Artificial intelligence Can Predict Which Patients Will Develop Breast Cancer Within a Year

A study published in Radiology found that a deep learning artificial intelligence (AI) model from IBM Research can predict the development of malignant breast cancer in patients within a year by linking health records and mammograms. 

In this retrospective study, 52,936 images were obtained in 13,234 women who underwent at least one mammogram and who had health records for at least 1 year before undergoing mammography.  The algorithm was trained on 9,611 mammograms and health records to predict biopsy malignancy and to differentiate between normal from abnormal screenings.

The AI could correctly forecasted respective development of 87 percent and 77 percent of cancerous and non-cancerous cases, and also identified breast cancer in 48 percent of patients that otherwise would have been overlooked, with accuracy comparable to radiologists therefore it has the potential to substantially reduce missed diagnoses of breast cancer.

Artificial Intelligence is Useful in the Interpretation of Screening Mammograms

A study published in Radiology found that breast radiologists had a slight higher diagnostic performance when using artificial intelligence (AI) with no additional time required.

Screening digital mammograms from 240 women (median age, 62 years; range, 39–89 years) performed between 2013 and 2017 were analyzed in this study. The mammograms were interpreted with and without AI support.

The researchers found that the cancer detection improved for all breast densities, and was independent of lesion type, vendor image quality, when radiologists used AI and interestingly did not lengthen interpretation time. The radiologists’ detection slightly improved when using AI support, with the average area under the receiver operating characteristic curve (AUC) increasing form 0,87 to 0.89.  Sensitivity increased with AI support 86% vs. 83%, whereas specificity improved slightly 79% vs. 77%. Reading time per case was for all practical purposes identical  (unaided, 146 seconds; supported by AI, 149 seconds).

The researchers concluded that AI assisted interpreting radiologists and improved their cancer detection at mammography when using AI without adding to the interpetation time.

Breast density assessment varies among radiologists

Sprangue et al published in Annals of Internal Medicine the findings of their study that suggests that radiologists often do not agree on what qualifies as dense breast.

The investigators looked at 216,783 mammograms from 145,123 women aged 40 to 89 years that were interpreted by 83 radiologists in 30 radiology facilities in 4 States.

Overall, 36.9% of mammograms were rated as showing dense breasts. Across radiologists, this percentage ranged from 6.3% to 84.5% (median, 38.7%). Examination of patient subgroups revealed that variation in density assessment across radiologists was pervasive in all but the most extreme patient age and BMI combinations. Among women who had consecutive mammograms interpreted by different radiologists, 17.2% (5909 of 34 271) of them suggested different density rating on the two tests.

The authors concluded because there is wide variation in density assessment across radiologists it is a fact should be carefully considered by providers and policymakers when considering supplemental screening strategies.

U.S is Spending $4 billion Annually Resulting from False Positive Mammograms and Unnecessary Treatments

According to a paper published in Health Affairs by Ong and Mendel the U.S. spends $4 billion a year on unnecessary medical costs due to mammograms that generate false alarms, and on treatment of certain breast tumors unlikely to cause problems.

 nt

They found that costs due to false-positive mammograms and breast cancer over diagnoses among women ages 40–59, based on expenditure data from a major US health care insurance plan for 702,154 women in the years 2011–13.

The average expenditures for each false-positive mammogram, invasive breast cancer, and ductal carcinoma in situ in the twelve months following diagnosis were $852, $51,837 and $12,369, respectively. This translates to a national cost of $4 billion each year.

The cumulative cost is as follows: $2.8 billion resulting from false-positive mammograms and another $1.2 billion attributed to treatment of tumors that grow slowly or not at all, and are unlikely to develop into life-threatening disease during a woman’s lifetime.

Screening has the potential to save lives. However, the economic impact of false-positive mammography results and breast cancer over diagnoses must be considered in the debate about the appropriate populations for screening.

Breast Density should not be the Sole Criterion in Determining High Risk for Breast Cancer

Research by Kerlikowske et al published in the Annals of Internal Medicine suggests that not all women with dense breasts are at high enough risk for breast cancer after a normal mammogram to justify having more diagnostic tests such as ultrasound or MRI.

