Lanny: There are three things that immediately come to mind. The first is obviously machine learning. I don’t think we have to, and I wouldn’t try to, cover that comprehensively. It impacts everything, and any attempt to be comprehensive is futile, it’s like talking about the impact of molecular biology. But I’m sure we’ll touch on it during at least part of this conversation.

The second is understanding and being able to gain actionable information from individual variation. For example, in oncology, looking at somatic driver mutations and tumor samples is now becoming more and more incorporated. I’ve heard that these kinds of oncogene panels are now being used actually quite routinely by both academic and community physicians. But there is even something above and beyond that, and that was brought home to me even just a few days ago, at the most recent ASH [American Society of Hematology] meeting, where I heard a talk about absolute neutrophil count: A certain blood group type, Duffy null, which is more prevalent in the African American community [than among Caucasian and Asian populations], results in the presence of a lower peripheral neutrophil count. There’s obviously a major attempt to bring more diversity, equity, and inclusion into clinical trials, and that’s something that we applaud and that has to be taken very seriously. But what I found interesting here is that, simply because of our prior lack of understanding of the effect of this blood group in this population, people were being excluded inappropriately; there were concerns that their peripheral neutrophil count made them ineligible for certain clinical trials. And now there’s talk about redoing and resetting the standards of ANC [absolute neutrophil count]. And that by itself, that kind of increased understanding, I think was an interesting thing to have occurred this year and will have ramifications going forward.

Then finally, the fact that in certain areas so many new drugs have been developed. Since 2003, there have probably been at least 20 new drugs developed for patients with myeloma. It has completely changed not only the prognosis for patients with myeloma, but the way clinical trials need to be done and the way the FDA needs to actually think about them. We were beginning to see in 2023—and I think it’s going to play out more in 2024—that the FDA maybe will be challenged to come up with better ways of defining efficacy. While perhaps still using things like overall survival for the safety indication, they may need to begin to consider other kinds of primary or intermediate endpoints, such as measurable residual disease (MRD, a molecular detection of tumor burden), in order to define efficacy in a timelier fashion that allows for more rapid and accelerated drug approvals.

Gary: I think we’re seeing liquid biopsies more and more because we understand the driver mutations a little bit better, thanks to the oncogene field and other fields. We are creating drug targets that allow new targeted therapies to be involved. Also, from a diagnostic point of view, we’re finding that certain liquid biopsy levels predict outcomes; therefore, the concept of minimal [or measurable] residual disease [MRD] is alive and well, and it should lead to improvements in patient outcome. So, I’d say that in addition to what we get from liquid biopsies (primarily nucleic acid content, sequences, and methylation patterns), the exosomes—which are extracellular vesicles that carry nucleic acids and are extruded when cells die—are adding to the information. Also, there are a lot of outcomes studies in Europe (the EPMA) and some in the US. It’ll be a few years before they read out their survival curves, but [these are] very, very interesting [developments].

The diagnostic industry is getting healthier, because of the increasing use of companion diagnostics (essential to determine the safe and effective use of a therapy) and complementary diagnostics (informs on factors affecting the benefit/risk of using a therapy but does not restrict its use). The Oncology Center of Excellence (OCE) of the FDA has helped with that with their many guidance documents; machine learning, AI, and digital path which allows for a better reading and what the pathologist sees with his eyes; and the immune microenvironment. Human oncology will continue, and it is a pleasant surprise for all of us how good these therapies are.

Alberto: So how do you expect some of these advances, and maybe other expected advances, to make their mark in 2024 and beyond?

Gary: I think, of course, you can’t prove anything very easily without randomized clinical trials. The level of the analyte you’re measuring has a certain near drop point that predicts clinical outcome. These studies are continuing, and you’ll get more and more evidence that when you reach a certain level of analyte, you can assume that you’ll get clinical utility or clinical outcome improvement. As our friend Bob Temple at FDA said, drugs are not approved unless they improve how a patient feels, functions, and survives. While the survival curves are the only thing that’s highly objective—and you need to show that in order to get a drug approved—as Lanny said, there are other target organs for which better surrogates exist.

Alberto: Both of you have mentioned machine learning, obviously a huge topic. But are there any specific areas where you think machine learning/AI could have an impact in the coming, say, 12 months? Looking beyond that is probably unrealistic. But anything that could happen in the next 12 months, that leverages machine learning in this space?

Lanny: I think we will see it infiltrate into almost every area over the next 12 months. It will influence everything from how medical students learn and are trained, to how pathologists sign out histopathology slides to make diagnoses [the way it’s been practiced] since the 1870s. So, I really think that this represents a fundamental sea change in how medicine is practiced. I’m obviously not alone, because you can’t open any journal or go to any session at a major conference these days without seeing at least one article or hearing at least one talk on the impact of machine learning on whatever the topic is that’s being discussed.

