Posted on by Dr. Francis Collins
My last post highlighted the use of artificial intelligence (AI) to create an algorithm capable of detecting 10 different kinds of irregular heart rhythms. But that’s just one of the many potential medical uses of AI. In this post, I’ll tell you how NIH researchers are pairing AI analysis with smartphone cameras to help more women avoid cervical cancer.
In work described in the Journal of the National Cancer Institute , researchers used a high-performance computer to analyze thousands of cervical photographs, obtained more than 20 years ago from volunteers in a cancer screening study. The computer learned to recognize specific patterns associated with pre-cancerous and cancerous changes of the cervix, and that information was used to develop an algorithm for reliably detecting such changes in the collection of images. In fact, the AI-generated algorithm outperformed human expert reviewers and all standard screening tests in detecting pre-cancerous changes.
Nearly all cervical cancers are caused by the human papillomavirus (HPV). Cervical cancer screening—first with Pap smears and now also using HPV testing—have greatly reduced deaths from cervical cancer. But this cancer still claims the lives of more than 4,000 U.S. women each year, with higher frequency among women who are black or older . Around the world, more than a quarter-million women die of this preventable disease, mostly in poor and remote areas .
These troubling numbers have kept researchers on the lookout for low cost, but easy-to-use, tools that could be highly effective at detecting HPV infections most likely to advance to cervical cancer. Such tools would also need to work well in areas with limited resources for sample preparation and lab analysis. That’s what led to this collaboration involving researchers from NIH’s National Cancer Institute (NCI) and Global Good, Bellevue, WA, which is an Intellectual Ventures collaboration with Bill Gates to invent life-changing technologies for the developing world.
Global Good researchers contacted NCI experts hoping to apply AI to a large dataset of cervical images. The NCI experts suggested an 18-year cervical cancer screening study in Costa Rica. The NCI-supported project, completed in the 1990s, generated nearly 60,000 cervical images, later digitized by NIH’s National Library of Medicine and stored away safely.
The researchers agreed that all these images, obtained in a highly standardized way, would serve as perfect training material for a computer to develop a detection algorithm for cervical cancer. This type of AI, called machine learning, involves feeding tens of thousands of images into a computer equipped with one or more high-powered graphics processing units (GPUs), similar to something you’d find in an Xbox or PlayStation. The GPUs allow the computer to crunch large sets of visual data in the images and devise a set of rules, or algorithms, that allow it to learn to “see” physical features.
Here’s how they did it. First, the researchers got the computer to create a convolutional neural network. That’s a fancy way of saying that they trained it to read images, filter out the millions of non-essential bytes, and retain the few hundred bytes in the photo that make it uniquely identifiable. They fed 1.28 million color images covering hundreds of common objects into the computer to create layers of processing ability that, like the human visual system, can distinguish objects and their qualities.
Once the convolutional neural network was formed, the researchers took the next big step: training the system to see the physical properties of a healthy cervix, a cervix with worrisome cellular changes, or a cervix with pre-cancer. That’s where the thousands of cervical images from the Costa Rican screening trial literally entered the picture.
When all these layers of processing ability were formed, the researchers had created the “automated visual evaluation” algorithm. It went on to identify with remarkable accuracy the images associated with the Costa Rican study’s 241 known precancers and 38 known cancers. The algorithm’s few minor hiccups came mainly from suboptimal images with faded colors or slightly blurred focus.
These minor glitches have the researchers now working hard to optimize the process, including determining how health workers can capture good quality photos of the cervix with a smartphone during a routine pelvic exam and how to outfit smartphones with the necessary software to analyze cervical photos quickly in real-world settings. The goal is to enable health workers to use a smartphone or similar device to provide women with cervical screening and treatment during a single visit.
In fact, the researchers are already field testing their AI-inspired approach on smartphones in the United States and abroad. If all goes well, this low-cost, mobile approach could provide a valuable new tool to help reduce the burden of cervical cancer among underserved populations.
The day that cervical cancer no longer steals the lives of hundreds of thousands of women a year worldwide will be a joyful moment for cancer researchers, as well as a major victory for women’s health.
 An observational study of Deep Learning and automated evaluation of cervical images for cancer screening. Hu L, Bell D, Antani S, Xue Z, Yu K, Horning MP, Gachuhi N, Wilson B, Jaiswal MS, Befano B, Long LR, Herrero R, Einstein MH, Burk RD, Demarco M, Gage JC, Rodriguez AC, Wentzensen N, Schiffman M. J Natl Cancer Inst. 2019 Jan 10. [Epub ahead of print]
 “Study: Death Rate from Cervical Cancer Higher Than Thought,” American Cancer Society, Jan. 25, 2017.
