🫀 Why Heart Cancer Barely Exists, Plus AI That Sees Tumors From The Future

Quick housekeeping before the science: we missed the last two weeks because one of us was enjoying a holiday, and then came back sick, and then Finland declared Helatorstai (Ascension Day) a national reason to do absolutely nothing, so we took it. Apologies for the radio silence. To make up for it, this week's lineup is enjoyably unhinged: heart-as-cancer-bouncer, AI that sees pancreatic tumors years before your radiologist can, the microplastic situation in your brain (worse than you think), HIV's most dramatic pharmacology comeback, and a Utah startup whose press release game is way ahead of its peer review game.

RESEARCH

Name someone you know with heart cancer. Couldn't, right? Primary cardiac cancers show up at autopsy at roughly 0.001% to 0.03%, which is medical-speak for "essentially never." Doctors have shrugged at this for a century. Theories floated around (cardiomyocytes don't divide much, the tissue is too dense, the resident macrophages are too mean), but none of them really nailed it.

Giulio Ciucci and Serena Zacchigna's team at ICGEB Trieste just nailed it in the most satisfying way possible. Their April 23 Science paper shows the heart doesn't passively resist cancer. It actively beats tumor cells into chromatin submission. Every contraction physically squeezes any cancer cell sitting in the tissue, and that mechanical force gets transmitted into the nucleus via a nuclear-envelope protein called Nesprin-2, which cranks up H3K9 methylation, compacts the DNA, and shuts down proliferation. The cell cycle just stalls. (A companion Perspective by Paltzer and Martin walks through why this changes how we think about cardiac-tumor biology.)

The proof is wild. They built a "two-heart mouse" with a donor heart grafted in, but not pumping. Same animal, same blood, same tumor cells. The parked heart got overrun by cancer in two weeks. The still-beating native heart shrugged it off. Knock out Nesprin-2, and even a normally beating heart loses its superpower.

Here's the fun part: Zacchigna's lab is already prototyping wearable devices that compress superficial tumors at heartbeat-like rhythms, starting with melanoma. "A kind of massage" of the tumor, she calls it. The same mechanism that prevents your heart from regenerating after a heart attack is what keeps cancer from moving in. Blessing, curse, and now possibly a therapy, all encoded in one nuclear-envelope protein you'd never heard of.

NEWS & RESEARCH

Pancreatic cancer is, statistically, the worst diagnosis you can get. The 5-year survival rate has been stuck at exactly 13% for three years running, and it's projected to become the second leading cause of cancer death by 2030. The reason is brutally simple. By the time anyone notices anything is wrong, it's basically too late.

Mayo Clinic's REDMOD model (Radiomics-based Early Detection Model) just rearranged the math. Published April 22 in Gut, the AI scans routine abdominal CTs that radiologists already read as "totally normal" and flags pancreatic cancer up to three years before clinical diagnosis. AUC of 0.82, 73% sensitivity at a median 16-month lead time versus 38.9% for the human radiologists looking at the exact same scans.

The model extracts hundreds of radiomic features (basically pixel-level texture math) that pick up tissue remodeling and fibrosis happening years before a visible mass appears. Senior author Ajit Goenka was emphatic that this isn't a "radiologists are bad" story. As he put it, pancreatic cancer at its earliest stages is "too subtle to be detected by the human eye alone". The signal exists. It just sits below the noise floor of human vision.

Validation ran across roughly 2,000 multi-institutional CTs. At lead times over 24 months, REDMOD hit 68% sensitivity versus radiologists' 23%, basically 3x. A prospective trial called AI-PACED is now deploying it for elevated-risk patients (new-onset diabetes after 50 is the canonical red flag). To be fair, the 6-to-1 control ratio is way richer than real-world population screening, so false-positive volume will spike in deployment. But this is the rare AI paper that, if specificity holds in the field, actually shifts a survival curve.

RESEARCH

You've heard "a spoonful of plastic in your brain." It's not quite accurate, but the truth is still gross. A new study from Beijing Tiantan Hospital, published April 20 in Nature Health, found microplastics and nanoplastics in 100% of healthy brain samples and 99.4% of brains with tumors. They sampled 156 diseased tissue specimens from 113 living patients during neurosurgery, plus 35 healthy brain samples from post-mortem donors.

The numbers, in case you were eating: median 50.3 micrograms of plastic per gram of healthy brain tissue, and 129 micrograms per gram in tumor-affected tissue. Nanoplastics outnumbered microplastics, with PET (water bottles), polyethylene (plastic bags), nylon, and PVC making up most of the polymer mix. Your microglia (the brain's resident janitors) are now gobbling up nanoplastic particles they cannot digest. They're hoarders with no recycling option.

