🤖 Robotic vs laparoscopic TME for mid/low rectal cancer. Randomised trials only this time — 4 RCTs, 1,952 patients. The pathology favours robotics: 📉 CRM positivity down (OR 0.58), complete mesorectum up (OR 1.55), conversion down (OR 0.41). Early morbidity: no difference. And the part to watch — 2 trials with 3-year data: 📉 lower locoregional recurrence (OR 0.43), a slight DFS edge (HR 0.78). OS unchanged. Robotics doesn't change survival, but it makes a hard operation more reproducibly complete — and the oncology signal is finally starting to show. doi.org/10.1007/s11701-026-0… #RoboticSurgery #RectalCancer

💊 FGFR2-rearranged cholangiocarcinoma, first line. Pemigatinib vs gem/cis. The response data are real. 🎯 ORR 47% vs 15%. PFS 8.3 vs 6.8 months, HR 0.58. Duration of response more than doubled. And yet: ⚠️ overall survival was flat. 24.4 vs 25.0 months. Why? Crossover. The chemo arm got pemigatinib later and caught up. The lesson isn't that the drug fails — it's that moving an effective drug earlier doesn't buy survival when everyone gets it eventually. First-line pemigatinib is for the patient who needs the response now. doi.org/10.1200/JCO-26-00788 #JCO #Cholangiocarcinoma

🧬 BRAF V600E metastatic colorectal cancer. First line. The original BREAKWATER paired encorafenib cetuximab with FOLFOX. Cohort 3 pairs it with FOLFIRI — and it holds. 147 patients vs FOLFIRI ± bevacizumab. 🎯 Response rate 64% vs 39%. 📈 Median PFS 15.2 vs 8.3 months. HR 0.44. OS prolonged too — HR 0.56. So the backbone is now a choice, not a constraint: irinotecan or oxaliplatin, decided on toxicity. BRAF V600E mCRC is firmly a first-line targeted-therapy disease. doi.org/10.1016/j.annonc.202… #AnnalsofOncology #ColorectalCancer

🧪 Resectable, PD-L1-positive gastric/GEJ cancer. Add a PD-1 inhibitor across the perioperative window? ASTRUM-006 says yes — and trims the adjuvant chemo. Neoadjuvant serplulimab SOX → adjuvant serplulimab, vs placebo SOX → adjuvant SOX. 588 patients. 📈 EFS at CPS ≥10: not reached vs 42 months. HR 0.65. ITT (CPS ≥5): HR 0.73. The adjuvant half is immunotherapy alone — chemo-sparing. ⚠️ And grade ≥3 toxicity was lower, not higher: 47% vs 59%. OS still immature. But the perioperative case keeps building. doi.org/10.1016/S0140-6736(2… #TheLancet #GastricCancer

🧪 Second-line metastatic pancreatic cancer. For years the honest answer was: more chemo, very little gained. RASolute 302 changes the sentence. Daraxonrasib — a pan-RAS(ON) inhibitor — vs investigator's-choice chemo. 500 patients, 92% RAS G12. 📈 Median OS 13.2 vs 6.6 months. HR 0.40. PFS 7.3 vs 3.5 months. And it was the better-tolerated arm: ⚠️ grade ≥3 events 62% vs 70%, discontinuations 1.2% vs 11%. A HR of 0.40 in pretreated PDAC is a number this field has basically never produced. doi.org/10.1056/NEJMoa260555… #NEJM #PancreaticCancer

🔬 Intrahepatic cholangiocarcinoma. Who actually benefits from adjuvant therapy? Maybe three stains can tell you. Multi-omics found a high-risk, mucin-enriched subtype. Marked by claudin-18, MUC1, MUC5AC. Worst survival when left untreated. 📈 But in 174 resected patients, the marker-positive ones did markedly better with adjuvant chemo or chemoembolisation. Marker-negative: no clear benefit. 💡 A small IHC panel to decide who to treat after resection. Needs prospective validation. And claudin-18 keeps showing up across the biliary tree. doi.org/10.1016/j.modpat.202… #Cholangiocarcinoma

🤖 Robotic vs laparoscopic gastrectomy. Japan's national registry. 2023–24. Nearly 11,000 matched pairs. Distal gastrectomy: 📉 Lower 30-day morbidity, robotic vs lap. 4.3% vs 4.9%. Less blood loss, fewer conversions, shorter stay. Total gastrectomy: Morbidity the same. 8.7% vs 8.3%. Robotic still bled less and converted less — but didn't move the main endpoint. 💡 The robotic edge in gastric cancer is real, but it's procedure-specific. Long-term oncology and cost still open. doi.org/10.1007/s10120-026-0… #GastricCancer

