Sources and notes 1. The study. Aleksandar Denic, Andrew Rule and colleagues at Mayo Clinic, with Cleveland Clinic, "evaluated 1,638 living kidney donors." Filters (glomeruli) per kidney were estimated from CT cortical volume multiplied by glomerular density on a biopsy. JASN, 2017. Open access. doi.org/10.1681/ASN.20160201… pmc.ncbi.nlm.nih.gov/article… 2. The numbers. "Donors 18 to 29 years old had a mean 990,661 nonsclerotic glomeruli" per kidney, which "progressively decreased to 520,410 nonsclerotic glomeruli per kidney... in donors 70 to 75 years old. Between the youngest and oldest age groups, the number of nonsclerotic glomeruli decreased by 48%, whereas cortical volume decreased by only 16%." Nonsclerotic means not scarred, so still working. The paper puts the loss at 7.3% per decade. 3. Function. GFR fell along with filter number, but these donors were healthy enough to donate. The authors: "Loss of nephrons with aging is grossly underappreciated by both the degree of glomerulosclerosis on biopsy and the cortical volume decline on imaging." 4. Limits. Different people at different ages, not the same people followed. Donors are healthier than average. 5. Slowing decline. DAPA-CKD enrolled 4,304 people with CKD, with or without type 2 diabetes. The HR was 0.61 (95% CI 0.51 to 0.72) for a composite of sustained 50% or greater eGFR decline, end-stage kidney disease, or death from renal or cardiovascular causes (Heerspink et al., NEJM, 2020). doi.org/10.1056/NEJMoa202481… EMPA-KIDNEY: HR 0.72 (0.64 to 0.82) (NEJM, 2023). doi.org/10.1056/NEJMoa220423… FLOW enrolled 3,533 people with type 2 diabetes and CKD. The HR for major kidney disease events was 0.76 (0.66 to 0.88) (Perkovic et al., NEJM, 2024). doi.org/10.1056/NEJMoa240334… 6. Neither class has been tested for slowing normal age-related filter loss in healthy people. The 16% figure is cortical volume, the kidney's outer layer.

Sources and notes 1. The trial. Met-HeFT, part of DANHEART, run from Aarhus University Hospital in Denmark. n=940 with heart failure and reduced ejection fraction, average age 70, only about 10% with diabetes (the rest had raised blood sugar or obesity). Metformin vs placebo, mean 3.7 years. Presented at ESC Congress, August 30, 2026. escardio.org/news/press/pres… clinicaltrials.gov/study/NCT… 2. The result. Death, worsening heart failure, heart attack or stroke hit 25.1% on metformin and 23.4% on placebo (HR 1.10, 95% CI 0.84 to 1.42, p=0.48). No cut in new diabetes. No associated harm. 3. Why it matters for aging. The anti-aging case for metformin leans partly on heart protection seen in people with diabetes. This is one of the few randomized tests of hard outcomes in people mostly without diabetes. 4. The original signal. UKPDS 34 (Lancet 1998): overweight people with diabetes on metformin had 36% fewer deaths over a median 10.7 years than those on diet alone. pubmed.ncbi.nlm.nih.gov/9742… 5. TAME. The proposed metformin trial for aging wants more than 3,000 adults aged 65 to 79 without diabetes, and it's still raising money to launch. afar.org/tame-trial 6. Next. The larger randomized test of metformin's heart claim outside diabetes is the VA's IMPACT trial (n=7,410, prediabetes plus artery disease), due around 2029. clinicaltrials.gov/study/NCT… 7. Limits. One disease group (weak hearts), a modest sample, and a symptom-plus-events endpoint rather than a pure aging outcome.

