Thread by @s_r_constantin

My current “impractical” project is building a comprehensive list of interventions that cause >20% complete tumor regression in metastatic solid tumors. Basically, the hardest of hard modes for cancer.

I don’t think I’m quite done (will put it up in a post when I’m satisfied), but here’s what I’m seeing so far.

1. There are some things I don’t expect to generalize much. Stereotactic radiation for lung metastases & intratumoral injections of Nasty Stuff for skin metastases — great where available but you can’t always access the tumors.

2. Metastatic melanoma is immunogenic, so it actually gets infiltrated by lymphocytes. If you expand those, and maybe transfer them with some specific target, you can strengthen that response and kill the tumors. But won’t work for many other cancers.

That’s really all we got for human studies; TILs, physical localization, and a couple cases where targeted therapies work on narrowly targeted populations. (Exemestane on hormone-positive breast cancer, erlotinib on EGFR-positive lung cancer, etc).

Animal studies are more exciting. (And, of course, less likely to translate.)

One cool thing is that there are a few trials of experimental treatments on companion animals. Dogs and cats.

So these are spontaneous tumors, more like someone actually getting cancer than injecting a healthy young mouse with a tumor cell line.

So one thing that sometimes works in dogs with metastatic cancers is transfecting them with the IL-12 gene, which is an anti-tumor cytokine. Gene therapy!

Another dog therapy (melanoma-specific) that replicated a couple times is a DNA vaccine that just straight-up kills the melanocytes. No melanocytes? no melanoma!

Can you really do this? Don’t you need your melanocytes? Well, I think these are local injections so you don’t lose *all* of them, just near the skin metastases?

People with vitiligo don’t have symptoms except the patches of pale skin, it seems, so I guess it’s fine.

Another dog therapy that was used on melanoma but doesn’t seem specific to it, was an AAV gene delivery of CD40L to the metastases. This activates dendritic cells and other anti-tumor immune responses.

Might be hard to get into disseminated tumors, though.

Moving into mice, we get less realistic tumor models, but way more studies, including quite a few with 100% complete tumor regressions in metastatic tumors.

A lot of these are gene or cell therapies expressing strongly tumoricidal stuff in the TNF family:

mesenchymal stem cells expressing TRAIL, AAV expressing TNFSF14, etc

Also intriguing: CAR-T working on metastatic pancreatic cancer! siRNA knockdowns of Met! Using Listeria as a gene delivery mechanism?!

IL-12, like IL-2, will kill you if you inject it in the bloodstream, but it’s really great if you can keep it in the tumor. frontiersin.org/articles/10.33…

One nice thing about gene therapy is that it is naturally local. You can either just inject it into the tumor, or you can inject it systemically but target the delivery mechanism to the type of tissue you want.

Interesting tradeoffs there. If you physically put it in the tumor you’re limited to reachable tumors; if you try to target it, your targeting mechanism might be noisy.

Fortunately, sometimes doing immunotherapy to the main tumor will simultaneously regress distant hard-to-reach metastases.

ziopharm.com/controlled-il-… These guys are developing an IL-12 gene therapy targeted to the tumor.

These guys have a targeted AAV that delivers IL-2 and TNF-alpha to cells with surface markers indicative of rapid cell division. tiltbio.com

I don’t know if this is irrational, but I feel good about TNF-alpha because it’s so basic. It directly causes an innate-immune inflammatory response & phagocytosis.

I don’t know enough immunology to have confidence in adaptive immune mechanisms. But I *know* TNF kills shit.

If you’re trying to make predictions about how experimental therapies will turn out, you need either super-specific domain expertise (not me!), overwhelming experimental evidence (by which point your take is stone cold) or SIMPLE mechanisms. I like simple.

Basically, the “safe”, reliable way to build a cancer therapy is to have an “if-then” agent. “If cancer, kill cell.”

The killing part is easy. Lots of very reliable cell-killing toxins, apoptosis genes, etc. We’ve known how to kill a cell since the 19th century.

The “if” part is the hard part. How do you select for all the cancer without getting too much of the non-cancer? Esp. given that cancer is diverse and evolves?

