You’ve heard of CAR-T already. We take T-cells out of a patient, stick synthetic receptors into them, grow them up, and graft the cells back into that patient. This technology has been quite powerful, and to understand what that actually means, let’s take a look at Multiple Myeloma (MM), one of the first diseases CAR-T has been approved for. Then, we’ll get to why the ex vivo vs. in vivo CAR-T discussion matters.

Multiple Myeloma is a liquid tumor

MM is a cancer of the plasma cells, which are the producers of antibodies. Generally when you think of cancer, you may think of a solid chunk of mass growing when it’s not supposed to - these are solid tumors. Plasma cells reside in the circulatory system - when they proliferate they don’t form solid, identifiable masses. Instead, they are still just proliferating throughout your blood and these cancers are therefore termed liquid tumors. Crucially, this means that unlike with solid tumors, you can’t surgically remove the tumor.

The main treatment historically has been using are proteasome inhibitors such as bortezomib. Since plasma cells are professional antibody factories, they can generate a lot of misfolded proteins by accident. This waste needs to be cleaned up by the proteasome, or else it can lead to apoptosis. Bortezomib leans into this mechanism - it prevents proteasome function, leading to these cells going all the way to committing apoptosis. Bortezomib is injected systemically and has no cell-type specificity, so it can cause toxicity in cells outside of the plasma cells you want to treat for MM.

Before I get too far, welcome to my Substack! This is my first solo post here - I also write for Decoding Bio which you should read. I’m going to use this space to write about interesting technology + problems + businesses that I learn about, generally within AI and Bio (FM, agents, synbio, protein design, cell/gene therapy). Please subscribe, and also if literally any of this is interesting to you, please reach out - I’d love to meet!

Subscribe now

Antibodies have improved Multiple Myeloma treatment

A wave of immunotherapies (antibody, T-cell engager, CAR-T) has pushed care past the efficacy of bortezomib alone. These all target either CD38 or the B-cell maturation antigen (BCMA), a marker for plasma cells. The current Standard Of Care (SOC) is a combination termed D-VRd of Darzalex (CD38 antibody) + Velcade (bortezomib) + Lenalidomide (immunomodulatory drug) + dexamethasone (steroid). VRd was the previous SOC, and you can see below the significant difference that adding the CD38 antibody made to this treatment.

Wait: What do these terms mean?

Learning about MM was my first intro to most of these terms used to describe cancer treatment outcomes. Just in case you’re a researcher who also hasn’t seen these, let’s review them.

Progression-free survival (PFS): how long a patient lives with cancer without their disease getting worse

Complete Response (CR): no detectable level of cancer (generally at a certain timepoint

MRDneg (10⁻⁵ or 10⁻⁶): MRD means at least 1 myeloma cell detected among 10⁵ or 10⁶ cells. MRD-neg refers to the proportion of patients for which no myeloma cells were detected at each level of detection, which is a good thing.

Adding the CD38 antibody improved the long-term outcomes of the treatment on all 4 of these metrics!

CAR-T seems to be even better

Two CAR-T products have been approved for MM therapy - Abecma and Carvykti. However, being approved is not the same as becoming the standard of care. Take Carvykti: which was initially approved for patients who have had 4 (!? ☹️) or more prior lines of therapy. As more longitudinal data from the Carvykti continue to come out, it has shown to have even stronger efficacy than the SOC. While MRDneg (10⁻⁵) for D-VRd is 75%, data from their phase III trial show the MRDneg (10⁻⁵) for Carvykti is 92%. Based on these updated clinical trial data, the FDA approved Carvykti to be used after 1 line of treatment. J&J is now testing Carvykti as a first-in-line treatment, which would determine if it could become the SOC itself.

That’s awesome! What’s the catch?

First, CAR-T causes some harsh side effects. Cytokine Release Syndrome (CRS) and Immune Effector Cell-Associated Neurotoxicity Syndrome (ICANS) are hyperactivated immune responses that are caused by immunotherapies like CAR-T and can be fatal. There’s a ton of technology development happening to make the actual CAR receptors safer, more efficacious, inducible, etc. so as to mitigate these side effects (the current FDA-approved CAR-T products are all “2nd generation” CAR-T designs, but we’re already on the 5th generation in vitro).

The bigger challenge is that CAR-Ts are currently really expensive and time-consuming to make. Let’s briefly run through the process, and why it is the way it is.

1. Take blood cells out of a patient’s body

2. Isolate T-cells

3. Genetically insert the CAR construct (usually with lentivirus)

4. Expand and activate the new CAR-T cells (you need tens to hundreds of millions of cells per patient). At this step, also do QC on the newly generated CAR-T cells. Did they get the CAR construct? Are they happy and not exhausted? Did you accidentally cause any critical genomic issues during installation of the CAR into the genome?

