There’s been a great deal of discussion about whether people who recover from coronavirus disease 2019 (COVID-19), have neutralizing antibodies in their bloodstream to guard against another infection. Lots of interesting data continue to emerge, including a recent preprint from researchers at Sherman Abrams Laboratory, Brooklyn, NY . They tested 11,092 people for antibodies in May at a local urgent care facility and found nearly half had long-lasting IgG antibodies, a sign of exposure to the novel coronavirus SARS-CoV-2, the cause of COVID-19. The researchers also found a direct correlation between the severity of a person’s symptoms and their levels of IgG antibodies.
This study and others remind us of just how essential antibody tests will be going forward to learn more about this challenging pandemic. These assays must have high sensitivity and specificity, meaning there would be few false negatives and false positives, to tell us more about a person’s exposure to SARS-CoV-2. While there are some good tests out there, not all are equally reliable.
Recently, I had a chance to discuss COVID-19 antibody tests, also called serology tests, with Dr. Norman “Ned” Sharpless, Director of NIH’s National Cancer Institute (NCI). Among his many talents, Dr. Sharpless is an expert on antibody testing for COVID-19. You might wonder how NCI got involved in COVID-19 testing. Well, you’re going to find out. Our conversation took place while videoconferencing, with him connecting from North Carolina and me linking in from my home in Maryland. Here’s a condensed transcript of our chat:
Collins: Ned, thanks for joining me. Maybe we should start with the basics. What are antibodies anyway?
Sharpless: Antibodies are proteins that your body makes as part of the learned immune system. It’s the immunity that responds to a bacterium or a virus. In general, if you draw someone’s blood after an infection and test it for the presence of these antibodies, you can often know whether they’ve been infected. Antibodies can hang around for quite a while. How long exactly is a topic of great interest, especially in terms of the COVID-19 pandemic. But we think most people infected with coronavirus will make antibodies at a reasonably high level, or titer, in their peripheral blood within a couple of weeks of the infection.
Collins: What do antibodies tell us about exposure to a virus?
Sharpless: A lot of people with coronavirus are infected without ever knowing it. You can use these antibody assays to try and tell how many people in an area have been infected, that is, you can do a so-called seroprevalence survey.
You could also potentially use these antibody assays to predict someone’s resistance to future infection. If you cleared the infection and established immunity to it, you might be resistant to future infection. That might be very useful information. Maybe you could make a decision about how to go out in the community. So, that part is of intense interest as well, although less scientifically sound at the moment.
Collins: I have a 3D-printed model of SARS-CoV-2 on my desk. It’s sort of a spherical virus that has spike proteins on its surface. Do the antibodies interact with the virus in some specific ways?
Sharpless: Yes, antibodies are shaped like the letter Y. They have two binding domains at the head of each Y that will recognize something about the virus. We find antibodies in the peripheral blood that recognize either the virus nucleocapsid, which is the structural protein on the inside; or the spikes, which stick out and give coronavirus its name. We know now that about 99 percent of people who get infected with the virus will develop antibodies eventually. Most of those antibodies that you can detect to the spike proteins will be neutralizing, which means they can kill the virus in a laboratory experiment. We know from other viruses that, generally, having neutralizing antibodies is a promising sign if you want to be immune to that virus in the future.
Collins: Are COVID-19 antibodies protective? Are there reports of people who’ve gotten better, but then were re-exposed and got sick again?
Sharpless: It’s controversial. People can shed the virus’s nucleic acid [genetic material], for weeks or even more than a month after they get better. So, if they have another nucleic acid test it could be positive, even though they feel better. Often, those people aren’t making a lot of live virus, so it may be that they never stopped shedding the virus. Or it may be that they got re-infected. It’s hard to understand what that means exactly. If you think about how many people worldwide have had COVID-19, the number of legitimate possible reinfection cases is in the order of a handful. So, it’s a pretty rare event, if it happens at all.
Collins: For somebody who does have the antibodies, who apparently was previously infected, do they need to stop worrying about getting exposed? Can they can do whatever they want and stop worrying about distancing and wearing masks?