Sources and notes 1. The paper. Ruijin Hospital, Shanghai Jiao Tong University (Qiongyi Hu, Chengde Yang) and Abogen (Bo Ying), Cell, September 18, 2026. ABO2203 is "a lipid nanoparticle-formulated messenger RNA (mRNA) encoding a CD19-targeting TCE" (T-cell engager). In "three patients with refractory secondary immune thrombocytopenia," it "achieved rapid and complete peripheral B cell depletion, with sustained depletion in bone marrow," with "durable platelet recovery, improved serology, and reduced disease activity through 6 months," and "only grade 1 and 2 adverse events without cytokine release syndrome." B cells that came back were mostly transitional and naive. doi.org/10.1016/j.cell.2026.… 2. Patients. Three women aged 35, 46 and 48, with lupus, Sjogren's syndrome and antiphospholipid syndrome. This detail is from VCBeat, not the abstract. vcbeathealth.com/article/629… 3. Disclosure. Abogen developed ABO2203 and co-authored the paper. The drug is given under the skin in Abogen's separate lymphoma study; the route in this study isn't in the abstract. 4. The CAR-T comparison. In Germany, 15 patients with lupus, myositis or systemic sclerosis got CD19 CAR-T, and immunosuppressive therapy "was completely stopped in all the patients" (Muller et al., NEJM, 2024). doi.org/10.1056/NEJMoa230891… The first lupus series was 5 patients (Mackensen et al., Nature Medicine, 2022). doi.org/10.1038/s41591-022-0… 5. Why it matters. The approved protein version of this idea, blinatumomab, runs as a continuous IV infusion for 28 days per cycle and carries a boxed warning for cytokine release syndrome (FDA label). accessdata.fda.gov/drugsatfd… 6. Limits. Three patients, open label, six months, and a different disease from the CAR-T series. 7. The industry bet. AbbVie agreed to buy Capstan Therapeutics for up to $2.1 billion (June 30, 2025). AstraZeneca bought EsoBiotec for up to $1 billion (March 17, 2025). Kite (Gilead) bought Interius for $350 million (August 21, 2025). All three are building in vivo CAR-T. prnewswire.com/news-releases… astrazeneca.com/media-centre… gilead.com/news/news-details… 8. The safety break. Novartis halted autoimmune trials of rapcabtagene autoleucel after three fatal cases of immune effector cell-associated hemophagocytic syndrome, and BMS paused zola-cel enrollment. This was reported by Fierce Biotech; I found no primary Novartis release. fiercebiotech.com/biotech/no… 9. Corrections in this version. No primary source confirms the patients' sex, so this version says "three patients." The side effects were grade 1 and 2, meaning mild to moderate. The route is under the skin, per Abogen's AACR 2026 release on the same drug.

Sources and notes: Zheludev et al., "Viroid-like colonists of human microbiomes," Cell, 30 October 2024, 187(23):6521-6536.e18. Andrew Fire's group at Stanford. 29,959 distinct obelisks identified from human metatranscriptome data. Present in roughly 50% of human oral metatranscriptomes examined (17 of 32) and roughly 7% of stool (29 of 440). They are circular RNA elements that replicate within gut bacteria, form their own phylogenetic group with no detectable sequence similarity to any known biological agent, and encode a novel protein superfamily the authors named Oblin. The authors report that they appear dispensable for host bacterial growth. No disease association has been established, and none is claimed here. doi.org/10.1016/j.cell.2024.…