5. Lymphodepletion: destroy the patient’s immune system (wait, what?). This is done to make sure the grafted CAR-T cells will be able to proliferate well, and essentially don’t have to compete with existing T-cells for space.

6. Graft the CAR-T cells back into the patient.

Okay, so this sucks for a couple of main reasons.

First, you have to do this for each patient. That’s because currently we don’t yet have good technology to produce T-cells “off-the-shelf” that can work for every patient, because your body might recognize it as foreign and reject the graft. This is time expensive (4-5 weeks): it takes real people to extract those cells, transduce them, and grow them. You can’t benefit from economies of scale as you can with most medicines. The current FDA approved CAR-T therapies will run you for about half a million dollars each. The drug developers are working on ways to make this process faster and require fewer cells.

Second, you have to destroy that patient’s immune system in order to let the grafted CAR-T cells grow. You can appreciate that by weakening the immune system, you open that patient up to infection.

Third, cells aren’t very happy ex vivo. (The reason that we’re able to do biomedical research on cells in vitro is because they’re all actually cancer cells and are much more resilient than normal cells). This one feels much more manageable than the others.

What if we just make the CAR-T cells inside of a patient?

This here is the holy grail. Instead of the expensive process of extracting cells out and turning into CAR-T cells, we can use technologies from gene therapy to directly deliver the CAR genes to T cells in our body. If done right, in vivo CAR-T therapies can be mass manufactured (therefore cheaper!), quicker for the patient (don’t have to wait 5 weeks for ex vivo production), and easier on the clinical system (don’t have to extract + develop CAR-T on a patient-by-patient basis). Gene therapy technologies have become better over the past decade, and industry has gotten better at manufacturing them at scale.

To execute this, you have two choices to make. The following are those choices and the current main tools in the toolbox companies are using.

Delivery: How do I get the CAR construct to the T cells?

1. Viral delivery. Utilizing targeted lentiviruses to deliver specifically to T-cells. (in theory AAV may be used here, but I haven’t seen anyone trying commerically). Companies like Umoja, Kelonia, Vyriad, Interius are using this.

2. Virus-like particle (VLP). These are similar to viral capsids, and don’t include any viral replication machinery. Companies like Ensoma, nChromaBio (formerly Nvelop), and Azalea are going this route (spinouts from labs of Hans-Peter Kiem, David Liu, Jennifer Doudna, respectively).

3. Lipid nanoparticle (LNP). These were used to deliver mRNA for the covid vaccines. Non-immunogenic and therefore re-dosable. Seems to be a little harder to target specific cell types of interest as compared to the viral methods (by default, lots of accumulation in the liver). Companies like Stylus and Capstan are using these.

CAR construct expression: what form do I want the CAR construct to be in? Do I want it to replicate?

1. mRNA (transient) - deliver strands of mRNA encoding the CAR construct. As the cell divides, this mRNA does not replicate, so eventually this will get diluted out. This makes it a little less problematic if you deliver the CAR construct to the wrong cell. Capstan and Cartesian are using this. Choosing a delivery vehicle that allows you to re-dose is important here.

2. genomic (permanent) - directly integrate the CAR construct into the genome, so the entire lineage of this cell will express the CAR construct. A single dose may afford longer-term protection. All of the lentiviral-based delivery vehicles use this. Targeted genomic insertions seem to also improve consistency and efficacy of the CAR-T cells - Stylus is using their Large Serine Recombinase technology to do this.

In vivo delivery has different challenges than ex vivo manufacturing. Gene delivery is still relatively hard, and precise gene delivery is even harder. Even with the best delivery vehicles in the world, you will get more precision in targeting CD8+ T-cells if you extract them and isolate them ex vivo than you will by developing a CD8+ targeting virus. With ex vivo manufacturing of CAR-T cells, you can also QC the product to make sure there are no genomic issues and track the quality of the CAR-T cells before grafting it back into the patient, which you can’t do in vivo.

The technology is still relatively early, so the verdict is still out on whether or not in vivo CAR-T will achieve the same safety/efficacy as ex vivo. Most of these are either preclinical or starting Phase I. If it pans out, there may be room for both - in vivo CAR-T for indications with larger patient populations and where you worry less about off-target delivery, and ex vivo CAR-T for patients whom you need precise on-target delivery.

I’m incredibly excited for this technology to be validated - and as CAR constructs themselves improve, they are being expanded for use in fibrosis, autoimmune disease, and even aging.