Alberto: Big picture: Would you expect, Lanny, that the processing, for example, of images and pathology samples, etc., will be vastly accelerated, maybe even made more accurate, if I may even go that far?

Lanny: I think there’s already evidence that it’s happening. Gary can talk at some length about, and he already mentioned, digital pathology and the kinds of things that are taking place, some of which are being evaluated in formal projects. And other things are being done as well as one-offs in certain centers that have access to these kinds of new approaches.

Gary: Yes, I think pattern recognition, i.e., the pattern the pathologist identifies as cancer [will be important]. Carcinomas are by far the majority of cancers, and when [the pathologist] sees the cells starting to invade the basal membrane of these epithelial sheets, that’s cancer. It’s not easy, but reasonably certain, that machine learning will allow recognition of this invasive process, which is the definition of cancer in pathology.

Lanny: In terms of complex diseases, and being able to assist in diagnoses, that’s another area [where machine learning and AI will be making its mark], because remember, this is the sort of thing that the machine learning program educates itself with for every case it sees. And if the patient metadata are made available, the number of cases that it can see and accumulate is enormous. And so, it is bound to not necessarily replace, but certainly frame a condition in a way that then allows people to start from that point and bring that into line with all of the other signs, symptoms, and other things that are going on with any given patient.

Alberto: Let me ask you about a topic that neither of you have mentioned, and that is CRISPR. The US FDA has just approved, as you know, a CRISPR-based gene therapy treatment for sickle cell disease. One could argue that this is an example of translational science at its best, where bedside success is efficiently turned into a product that treats patients with a serious condition for which there has been no cure until now. So, the question is, what if anything can we take from this example, for the purpose of finding cures for other debilitating diseases and unmet medical needs? Where do you see the true sustainable success story here? Could it be applied to other diseases?

Lanny: That technology absolutely can. Again, there are certain requirements. In other words, it has to be something that can be corrected at the genomic level, and whose correction achieves a certain threshold where it mitigates the debilitating effects of the disease. Sickle cell is a perfect example, thalassemia’s another one—those sorts of things. That’s correction at the genome level.

Gary has already mentioned other areas where I think there will be major therapeutic advances where we are already seeing that as well: in immune profiling and harnessing, amplifying, or blocking the immune response—amplifying it in situations where we see antipathogen- or anticancer-specific antigens, and blocking it in contexts like autoimmunity, multiple sclerosis, etc.

I’m not saying that we will achieve them necessarily in 2024, but I think progress will be made. I hope that we’ll even begin to see [these advances] in things like Type 1 diabetes: The antigen is probably identifiable; then you can either block the T cells that are causing the pancreatic destruction or mask the antigens that are the targets of those T cells. That would be a very significant step in affecting millions of people with a very novel approach.

And then finally, (Gary mentioned this), measurable residual disease. It’s not a surrogate, it is a direct measure of the tumor itself in oncology. So, as more and more sensitive techniques that are accurate and standardized emerge, they really do provide an early measure that can allow one to either intensify or de-intensify a particular therapy, and therefore either intervene at an earlier time point, before excessive morbidity is present, or avoid morbidity by discontinuing therapy or by reducing therapy based on very sensitive assessments. We’re already seeing readouts of examples where that’s occurring, and it is prolonging life and improving quality of life of patients with many different diseases.

Alberto: Gary, to conclude, let me ask you a final question: What translational science and medicine articles can we look forward to reading in Global Forum in 2024?

Gary: We’ve mentioned the areas of interest. We’ll be inviting senior authors with key opinions on the sickle cell problem. I think the reason that one of the first targets is hemoglobin is that one of the 146 amino acids in the beta chain isn’t the right one, and thanks to Charpentier and her colleague and co-Nobel laureate Doudna, you can read down to the wrong sequence, the wrong codon, and then insert the right codon with the CRISPR-Cas9 excision tool. And so, for any genetic disease where the sequence is known and [the error] is hopefully monogenic, it can be corrected. Sickle cell is one of the obvious ones, because it’s been understood for a long time. And unfortunately, or fortunately, we’re only seeing in vitro excision. The challenge is going to be to be able to give the excision in an in vivo situation, which has not yet happened. I doubt it will happen in the next year, but it will in the ensuing years.

Alberto: That brings us to the close of our conversation. Lanny, Gary, really want to thank you again for our annual conversation on trends, what has happened in the current year and what will happen in the coming year. Fascinating progress in science as we’ve seen and will see much more of, published in our Global Forum. I’d like to thank you both and wish you and our readers a happy holiday season.