Most effective cancer therapies actually dodge the question.

Chemo and radiation kill everything they touch, but kill cancer a bit faster than they kill the rest of you, taking advantage of cancer’s rapid cell division and genetic instability.

CAR-T therapies for leukemia and lymphoma mostly target their T-cells to kill the *entire cell category* affected by cancer. All the B cells, for instance. No B cells? no B-cell lymphoma!

Lots of monoclonal antibodies are the same way. Rituximab is an anti-CD20 chimeric antibody. Kills all your B cells.

If you can just *point* and say “here’s the tumor, kill THIS”, that’s great. Surgery, targeted radiation, stereotactic radiation, etc. Trouble is, it tends to miss microscopic or otherwise undetected cancer cells.

Immune therapies often hope to dodge the question another way. “Your body already knows how to attack cancer; let’s take away a barrier and let it do its thing more!”

Hence, CTLA4 and PD1 antibodies; tumor-infiltrating lymphocytes; etc.

“Strengthen anti-tumor response” can be very effective, but seems very hard to make a priori predictions about. You don’t know it’s gonna work till it does. (At least I don’t.) Jim Allison’s earned his Nobel, but I can’t guess who the next Allison’s going to be.

What seems intriguing to me is to find a “universal indicator of cancer” and a way to attach a “detector” to a local “kill-switch” (like expressing a cell death signal in that specific cell.)

Here’s a striking claim: all cancer cells are negatively charged, and no normal cells are. link.springer.com/article/10.100…

There’s a reason for this: the Warburg Effect. Cancer cells do a lot of glycolysis.

I’ve played around a bit with tumor metabolomics data, and principal component between “tumor” and “normal” cells, across tumor types, is all glycolysis-related metabolites. It’s hard to miss.

Glycolysis makes lactic acid. Dissolve lactic acid (like any acid) in water and it releases positively charged hydrogen ions, leaving negatively charged lactate.

ncbi.nlm.nih.gov/pmc/articles/P…

You can detect negatively charged cells with magnetic nanoparticles. Then the nanoparticles can be detected via fluorescence microscopy.

As their pictures show, positively-charged nanoparticles, but not negatively-charged ones, bind to HeLa cells, a well-known cancer cell line.

In fact, positive nanoparticles bind to all 22 cancer cell lines tested and none of the 4 non-cancer cell cultures.

In a mixture of breast-cancer cells and non-cancer blood cells, the nanoparticles were 90% accurate at picking out the cancer cells.

You can literally just draw the nanoparticle-coated cancer cells to the side of the test tube with a magnet. This extracts a mixture that’s 80% cancer cells, along with some granulocytes from blood that seem to interact w/ the cancer cells.

They determine the negative charge is coming from lactate because it correlates directly with lactate levels & inversely with medium glucose concentrations. Glycolysis inhibiting drugs also reduce the negative charge.

jnanobiotechnology.biomedcentral.com/articles/10.11… Here they take blood from a mouse with sarcoma, incubate it with positive nanoparticles, and observe a bunch of nanoparticle-labeled cells, vs. none with healthy mouse blood.

“But wait!” you say. “Aren’t all cell membranes negatively charged?”

Phospholipids are negatively charged, yes, but they immediately attract positive ions in aqueous solution.

If you put a microelectrode in solution with cells, they’ll be attracted to a positive charge, I suppose because it drives off the positive ions & attracts the phospholipids? But this might not happen with positive nanoparticles?

m.scirp.org/papers/97907 Another paper noting that cell surface charge is more negative in cancer cells and cells undergoing mitosis.

europepmc.org/article/pmc/pm… This experiment shows that breast cancer cells have stronger negative surface charge than normal cells at physiological pH ranges. Both are negatively charged, but the cancer cells are more so.

In fact, the more metastasis-prone a cell line is, the more negative its zeta potential (a measure of negative surface charge). auajournals.org/article/S0022-…

If this is for real, there’s a simple model for a cancer drug: positively charged nanoparticles bound to something cytotoxic or a gene vector bearing a suicide gene. “If I’m very electronegative, kill me.”

*transfect, stupid autocorrect