🧠 A "low-risk" pancreatic cyst. Stable for years. Safe to stop watching? 1,494 IPMNs under surveillance vs the SEER population: 📈 ~10x the pancreatic cancer risk overall. Main duct ≥5 mm: 26x. BMI ≥30: 21x. Here is the part that matters: Even guideline-negative cysts ran 8x the risk. After 5 stable years, the curve kept climbing. No plateau. ⚠️ "Low-risk" is not "no-risk." And year 5 is not a finish line. doi.org/10.1097/SLA.00000000… #AnnalsofSurgery

🩺 Oesophageal cancer. Restaging after chemoradiation comes back "uncertain." Watch, or operate? SANO looked at 272 uncertain responders. 205 went to surgery. Only 15% had no residual tumour. So 85% still had cancer. 🎯 "Uncertain" is not "complete." Default to surgery. One exception: squamous cancer a non-traversable stricture. There, 33% had a pathologic complete response. That subgroup earns a real conversation — not a reflex resection. doi.org/10.1097/SLA.00000000… #AnnalsofSurgery

🧪 HER2-positive gastroesophageal cancer, first line. For 15 years the answer was trastuzumab chemo. HERIZON-GEA-01 says that can change. Zanidatamab is a dual-HER2 bispecific. It beat trastuzumab on PFS. HR 0.65. 📈 Add tislelizumab and median OS reaches 26.4 months. HR for death 0.72. The catch: the survival win needs the PD-1. Zanidatamab alone — OS 24.4 months, not significant. ⚠️ And grade ≥3 toxicity ~73%. Diarrhoea in a quarter. Real gain. Real toxicity. Pick the patient. doi.org/10.1056/NEJMoa251772… #NEJM #GastricCancer

Nationwide Dutch data (18 hospitals, 2,384 patients): multimodal prehabilitation before colorectal cancer surgery cut overall complications from 37.8% to 30.1% (OR 0.71), nearly halved medical complications, and shaved a day off length of stay — across every age and ASA grade. The complication you prevent in clinic is the one you don't fight in the ICU. doi.org/10.1001/jamasurg.202… #JAMASurgery

Intraoperative AI that flags the pancreas, common hepatic artery, left gastric artery and vein during robotic gastrectomy cut unsafe incision-line choices (OR 0.25) and found the hepatic artery faster — across 67 videos and 20 surgeons. Vessel recognition is still modest. A cognitive aid for suprapancreatic dissection, not an autopilot. doi.org/10.1007/s10120-026-0… #GastricCancer

RESPONDER (n=844, 8 centres): in rectal cancer, when lateral pelvic nodes ≥5 mm shrink to under 5 mm after chemoradiation, omitting lateral node dissection gave the same 5-year local recurrence (5.3% vs 3.1%) and disease-free survival — with less operative morbidity. The response carries the prognosis. Dissect the node that doesn't shrink; spare the one that does. doi.org/10.1097/SLA.00000000… #AnnalsofSurgery

Tissue-free ctDNA in stage III colon cancer, validated in 2,260 patients from Alliance N0147. Detectable after surgery in 1 in 5. Recurrence HR ~6. Five-year disease-free survival 27.7% if positive vs 77.1% if negative — and the gap was widest in the lower-risk and dMMR patients we're least certain about. No tumour block required. Clearly prognostic now; whether acting on it improves outcomes is the next trial. doi.org/10.1200/JCO-25-02086 #JCO

DRAGON-01 (phase 3, n=222, gastric cancer with peritoneal-only metastasis): adding intraperitoneal to IV paclitaxel S-1 improved median OS from 13.9 to 19.4 months (HR 0.67), and 3-year OS from 12% to 25%. Conversion surgery 51% vs 35%; positive cytology cleared in 84% vs 53%. No extra grade 3–4 toxicity. First positive phase 3 for intraperitoneal chemo here — PHOENIX-GC was negative. The backbone predates chemo-immunotherapy, so integration is the next question. doi.org/10.1001/jamaoncol.20… #JAMAOncology

MATTERHORN final OS: durvalumab FLOT beats FLOT alone in resectable G/GEJ. HR 0.78; p=0.021. 🎯 The first checkpoint regimen to move OS in this curative-intent setting. KEYNOTE-585 couldn't. doi.org/10.1056/NEJMoa250370…

One dose of pembrolizumab. 44% pCR in stage I–III dMMR colon cancer. RESET-C in JCO. 💡 The non-operative-management trial design writes itself from here. doi.org/10.1200/JCO-25-02274

NORDICC, 13-year follow-up. Distal CRC: RR 0.79. Proximal CRC: RR 0.91, not significant. ⚠️ The right colon remains the screening problem nobody has solved. doi.org/10.1016/S0140-6736(2…