The researchers used data from Breast Cancer Surveillance Consortium (BCSC) breast imaging facilities.

Studies of 365 426 women aged 40 to 74 years who had 831 455 digital screening mammographic examinations were reviewed.

BI-RADS breast density, BCSC 5-year breast cancer risk, and interval cancer rate (invasive cancer ≤12 months after a normal mammography result) per 1000 mammography examinations. High interval cancer rate was defined as more than 1 case per 1000 examinations.

High interval cancer rates were observed for women with 5-year risk of 1.67% or greater and extremely dense breasts or 5-year risk of 2.50% or greater and heterogeneously dense breasts (24% of all women with dense breasts). The interval rate of advanced-stage disease was highest (>0.4 case per 1000 examinations) among women with 5-year risk of 2.50% or greater and heterogeneously or extremely dense breasts (21% of all women with dense breasts). Five-year risk was low to average (0% to 1.66%) for 51.0% of women with heterogeneously dense breasts and 52.5% with extremely dense breasts, with interval cancer rates of 0.58 to 0.63 and 0.72 to 0.89 case per 1000 examinations, respectively.

In this study half of women had mammograms that showed dense breasts. For most women who had a mammogram, the risk for breast cancer after a normal mammogram was low, even for those who had dense breasts and low 5-year breast cancer risk. Two groups of women had the highest risk for breast cancer after a normal mammogram: those with extremely dense breasts and an intermediate or high 5-year cancer risk, and those who had different patterns of breast density and a high or very high 5-year cancer risk.

Breast density should not be the sole criterion for deciding whether supplemental imaging is justified because not all women with dense breasts have high interval cancer rates. BCSC 5-year risk combined with BI-RADS breast density can identify women at high risk for interval cancer.

MRI and Mammography Combined are Effective in Detecting Breast Cancer in Women at High Risk

Chiarelli et al in their JCO article report on Ontario’s Breast Screening Program of women age 30 to 69 years at high risk for breast cancer with annual magnetic resonance imaging (MRI) and digital mammography.

The study cohort consisted of 2,359 women. The following criteria were used to determine eligibility: known BRCA1, BRCA2 mutation, or other gene predisposing to a markedly increased risk of breast cancer, untested first-degree relative of a gene mutation carrier, family history consistent with hereditary breast cancer syndrome and estimated personal lifetime breast cancer risk of 25% or higher, or radiation therapy to the chest before age 30 years.

Digital mammograms were performed with standard craniocaudal and mediolateral oblique projections.  The minimum MRI standards were 1.5 Tesla units, gadolinium injection (0.1 to 0.2 mmol/kg) and a dedicated breast coil. Both breasts were imaged in axial and sagittal planes.  Most of the eligible women (90.7%) had their MRI within a month from their mammograms.  Of the 2,359 eligible women 2,290 were screened.  Of the women screened 2,157 had an MRI and were included in the study as women who had only a mammogram were excluded.

The recall rate of 15% was significantly higher among women who had abnormal MRI alone compared with 6.4% when mammogram alone was used.  Of the 35 breast cancers detected (16.3 per 1,000), none were detected by mammography alone, while 23 (65.7%) were detected by MRI alone (10.7 per 1,000), and 25 (71%) were detected among women who were known gene mutation carriers (30.8 per 1,000). The positive predictive value of 12.4% for detection was highest when findings from mammogram and MRI were combined.  Overall, the cancer detection rate was significantly higher for invasive cancers (12.6 per 1,000) compared with DCIS (3.7 per 1,000). Cancer detection rates were higher among women age 50 years (23.3 per 1,000) compared with women younger than age 50 years (13.3 per 1,000) and significantly higher among those who were known gene mutation carriers (30.8 per 1,000) compared with those with a family history plus an estimated lifetime cancer risk of 25% (6.9 per 1,000).

The authors conclude that screening with annual MRI combined with mammography is effective and could be implemented into an organized breast screening program for women at high risk for breast cancer as mammograms alone failed to detect early breast cancers. 

An editorial by Dr. Wendie Berg with comments on this topic was published by the Journal of Clinical Oncology.