This is essentially the living-patient sequel to the famous 2025 Campen study that found brain microplastic concentrations 7 to 30 times higher than in the liver or the kidney. The Chen team improved on Campen's methodology with rigorous controls for operating-room contamination, which had been the main critique of the original. They literally ran a Py-GC/MS audit on every syringe and glove in the OR before counting their results.

To be fair, this is correlation, not causation. The higher peritumoral concentrations probably reflect a compromised blood-brain barrier in cancer (the "blood-tumor barrier"), not plastics actually causing tumors. But the bigger picture, that nanoplastics cross into the most lipid-rich organ in your body and apparently don't leave, is harder to wave away. The real question isn't "should I be worried?" It's "where does any of this stuff go?"

NEWS

On April 21, the FDA approved Idvynso, Merck's once-daily two-drug HIV pill combining doravirine and a microscopic dose of islatravir. Surface-level, it's just another switch regimen, non-inferior to Biktarvy at 48 weeks. The actual story is the dramatic resurrection of islatravir from clinical death.

Islatravir was supposed to be the future. As the first-in-class nucleoside reverse transcriptase translocation inhibitor, it had an absurdly long intracellular half-life and was being developed for monthly or even yearly dosing. Then in December 2021, Merck disclosed that participants in higher-dose trials showed unexplained drops in lymphocyte and CD4 counts, and the FDA promptly put clinical holds on basically every islatravir study. The PrEP dream died. The long-acting injectable dream died.

What survived is Idvynso at 0.25 mg of islatravir, a fraction of the originally planned dose. At that level, lymphocyte and CD4 changes are indistinguishable from approved ART. In the head-to-head phase 3 trial, 92% of patients who switched from Biktarvy to Idvynso maintained viral suppression at 48 weeks, statistically indistinguishable from the 94% who stayed on Biktarvy.

Why does this matter beyond "another pill"? Most modern HIV regimens lean on integrase inhibitors (Biktarvy, Triumeq), which have known weight-gain and intolerance issues. Idvynso is the first INSTI-free, tenofovir-free, two-drug single-tablet regimen approved for switch therapy. It's only for virologically suppressed adults with no resistance history, so this is a switch drug, not a starter. Merck hasn't disclosed pricing, which means it's probably as eye-watering as everything else in this category. A clinical resurrection by aggressive de-escalation.

NEWS

Picture this: a Salt Lake City company holds a press conference, claims it grew functional human sperm in a dish from testicular biopsy cells, then used those sperm to fertilize human eggs and form embryos. No peer-reviewed paper. No preprint. No technical data released. Just press releases and microscope photos of what they say are sperm tails. Welcome to Paterna Biosciences.

The team has a legitimate pedigree (founders include University of Utah reproductive urologist Alexander Pastuszak and HHMI investigator Brad Cairns), and the technical approach (in vitro spermatogenesis from spermatogonial stem cells) is biologically more tractable than full IVG from skin cells. They've raised about $11.4 million from IVIRMA, Mayo Clinic, and SpringTide Ventures. This isn't garage biotech.

But the rollout is a near-exact copy of Kallistem in 2015, a French startup that announced "complete human spermatogenesis in vitro" via press conference, promised 2017 clinical trials, never published, and faded. The strongest outside endorsement Paterna has gotten ("This is huge," from Baylor's Larry Lipshultz) is from Pastuszak's former training mentor, a conflict no coverage seems to mention. Meanwhile, Katsuhiko Hayashi, the literal pioneer of mammalian IVG, has estimated that viable lab-grown human sperm is about seven years away. Paterna says they'll start pregnancy trials next year.

"We fertilized eggs and formed embryos" sounds impressive until you remember ICSI with immature round spermatids has done that since the 1990s, with abnormal outcomes. Without verified meiosis, methylation data, or blastocyst progression rates, this is marketing dressed in a lab coat. We'll be watching the bioRxiv feed.

This week was peak biotech absurdity: a mechanical organ that physically deletes cancer's instruction manual, an AI that sees a tumor's shadow years before it casts one, plastic confetti permanently squatting in your brain, a Merck molecule that survived clinical death and came back at one-tenth strength, and a Utah biotech running the Kallistem speedrun. The common thread is "your body is doing things you didn't know about, and we keep finding more."

If a friend would enjoy reading about heart-as-cancer-bouncer or the inevitable plastic-in-your-microglia situation, send this their way. And we'd genuinely love to hear from you: which of these stories made you most "wait, what"? Reply and tell us, or just send us your favorite obscure organ.

Keep questioning everything (and your microglia),

Prateek & Jere

P.S. We're accepting submissions for the next "organ you've never heard of getting cancer" mystery box. Lungs and breasts need not apply.