CHICAGO – Precision medicine is here, but it still has plenty of room for improvement.
"We actually are really in the age of precision," Olufunmilayo Olopade, founding director of the Center for Clinical Cancer Genetics and Global Health at University of Chicago Medicine, said during the closing plenary session of the American Medical Informatics Association (AMIA) Informatics Summit here Thursday.
"We can really target mutations, both germline and somatic, and now we actually have data. We can get granular data, to look where people live. We can know what communities they're coming from, and we can actually integrate the social environment and the social stressors into how we develop targeted therapies and targeted screening and prevention or cancer interception," said Olopade, who started a cancer risk clinic in 1992, long before the first human genome was sequenced.
However, the vision of precision medicine will not be fully realized until researchers and clinicians alike are able to accurately assess patients based on diverse datasets, according to the well-known genetic oncologist, who won a MacArthur Fellowship (aka Genius Grant) in 2005.
"We have different shades [of skin color] and different genetic ancestry," Olopade said. "And then, if we're really thinking about breast cancer, there is heterogeneity. You don't know whether you're going to be at risk for ER-negative breast cancer, so you don't know how often you should go for mammogramming."
That is where genetic heterogeneity comes in to inform clinical decisions.
Olopade said that "one of the most transformative events in my career" involved a 34-year-old Black woman who had been diagnosed with early-stage triple-negative breast cancer.
At that point, mastectomies had fallen out of favor, and Olopade said that she "gained the confidence" of the woman by offering lumpectomy, radiation therapy, or chemotherapy as options rather than mastectomy. A rich family history from five generations of genealogy research led Olopade to order genetic testing on a number of the patient's relatives, from which she discovered a BRCA1 mutation in many family members.
Olopade recommended an oophorectomy to prevent the woman from developing ovarian cancer, but the patient had lost her health insurance when she stopped working due to a side effect of chemotherapy. The woman subsequently developed ovarian cancer and died from it.
That unfortunate case led Olopade to collaborate with Mary-Claire King, the first to identify BRCA1 and to link mutations in that gene to heritable breast cancer. Olopade and King researched how BRCA1 behaves in different populations, not just women of European ancestry, as earlier work had been limited to.
Diversifying genetic datasets has since become a central focus of Olopade's work.
"I began to think about why Black women were dying beyond the fact that they don't have insurance," Olopade said. "The question then became: How do we get to the root of breast cancer heterogeneity by studying Black women across the African diaspora?"
In her work to examine factors that contribute to poor cancer outcomes, Olopade and colleagues discovered that about 10 percent of the genome in populations of African origin was missing from the reference genome they were using to assess breast cancer risk. That has led to informatics algorithms based on incomplete genomes.
"Now that we have the tools to do whole-genome sequencing and look at mutation signatures, we can actually begin to look at particles and we can then really begin to integrate germline genetics with somatic genetics," Olopade said.
There frequently was a delay in reaching a diagnosis in Olopade's clinic and in overseas cases she had seen in nonwhite women younger than 40 who did not fit the genetic profile for breast cancer from medical literature. Clearly, a first mammogram at age 50 was not going to catch many cancers in these populations.
It actually works both ways. "When a woman goes in for a mammogram, is that a missed opportunity if you don't provide them genetic testing?" Olopade wondered aloud during her presentation.
Olopade published research in Cancer Cell last year that discussed genetic variation in nonwhite populations, including those with high rates of genetic admixture. She said that the paper will provide her clinic with the opportunity to develop biomarker-informed clinical trials that integrate germline and tumor genetics alike.
"When you have the right tools to classify tumor-normal and when you sequence whole genomes and you actually have [information] about the whole genome, then you are able to do better classification," Olopade said.
A key turning point in the diversification of breast cancer genomics for Olopade was the 2019 revelation by Mathew Knowles, father of singers Beyoncé and Solange Knowles, that he had breast cancer related to a BRCA2 mutation.
"A lot of what's been in the literature has been about Jewish women and risk of BRCA1 and BRCA2," Olopade said. The Knowles news showed the world that the literature leaves out so many other populations.
However, Olopade, like so many others, has had trouble extracting proof of mutations from electronic health records. "That's why we need all the informatics people," she told the AMIA audience. "If [data] gets locked into the EHR, no one is going to act on it, and that's what we're finding in our own hospitals."
Ideally, clinical decision support systems, fed by EHRs, would tell clinicians when a patient has a mutation that might warrant further testing. University of Chicago has started offering MRIs to women at high risk for breast cancer to improve their ability to detect tumors, for example.
"I've been really very interested in developing tools to predict, preempt, and prevent attacks," Olopade. "Let's really make a case for population risk stratification and population health management." That will take informatics professionals.
To address the informatics side, Olopade and her daughter, Feyi Olopade Ayodele, founded CancerIQ. Ayodele serves as CEO, while her mother is chief scientific adviser for the Chicago-based firm, which closed a $14 million Series B investment round this month.
Olopade discussed a CancerIQ collaboration with Adventist Health. The Roseville, California-based health system is comprehensively evaluating patients in the primary care setting and within breast imaging centers for their inherited predisposition for cancer through the Adventist Health Early All-Around Detection (AHEAD) program, a multidisciplinary initiative to improve assessment of hereditary cancer risk.
Although only a subset of patients will meet the criteria for genetic testing through this strategy, the program is enabling the organization to better identify those who should be tested according to current guidelines.
Adventist said in 2020 that AHEAD has more tightly integrated genomics into primary care. It has also increased revenue, and potentially will save money and lives in the long term. Eventually, the health system would like to see current guidelines evolve to account for the insufficiency of relying heavily on family history to decide which patients should receive genetic testing.
In an interview with GenomeWeb after her AMIA keynote, Olopade said that genetic screening for cancer risk has to start at the primary care level. "Even primary care providers are embracing home-based testing and community-based interventions, [so] this should be a lot easier to manage," she said.
On the other side, Olopade said that no matter how diverse a dataset or reference genome is, it is not clinically useful if the information does not get to clinicians at the point of care.
Olopade said that her clinic at University of Chicago has built its informatics infrastructure so that genotypes are put into the relational database that drives clinical decisions, but that is a rare instance based on custom programming.
"To get to precision medicine, we're going to need to figure out how to do the genotypes [in EHRs]," Olopade said. "You don't need the whole genome in there. You just need the genotypes. If every patient has five variants that they need to worry about, then put the genotypes of those five variants. That's all you need."