Sharpless: No, not yet. To use antibodies to predict who’s likely to be immune, you’ve got to know two things.
First: can the tests actually measure antibodies reliably? I think there are assays available to the public that are sufficiently good for asking this question, with an important caveat. If you’re trying to detect something that’s really rare in a population, then any test is going to have limitations. But if you’re trying to detect something that’s more common, as the virus was during the recent outbreak in Manhattan, I think the tests are up to the task.
Second: does the appearance of an antibody in the peripheral blood mean that you’re actually immune or you’re just less likely to get the virus? We don’t know the answer to that yet.
Collins: Let’s be optimistic, because it sounds like there’s some evidence to support the idea that people who develop these antibodies are protected against infection. It also sounds like the tests, at least some of them, are pretty good. But if there is protection, how long would you expect it to last? Is this one of those things where you’re all set for life? Or is this going to be something where somebody’s had it and might get it again two or three years from now, because the immunity faded away?
Sharpless: Since we have no direct experience with this virus over time, it’s hard to answer. The potential for this cell-based humoral immunity to last for a while is there. For some viruses, you have a long-lasting antibody protection after infection; for other viruses, not so much.
So that’s the unknown thing. Is immunity going to last for a while? Of course, if one were to bring up the topic of vaccines, that’s very important to know, because you would want to know how often one would have to give that vaccine, even under optimal circumstances.
Collins: Yes, our conversation about immunity is really relevant to the vaccines we’re trying to develop right now. Will these vaccines be protective for long periods of time? We sure hope so, but we’ve got to look carefully at the issue. Let’s come back, though, to the actual performance of the tests. The NCI has been right in the middle of trying to do this kind of validation. How did that happen, and how did that experience go?
Sharpless: Yes, I think one might ask: why is the National Cancer Institute testing antibody kits for the FDA? It is unusual, but certainly not unheard of, for NCI to take up problems like this during a time of a national emergency. During the HIV era, NCI scientists, along with others, identified the virus and did one of the first successful compound screens to find the drug AZT, one of the first effective anti-HIV therapies.
NCI’s Frederick National Lab also has a really good serology lab that had been predominantly working on human papillomavirus (HPV). When the need arose for serologic testing a few months ago, we pivoted that lab to a coronavirus serology lab. It took us a little while, but eventually we rounded up everything you needed to create positive and negative reference panels for antibody testing.
At that time, the FDA had about 200 manufacturers making serology tests that hoped for approval to sell. The FDA wanted some performance testing of those assays by a dispassionate third party. The Frederick National Lab seemed like the ideal place, and the manufacturers started sending us kits. I think we’ve probably tested on the order of 20 so far. We give those data back to the FDA for regulatory decision making. They’re putting all the data online.
Collins: How did it look? Are these all good tests or were there some clunkers?
Sharpless: There were some clunkers. But we were pleased to see that some of the tests appear to be really good, both in our hands and those of other groups, and have been used in thousands of patients.
There are a few tests that have sensitivities that are pretty high and specificities well over 99 percent. The Roche assay has a 99.8 percent specificity claimed on thousands of patients, and for the Mt. Sinai assay developed and tested by our academic collaborators in a panel of maybe 4,000 patients, they’re not sure they’ve ever had a false positive. So, there are some assays out there that are good.
Collins: There’s been talk about how there will soon be monoclonal antibodies directed against SARS-CoV-2. How are those derived?
Sharpless: They’re picked, generally, for appearing to have neutralizing activity. When a person makes antibodies, they don’t make one antibody to a pathogen. They make a whole family of them. And those can be individually isolated, so you can know which antibodies made by a convalescent individual really have virus-neutralizing capacity. That portion of the antibody that recognizes the virus can be engineered into a manufacturing platform to make monoclonal antibodies. Monoclonal means one kind of antibody. That approach has worked for other infectious diseases and is an interesting idea here too.
Collins: I can say a bit about that, because we are engaged in a partnership with industry and FDA called Accelerating COVID-19 Therapeutic Interventions and Vaccines (ACTIV). One of the hottest ideas right now is monoclonal antibodies, and we’re in the process of devising a master protocol, one for outpatients and one for inpatients.
