BOSTON—Thanks to the speed and interpretation abilities of artificial intelligence (AI) platforms, they can search through massive databases and cross-reference related literature much faster than existing technologies. As such, AI is seeing increased use in the industry as companies gain more examples of its potential in identifying new avenues and new targets for drug discovery. But a team from the Dana-Farber Cancer Institute and Brigham and Women’s Hospital (BWH) is looking to use AI even earlier in the disease process—not to find new treatments, but for diagnostics.
The researchers singled out a network of seven circulating microRNAs that are associated with the risk of ovarian cancer and are detectable in blood samples. Their work was published online in eLife under the title “Diagnostic potential for a serum miRNA neural network for detection of ovarian cancer.”
While early diagnostic options for ovarian cancer already exist, including ultrasound or the detection of the protein CA125, these tests have high false-positive rates, and using them to detect early-stage cancer in clinical trials has not led to a significant impact on survival rates. For those whose cancer is found early, survival rates are much higher than traditional rates, as most ovarian cancer diagnoses come when the cancer is already at an advanced stage and only around 25 percent of patients survive for five years after diagnosis.
According to the eLife paper, “miRNAs have several advantages over protein measures: (1) PCR amplifies detection of rare transcripts in blood; (2) all miRNAs use the same units of measure, easing incorporation into multiplexed panels; and (3) miRNAs play a critical role in ovarian cancer biology, whereas the function of CA125 is unknown (Deb et al., 2017; Katz et al., 2015). Moreover, non-invasive sampling of circulating miRNAs has a clear advantage over analytes obtained through biopsy.”
The Dana-Farber/BWH team found that ovarian cancer cells and normal cells present with different microRNA profiles. microRNAs circulate in the blood, which enables their measurement via serum samples. The microRNA levels of 135 women who had not had surgery or chemotherapy were measured via blood samples to produce a “training set” for a computer program, so that it would know how to differentiate between ovarian cancer, benign tumors, non-invasive tumors and healthy tissue. The model that was most capable of differentiating benign tissue from cancerous tissue is known as a neural network model.
“This project exemplifies the synergy of the two institutes—Dana-Farber and Brigham and Women’s—and the power of clinicians working closely with lab-based scientists. My lab has been working on miRNAs for a decade, and when Kevin [lead author Dr. Kevin Elias of BWH’s Department of Obstetrics and Gynecology] came to us with the patient samples, it was a no-brainer to initiate this project,” said Dr. Dipanjan Chowdhury, senior author of the study and chief of the Division of Radiation and Genomic Stability in the Department of Radiation Oncology at Dana-Farber.
That sequencing model was then applied in a set of 44 women to evaluate its accuracy. When that confirmed its accuracy, it was applied in several patient sample sets, with a total of 859 patient samples used to evaluate the model’s sensitivity and specificity. The neural network model proved vastly more accurate and effective than ultrasound—while less than 5 percent of abnormal test results detected with ultrasound proved to be ovarian cancer, nearly 100 percent of abnormal results detected with the new model represented ovarian cancer.
Following the promising sample results, the team deployed the microRNA test to predict the diagnoses of 51 patients presenting for surgical care in Lodz, Poland, in which 91.3 percent of the abnormal test results were genuine cases of ovarian cancer. The test’s negative results correctly predicted absence of cancer approximately 80 percent of the time, similar to the accuracy of a Pap smear test.
The next step in this work will be to monitor how an individual’s microRNA signature changes over time as ovarian cancer risk increases. This will require prospectively collected longitudinal samples from women who can be followed over a period of time.
“The key is that this test is very unlikely to misdiagnose ovarian cancer and give a positive signal when there is no malignant tumor. This is the hallmark of an effective diagnostic test,” said Chowdhury.
The team also looked for evidence of biological relevance for the distinguishing microRNAs. They found changes in the quantity of these microRNAs in blood samples collected before and after surgery, suggesting that the microRNA signal decreases after the cancerous tissue is removed. They also took actual patient samples and imaged the microRNAs in the cancerous cells, demonstrating that the serum signal was coming from the cancerous tissues.
This is not the only cancer partnership for the two organizations in recent months. Dana-Farber and the Brigham and Women’s Cancer Center announced in December that they would be launching a program targeting advanced and aggressive thyroid cancers: the Thyroid Cancer 360 Program. This initiative will focus on improving patient outcomes by offering targeted treatments tailored to individuals’ genomics and biology. The program is led by specialists in the Thyroid Cancer Center, and the Center is enrolling patients for the first clinical immunotherapy trial designed for patients with thyroid cancer.
