BI-RADS C/D) using the Mann-Whitney (MW) U test and the Student’s t-test

oncRNA Profiling & AI-Driven Model for Breast Cancer Prediction

Figure 2. Schematic of oncRNA Profiling and Modeling Pipeline

Our generative AI model utilizes tumor-derived oncRNAs discovered in TCGA breast cancer samples for downstream applications in serum. The model is trained to predict presence or absence of invasive breast cancer while minimizing the impact of technical variations such as supplier and batch effects.

Cohort Demographics

• Of 68 women with breast cancer, 25 had fatty breast tissue and 43 had dense breast tissue
• Cohort characteristics were similar between women with fatty and dense breast tissue (cohort mean age: 58.9±11.6yr, BMI: 30.11±7.57, 74% Caucasian/White, 18% Hispanic, and 9% Black
• In the overall cohort, all cancer stages were represented (I: n = 36, II: n = 18, III: n = 10, IV: n = 4), T-stage (T1: n = 36, T2: n = 23, T3: n = 4, T4: n = 5). The proportion of early stage (I/II) patients is comparable between women in the two breast density groups: fatty breast tissue, (21/25 = 84%) and dense breast tissue (33/43 = 77%)

Results

Table 1. Characteristics of the small RNA sequencing data generated using serum samples from invasive breast cancer patients.

Figure 3. Exai Bio’s oncRNA-based AI model score and oncRNA burden among invasive breast cancer patients stratified by BI-RADS density level (A/B versus C/D). ns: not significant (p-value > 0.01).

Table 2. Summary statistics of the AI model score and oncRNA burden per BI-RADS density group.

Table 3. Summary of multivariate regression analysis between AI model score/oncRNA burdenand BI-RADS density group, cancer stage group, patient age and BMI

Conclusions

• We did not observe differences in the distribution of oncRNA burden or score between women with dense breasts and women with non-dense breasts. These results were unchanged even after adjusting for cancer stage, age, and BMI
• Unlike mammograms, the use of oncRNAs for breast cancer detection is not significantly affected by breast density. Further research will focus on oncRNA-based early breast cancer detection in women with dense breasts
• In April 2024, the USPSTF stated that additional screening methods (beyond mammography) might help women with dense breasts (>40% of the screening population) find cancers earlier, and they are urgently calling for more research⁷
• Exai's high performing, minimally invasive blood-based oncRNA platform has the potential to improve breast cancer diagnosis over currently available methods in a cost-effective manner