Janet Woodcock of Operation Warp Speed tells me 21 companies are developing monoclonal antibodies. While doing these trials, we’d love to do comparisons, which is why it’s good to have an organization like ACTIV to bring everybody together, making sure you’re using the same endpoints and the same laboratory measures. I think that, maybe even by late summer, we might have some results. For people who are looking at what’s the next most-hopeful therapeutic option for people who are really sick with COVID-19, so far we have remdesivir. It helps, but it’s not a home run. Maybe monoclonal antibodies will be the next thing that really gives a big boost in survival. That would be the hope.
Ned, let me ask you one final question about herd, or group, immunity. One hears a bit about that in terms of how we are all going to get past this COVID-19 pandemic. What’s that all about?
Sharpless: Herd immunity is when a significant portion of the population is immune to a pathogen, then that pathogen will die out in the population. There just aren’t enough susceptible people left to infect. What the threshold is for herd immunity depends on how infectious the virus is. For a highly infectious virus, like measles, maybe up to 90 percent of the population must be immune to get herd immunity. Whereas for other less-infectious viruses, it may only be 50 percent of the population that needs to be immune to get herd immunity. It’s a theoretical thing that makes some assumptions, such as that everybody’s health status is the same and the population mixes perfectly every day. Neither of those are true.
How well that actual predictive number will work for coronavirus is unknown. The other thing that’s interesting is a lot of that work has been based on vaccines, such as what percentage do you have to vaccinate to get herd immunity? But if you get to herd immunity by having people get infected, so-called natural herd immunity, that may be different. You would imagine the most susceptible people get infected soonest, and so the heterogeneity of the population might change the threshold calculation.
The short answer is nobody wants to find out. No one wants to get to herd immunity for COVID-19 through natural herd immunity. The way you’d like to get there is with a vaccine that you then could apply to a large portion of the population, and have them acquire immunity in a more safe and controlled manner. Should we have an efficacious vaccine, this question will loom large: how many people do we need to vaccinate to really try and protect vulnerable populations?
Collins: That’s going to be a really critical question for the coming months, as the first large-scale vaccine trials get underway in July, and we start to see how they work and how successful and safe they are. But I’m also worried seeing some reports that 1 out of 5 Americans say they wouldn’t take a vaccine. It would be truly a tragedy if we have a safe and effective vaccine, but we don’t get enough uptake to achieve herd immunity. So, we’ve got some work to do on all fronts, that’s for sure.
Ned, I want to thank you for sharing all this information about antibodies and serologies and other things, as well as thank you for your hard work with all your amazing NCI colleagues.
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.
Caption: Triple immunohistochemical stained oral squamous cell carcinoma: nuclei in brown, cytoplasm in red, and cytoplasmic membranes in blue green. Credit: Alfredo A. Molinolo, Oral and Pharyngeal Cancer Branch, National Institute of Dental and Craniofacial Research, NIH
An exciting new era in cancer research is emerging, called precision oncology. It builds on decades of research establishing that cancers start with glitches in the genome, the cell’s instruction book. Researchers have now identified numerous ways that mutations in susceptible genes can drive the cancer process. Knowing where and how to look for them brings greater precision to diagnosing cancers and gives doctors key clues about which treatments might work and which ones won’t.
To build a firmer evidence base for precision oncology, more and more cancer genomes, from many different body sites, must be analyzed for clues about the drivers of the malignant process. That’s why it’s always exciting to see a new genomic analysis that adds substantially to our understanding of a common tumor. The latest to appear, published online at the journal Nature, comes from an NIH-supported study on the most common type of head and neck cancer, called squamous cell carcinoma. The technologically advanced analysis confirms that many previously suspected genes do indeed play a role in head and neck cancer. But that’s not all. The new data also identify several previously unknown subtypes of this cancer. The first descriptions of the abnormal molecular wiring in these subtypes are outlined, suggesting possible strategies to neutralize or destroy the cancer cells. That’s potentially good news to help guide and inform the treatment of the estimated 55,000 Americans who are diagnosed with a head and neck cancer each year.