Sources and notes: 1. What a negative result means. GRAIL's patient page says a "No Cancer Signal Detected" result "does not rule out cancer" and that the test "should be used in addition to routine cancer screening tests." galleri.com/patient/the-gall… In the trial, the test caught 37% of the cancers diagnosed within 12 months of the first blood draw, and 27% in each of the next two rounds (all cancers). For the 12 cancers that cause most cancer deaths, it caught 48% to 63%. About 99% of negative results were correct, because most people tested don't have cancer. The problem is the cancers it doesn't see. Test performance paper, Nature Medicine: doi.org/10.1038/s41591-026-0… 2. The trial. NHS-Galleri, New England Journal of Medicine, Sept 22, 2026. 142,250 people aged 50 to 77 in England were randomly split. Half had their blood tested at up to three yearly visits. The other half had their blood stored and not tested. GRAIL funded the trial; Queen Mary University of London analyzed the data independently. doi.org/10.1056/NEJMoa250572… clinicaltrials.gov/study/NCT… 3. What a positive result means. 1,801 people had a positive result, and 937 of them (52%) were diagnosed with cancer (58%, 50% and 46% in rounds one, two and three). A positive result led to an NHS urgent cancer referral, with tests chosen by the organ the result pointed to. The test's top guess at the organ was right for 87% of the cancers found. Of the first-round positives where no cancer was found at the time, 18% (54 of 303) were diagnosed with cancer in a later round. 4. Age. NHS-Galleri enrolled people aged 50 to 77. PATHFINDER 2, the large US study, enrolled people 50 and over. GRAIL recommends the test for "adults with an elevated risk for cancer, such as those age 50 or older" and advises against it for people 21 or younger. No trial has tested it in people in their 20s, 30s or 40s. Within the trial, the share of positives that turned out to be cancer was similar from the 50s to the 70s. clinicaltrials.gov/study/NCT… 5. Where it helps. 72.8% of the cancers the test found were not breast, cervical or bowel cancer, the cancers England screens everyone for (58.7% with lung excluded too). Compared with the untested group, the tested group had: - more cancers found at stage I or II (647 vs 559) - 21% fewer diagnosed as an emergency (225 vs 286) - fewer stage IV cancers (341 vs 394). This is a rate ratio of 0.86 (95% CI 0.74 to 1.00), a secondary result the trial's plan didn't allow it to test formally. 6. The main goal. Fewer stage III or IV cancers among 12 cancer types that cause more than 60% of cancer deaths. The designers projected about 20% fewer. The result was 706 people in the tested group vs 688 in the untested group (rate ratio 1.03, 95% CI 0.92 to 1.14, P=0.63). Stage IV fell, but stage III rose (360 vs 287). 7. Deaths. The trial wasn't designed to measure deaths. Cancer deaths will be analyzed 3 and 6 years after the final trial visit. The closest precedent is the UKCTOCS ovarian screening trial of 202,562 women: it cut stage IV ovarian cancer by 24.5% and didn't reduce ovarian cancer deaths after 16 years. doi.org/10.1016/S0140-6736(2… 8. FDA and price. The FDA's Molecular and Clinical Genetics Panel reviews Galleri on Sept 23, 2026, and GRAIL is seeking approval for adults 50 and over. The panel advises; the FDA decides. The US list price is $949. fda.gov/advisory-committees/… galleri.com/cost

Correction on the head-to-head: the 37 vs 20 at hour eight was for danavorexton, given IV at its top 112 mg dose in sleep-deprived healthy men. Oveporexton, the approved pill, has not been run against modafinil that way. My source’s reply has it right; the post does not.

Sources and notes: Ikeda et al., "Immune evasion through mitochondrial transfer in the tumour microenvironment," Nature, 22 January 2025, from Togashi's group at Chiba Cancer Center. Open access. Tumour-infiltrating lymphocytes were found carrying mitochondrial DNA mutations identical to those in the patient's cancer cells, which is how the transfer was detected in the first place. The transferred mitochondria resist the recipient cell's mitophagy, partly because inhibitory molecules including USP30 travel with them, and then undergo homoplasmic replacement, meaning they displace the T cell's own mitochondrial population completely rather than mixing with it. The recipient T cells become senescent and functionally impaired. In melanoma and non-small-cell lung cancer, tumour mtDNA mutation was associated with worse response to checkpoint inhibitors. That association is correlative. Human specimens plus mouse and in vitro work. doi.org/10.1038/s41586-024-0…

Sources and notes: Wang et al., "Transplantation of chemically induced pluripotent stem-cell-derived islets under abdominal anterior rectus sheath in a type 1 diabetes patient," Cell, 25 September 2024. Trial ChiCTR2300072200, Deng Hongkui's group. One patient. Her own cells were reprogrammed to pluripotency using small molecules only, with no Yamanaka transcription factors and no viral vector, then differentiated into islets and transplanted under the abdominal anterior rectus sheath rather than into the hepatic portal vein. The site choice matters: it is imageable and the graft is retrievable. Time in target glycaemic range rose from 43.18% at baseline to 96.21% by month 4, then stayed above 98%. Insulin independence from day 75, HbA1c around 5% at one year, no transplant-related abnormalities reported. This is a first-in-human phase 1 in a single person. It doesn't establish safety or efficacy in anyone else. doi.org/10.1016/j.cell.2024.…