 “World Cancer Day,” World Health Organization, Feb. 2, 2017.
Posted on by Dr. Francis Collins
Forty-five years ago, when I was a first-year medical student, a lecturer introduced me to a young man with sickle cell disease (SCD). Sickle cell disease is the first “molecular disease”, with its cause having been identified decades ago. That helped me see the connection between the abstract concepts of molecular genetics and their real-world human consequences in a way no textbook could. In fact, it inspired some of my earliest research on human hemoglobin disorders, which I conducted as a postdoctoral fellow.
Today, I’m heartened to report that, thanks to decades of biomedical advances, we stand on the verge of a cure for SCD. While at the American Society of Hematology meeting in San Diego last week, I was excited to be part of a discussion about how the tools and technologies arising from the Human Genome Project are accelerating the quest for cures.
The good news at the meeting included some promising, early results from human clinical trials of SCD gene therapies, including new data from the NIH Clinical Center. Researchers also presented very encouraging pre-clinical work on how gene-editing technologies, such as CRISPR, can be used in ways that may open the door to curing everyone with SCD. In fact, just days before the meeting, the first clinical trial for a CRISPR approach to SCD opened.
One important note: the gene editing research aimed at curing SCD is being done in non-reproductive (somatic) cells. The NIH does not support the use of gene editing technologies in human embryos (germline). I recently reiterated our opposition to germline gene editing, in response to an unethical experiment by a researcher in China who claims to have used CRISPR editing on embryos to produce twin girls resistant to HIV.
SCD affects approximately 100,000 people in the United States, and another 20 million worldwide, mostly in developing nations. This inherited, potentially life-threatening disorder is caused by a specific point mutation in a gene that codes for the beta chain of hemoglobin, a molecule found in red blood cells that deliver oxygen throughout the body. In people with SCD, the mutant hemoglobin forms insoluble aggregates when de-oxygenated. As a result the red cells assume a sickle shape, rather than the usual donut shape. These sickled cells clump together and stick in small blood vessels, resulting in severe pain, blood cell destruction, anemia, stroke, pulmonary hypertension, organ failure, and much too often, early death.
The need for a widespread cure for SCD is great. Since 1998, doctors have used a drug called hydroxyurea to reduce symptoms, but it can cause serious side effects and increase the risk of certain cancers. Blood transfusions can also ease symptoms in certain instances, but they too come with risks and complications. At the present time, the only way to cure SCD is a bone marrow transplant. However, transplants are not an option for many patients due to lack of matched marrow donors.
The good news is that novel genetic approaches have raised hopes of a widespread cure for SCD, possibly even within five to 10 years. So, in September, NIH’s National Heart, Lung, and Blood Institute launched the Cure Sickle Cell Initiative to accelerate development of the most promising of these next generation of therapies
At the ASH meeting, that first wave of this progress was evident. A team led by NHLBI’s John Tisdale, in collaboration with Bluebird bio, Cambridge, MA, was among the groups that presented impressive early results from human clinical trials testing novel gene replacement therapies for SCD. In the NIH trial, researchers removed blood precursor cells, called hematopoietic stem cells (HSCs), from a patient’s own bone marrow or bloodstream and used a harmless virus to insert a sickle-resistant hemoglobin gene. Then, after a chemotherapy infusion to condition the patient’s existing bone marrow, they returned the corrected cells to the patient.
So far, nine SCD patients have received the most advanced form of the experimental gene therapy, and Tisdale presented data on those who were farthest out from treatment [1,2]. His team found that in the four patients who were at least six months out, levels of gene therapy-derived hemoglobin were found to equal or exceed their levels of SCD hemoglobin.
Very cool science, but what does this mean for SCD patients’ health and well-being? Well, none of the gene therapy trial participants have required a blood transfusion during the follow-up period. In addition, improvements were seen in their hemoglobin levels and key markers of blood-cell destruction (total bilirubin concentration, lactate dehydrogenase, and reticulocyte counts) compared to baseline. Most importantly, in the years leading up to the clinical trial, all of the participants had experienced frequent painful sickle crises, in which sickled cells blocked their blood vessels. No such episodes were reported among the participants in the months after they received the gene therapy.
