| Setting | Treatment |
|---|---|
| HGD or T1a MUCOSAL ≤2 cm, well/moderately differentiated, no ulceration, no LVI | ENDOSCOPIC RESECTION (EMR / ESD) — preferred over oesophagectomy or gastrectomy Barrett's-related → endoscopic resection of the visible lesion + RFA ablation of the residual Barrett's segment ⚠ T1b (submucosal), poor differentiation, or LVI → formal resection
nodal risk rises steeply |
| RESECTABLE OESOPHAGEAL SCC | Neoadjuvant chemoradiation → oesophagectomy (CROSS)
carboplatin AUC2 + paclitaxel 50 mg/m² weekly ×5 with 41.4 Gy in 23 fractions, then surgery SCC is the radiosensitive histology
pCR 49% (vs 23% in adenocarcinoma); median OS 81.6 vs 21.1 months Because of that, definitive chemoradiation is a legitimate curative alternative
see the organ-preservation row below |
| RESECTABLE OESOPHAGEAL / GOJ ADENOCARCINOMA | ① Perioperative FLOT (FLOT4)
docetaxel + oxaliplatin + 5-FU/leucovorin, 4 cycles pre- and 4 post-op; median OS 50 vs 35 months · ESMO increasingly favours this for GOJ / gastric-type adenocarcinoma ② Neoadjuvant chemoradiation → surgery (CROSS) ⚠ A genuine NCCN-vs-ESMO divergence, unresolved by head-to-head data
NCCN lists both |
| RESECTABLE GASTRIC ADENOCARCINOMA | Perioperative FLOT (FLOT4) + D2 lymphadenectomy at a high-volume centre ⚠ If they went straight to surgery with a D0/D1 resection and no perioperative chemo → adjuvant chemoradiation (INT-0116)
5-FU/leucovorin + 45 Gy; median OS 36 vs 27 months, but poorly tolerated — 32% grade 4 toxicity, 31% did not complete After an adequate D2 resection it adds little (ARTIST)
reserve it for the under-staged / under-resected patient |
| AFTER neoadjuvant chemoRT — RESIDUAL pathological disease i.e. not ypT0N0 | Adjuvant nivolumab ×1 year (CheckMate-577)
median DFS 22.4 vs 11.0 months ⚠ OS difference did NOT reach significance — this is a DFS-based standard |
| 🔴 ORGAN PRESERVATION — inoperable, declines surgery, cervical oesophagus, or SCC choosing preservation | DEFINITIVE CHEMORADIATION (RTOG 85-01)
cisplatin + infusional 5-FU concurrent with RT · 5-year OS 26–27% vs 0% for radiotherapy alone DOSE — DO NOT ESCALATE: 50.4 Gy in 28 fractions is the standard
(INT-0123) / (RTOG 94-05) tested escalation to 64.8 Gy and found NO gain — no survival difference, no better locoregional control, no QoL benefit In SCC, definitive chemoradiation may equal preoperative chemoRT + surgery
NCI PDQ ⚠ Locoregional failure is the weak point
~25% persistence, ~21% recurrence → these patients need structured endoscopic surveillance, not discharge Salvage oesophagectomy
for local persistence / isolated recurrence in fit patients at high-volume centres |
| 🔴 PERITONEAL-POSITIVE M1 staging laparoscopy: visible deposits OR positive cytology alone | Systemic therapy as for M1 — curative resection is NOT indicated ⚠ CRS/HIPEC remains INVESTIGATIONAL
trials only, at peritoneal-malignancy centres — (PHOENIX-GC) showed no OS difference; (GASTRICHIP) ongoing Positive cytology with no visible disease
a distinct, better-prognosis subgroup — practice varies |
| METASTATIC — FIRST LINE, biomarker-directed interrogate in this order | ① MSI-H / dMMR → pembrolizumab (KEYNOTE-158)
ORR 30.8% · ⚠ MSI status should also change your LOCALISED plan — perioperative chemotherapy does not improve survival in dMMR/MSI-H gastric cancer ② HER2-positive → trastuzumab + chemotherapy (ToGA)
IHC 3+, or 2+ with ISH amplification · add pembrolizumab if CPS ≥1 (KEYNOTE-811) ③ HER2-negative + Claudin 18.2-positive → zolbetuximab + CAPOX/mFOLFOX6 (SPOTLIGHT) (GLOW)
moderate-to-strong (2+/3+) membranous staining in ≥75% of cells · FDA-approved Oct 2024 ④ HER2-negative adeno → nivolumab + FOLFOX/CAPOX (CheckMate-649)
NCCN category 1 at CPS ≥5; median OS 14.4 vs 11.1 months · ⚠ DIVERGENCE: ESMO/EMA restrict to CPS ≥5; NCCN allows CPS 1–4 as category 2B · or pembrolizumab + chemo (KEYNOTE-859) at CPS ≥1 ⑤ Oesophageal SCC → nivolumab + chemo OR nivolumab + ipilimumab (CheckMate-648)
⚠⚠ scored as TUMOUR-CELL PD-L1 ≥1%, NOT CPS — a different denominator and an easy, dangerous error · or pembrolizumab + chemo (KEYNOTE-590), benefit concentrated at CPS ≥10 |
| ⚠ REGULATORY — affects every PD-1 row above | ⚠ June 2025 — FDA NARROWED pembrolizumab and nivolumab labels to PD-L1 CPS ≥1
in advanced gastric / GOJ / oesophageal cancer, following a September 2024 ODAC vote that benefit did not outweigh risk in PD-L1-negative disease — the older all-comers approvals no longer stand ⚠ "PD-L1 positive" is NOT a category
always state the NUMBER and the ASSAY (CPS vs tumour-cell %) |
| SECOND LINE | Ramucirumab + paclitaxel (RAINBOW)
median OS 9.6 vs 7.4 months — the standard doublet Taxane-unfit or neuropathic → ramucirumab alone (REGARD) HER2-positive → trastuzumab deruxtecan (DESTINY-Gastric04)
now the new standard in this line, beating ramucirumab+paclitaxel head-to-head (30% reduction in risk of death; median OS 14.7 vs 11.4 months) ⚠ Re-test HER2 at progression where feasible — it can be lost |
| THIRD LINE + | Trifluridine/tipiracil (TAGS)
median OS 5.7 vs 3.6 months · irinotecan is an accepted alternative |
| CDH1 CARRIER — hereditary diffuse gastric cancer, no cancer yet | Prophylactic total gastrectomy, conventionally from age 20
guideline range 20–30 If deferred → annual surveillance endoscopy with multiple random biopsies (Cambridge protocol)
from age 40, or 10 years before the youngest family case Women also need breast MRI from age 30
lobular breast cancer risk ⚠ Counsel with CURRENT figures
cumulative advanced gastric cancer risk by age 80 is ~10.3% (men) / 6.5% (women) — far lower than the historic 70–80% estimates, and it materially changes the consent conversation |
| RISK REDUCTION & EMERGENCIES | H. pylori is a causal risk factor — test and eradicate
including in first-degree relatives, and after endoscopic resection of early gastric cancer Malignant dysphagia → self-expanding metal STENT is the fastest palliation
palliative RT 20 Gy/5 or 30 Gy/10 · intraluminal brachytherapy gives longer dysphagia-free survival where life expectancy is longer ⚠ NEVER place a PEG in a potential oesophagectomy candidate
it compromises the gastric conduit — use jejunostomy or an NJ tube Bleeding tumour → endoscopic haemostasis, haemostatic RT, or embolisation |