Researchers did report that one patient receiving this form of gene therapy developed myelodysplastic syndrome (MDS), a serious condition in which the blood-forming cells in the bone marrow become abnormal. However, there is no indication that the gene replacement technology itself caused the problem, and MDS has previously been linked to the chemotherapy drugs used in conditioning regimens before bone marrow transplants.
The NIH trial is just one of several clinical trials for SCD that are using viral vectors to deliver a variety of genes with therapeutic potential. Other trials actively recruiting are led by researchers at Boston Children’s Hospital, Cincinnati Children’s Hospital, and the University of California, Los Angeles.
While it’s hoped that genes inserted by viral vectors will provide long-lasting or curative treatment, other researchers are betting that new gene-editing technologies, such as CRISPR, will offer the best chance for developing a widespread cure for SCD. One strategy being eyed by these “gene editors” is to correct the SCD mutation, replacing it with a normal gene. Another strategy involves knocking out certain DNA sequences to reactivate production of fetal hemoglobin (HbF).
The HbF protein is produced in the developing fetus to give it better access to oxygen from the mother’s bloodstream. But shortly after birth, the production of fetal hemoglobin shuts down, and the adult form kicks in. Adults normally have very low levels of fetal hemoglobin, which makes sense. However, from genome-wide association studies of human genetic variation, we know that that actual levels of HbF are under genetic control.
A major factor has been mapped to the BCL11A gene, which has subsequently been found to be a master mediator for the fetal to adult hemoglobin switch. Specifically, variations in a red cell specific enhancer of BCL11A affect an adult’s level of HbF— levels of BCL11A protein lead to higher amounts of fetal hemoglobin. Furthermore, it’s been known for some time that rare individuals keep on producing relatively high levels of hemoglobin into adulthood. If people with SCD happen to have a rare mutation that keeps fetal hemoglobin production active in adulthood (the first of these was found as part of my postdoctoral research), their SCD symptoms are much less severe.
Currently, two groups—CRISPR Therapeutics/Vertex Pharmaceuticals and Sangamo Therapeutics/Bioverativ—are gearing up to begin the first U.S. human clinical trials of gene-editing for SCD within the next few months. While they employ different technologies, both approaches involve removing a patient’s HSCs, using gene editing to knock out the BCL11A red cell enhancer, and then returning the gene-edited cells to the patient. The hope is that the gene-edited cells will greatly boost fetal hemoglobin production, thereby offsetting the effects of SCD.
All of this is exciting news for the 100,000 people living in the United States who have SCD. But what about the 300,000 babies born with SCD every year in other parts of the world, mostly in low- and middle-income countries?
The complicated, high-tech procedures that I just described may not be practical for a very long time in places like sub-Saharan Africa. That’s one reason why NIH recently launched a new effort to speed the development of safe, effective genome-editing approaches that could be delivered directly into a patient’s body (in vivo), perhaps by infusion of the CRISPR gene editing apparatus. Recent preclinical experiments demonstrating the promise of in vivo gene editing for Duchenne muscular dystrophy make me optimistic that NIH’s Somatic Cell Genome Editing Program, which is hosting its first gathering of investigators this week, will be able to develop similar approaches for SCD and many other conditions.
While moving forward in this fast-paced field, it is important that we remain ethical, but also remain bold on behalf of the millions of patients with genetic diseases who are still waiting for a cure. We must continue to assess and address the very serious ethical concerns raised by germline gene editing of human embryos, which will irreversibly alter the DNA instruction book of future children and affect future generations. I continue to argue that we are not ready to undertake such experiments.
But the use of gene editing to treat, perhaps even to cure, children and adults with genetic diseases, by correcting the mutation in their relevant tissues (so-called somatic cell gene editing), without risk of passing those changes on to a future generation, holds enormous promise. Somatic cell gene editing is associated with ethical issues that are much more in line with decades of deep thinking about benefits and risks of therapeutic trials.
Finally, we must recognize that somatic cell gene editing is a profoundly promising approach not only for people with SCD, but for all who are struggling with the thousands of diseases that still have no treatments or cures. Real hope for cures has never been greater.
 NIH researcher presents encouraging results for gene therapy for severe sickle cell disease. NIH News Release. December 4, 2018
Sickle Cell Disease (National Heart, Lung, and Blood Institute/NIH)
Cure Sickle Cell Initiative (NHLBI)
John Tisdale (NHLBI)
Somatic Cell Genome Editing Program (Common Fund/NIH)
NIH Support: National Heart, Lung, and Blood Institute; Common Fund