Cutting Through the Complexity: Precision Medicine in Gastric Cancer with Perioperative Immunotherapy

Precision Medicine in Gastric Cancer with Perioperative Immunotherapy

Hematology / Oncology · Seminar week 19 · released July 30, 2026 · includes a discussion video

Recent developments in the integration of immunotherapy in resectable gastric cancer, as demonstrated in the Matterhorn trial, have reshaped standard care. This is essential as…

Learning Objectives

Stage resectable gastric and gastroesophageal junction adenocarcinoma using an anatomically and biologically complete pretreatment evaluation.
Contrast surgery-alone, postoperative, chemoradiation, and perioperative treatment strategies using landmark trial evidence.
Explain the mechanistic and clinical rationale for combining FLOT chemotherapy with immune-checkpoint blockade.
Appraise the efficacy, safety, surgical, biomarker, and regulatory findings of the MATTERHORN trial.
Plan gastrectomy or gastroesophagectomy safely after chemoimmunotherapy.
Recognize and manage overlapping cytotoxic, immune-mediated, nutritional, and postoperative toxicities.
Integrate MSI/MMR, HER2, PD-L1, CLDN18.2, FGFR2b, and circulating tumor DNA into contemporary clinical reasoning without overextending immature evidence.

Overview of Gastric Cancer and Historical Treatment Options

%%FIG0%% Gastric cancer is not one disease. Intestinal-type tumors often develop through chronic inflammation, atrophy, intestinal metaplasia, and dysplasia—frequently driven by Helicobacter pylori—whereas diffuse and signet-ring carcinomas may infiltrate the stomach wall without forming a discrete mass and have a particular propensity for peritoneal dissemination. Relevant etiologies include smoking, high-salt preserved foods, autoimmune gastritis, Epstein–Barr virus, and inherited syndromes such as germline CDH1-associated hereditary diffuse gastric cancer and Lynch syndrome. The Cancer Genome Atlas framework—chromosomal instability, microsatellite instability, Epstein–Barr virus, and genomically stable disease—helps explain biological heterogeneity, although treatment decisions still rely predominantly on stage, anatomy, histology, performance status, and validated biomarkers.

MUST ACT: Establish whether the cancer is truly localized before committing to curative-intent treatment. Evaluation should include high-quality esophagogastroduodenoscopy with sufficient biopsies, contrast-enhanced CT of the chest, abdomen, and pelvis, multidisciplinary pathology review, and selective endoscopic ultrasonography when better definition of T or regional nodal stage would change management. PET/CT may clarify equivocal distant disease but is less sensitive in diffuse, mucinous, and signet-ring cancers. Diagnostic laparoscopy with peritoneal washings is particularly important for cT3–T4, node-positive, diffuse-type, or linitis plastica presentations because CT can miss low-volume peritoneal metastases. Positive peritoneal cytology constitutes metastatic disease even when no implants are visible.

Very early, properly selected T1a cancers can be cured by endoscopic submucosal dissection. Most medically fit patients with cT2 or node-positive, M0 disease require multimodality treatment. The surgical foundation is an oncologic subtotal or total gastrectomy, selected according to tumor location and achievable margins, with a formal D2 lymphadenectomy at an experienced center. At least 16 nodes are required for adequate pathologic staging, but a well-executed D2 dissection usually retrieves more.

Surgery alone historically produced frequent locoregional and distant relapse. INT-0116 established postoperative fluoropyrimidine-based chemoradiotherapy as superior to surgery alone, with median overall survival of 36 versus 27 months (PMID: 11547741). Its interpretation is tempered by limited lymphadenectomy: many participants had D0 or D1 rather than D2 surgery. Adjuvant chemoradiation therefore remains especially relevant after inadequate nodal dissection, an R1 margin when re-resection is not feasible, or selected high-risk situations—not as an automatic substitute for appropriate surgery and systemic therapy.

East Asian trials demonstrated that adjuvant chemotherapy improves outcomes after D2 gastrectomy. CLASSIC used eight postoperative cycles of capecitabine plus oxaliplatin and improved five-year disease-free survival from 53% to 68% and overall survival from 69% to 78% (PMIDs: 22226517, 25439693). ACTS-GC similarly supported postoperative S-1. These strategies remain important when cancer is discovered only after upfront gastrectomy or in regional practice patterns favoring surgery first.

The Western perioperative model began with MAGIC: three cycles of ECF before and after surgery improved five-year survival from approximately 23% to 36% versus surgery alone (PMID: 16822992). FLOT4 then replaced epirubicin-based therapy. Four preoperative and four postoperative doses of FLOT improved median overall survival to 50 months versus 35 months with ECF/ECX, with a hazard ratio for death of 0.77 (PMID: 30982686).

Nuance: CROSS is a landmark for esophageal and junctional—not distal gastric—cancer. It used weekly carboplatin AUC 2 and paclitaxel 50 mg/m² with 41.4 Gy in 23 fractions before surgery and improved survival in esophageal/GEJ cancer (PMID: 22646630). Its frequently cited complete-response advantage was greatest in squamous carcinoma. For adenocarcinoma centered at the GEJ, the choice between perioperative FLOT and chemoradiation must reflect Siewert anatomy, anticipated operation, nodal distribution, dysphagia, and local expertise.

Framework: Ask three questions: Is the disease endoscopically curable? Is it resectable but at meaningful systemic-relapse risk? Is there occult or overt metastatic disease? Errors at this branch point cannot be rescued by simply intensifying therapy later.

Audience Poll: Which finding most often changes your initial plan: positive peritoneal cytology, a Siewert I rather than Siewert III location, MSI-high status, or marginal physiologic fitness for FLOT?

Perioperative Chemotherapy and Immunotherapy: Definitions and Rationale

%%FIG1%% “Neoadjuvant” treatment occurs before surgery; “adjuvant” treatment follows surgery; “perioperative” therapy deliberately spans both periods. This distinction matters. Preoperative treatment attacks micrometastases early, may reduce tumor and nodal burden, tests disease biology, and is delivered when nutrition and performance status are generally better. Surgery provides definitive local control and pathologic staging. Postoperative therapy targets residual systemic disease, although gastrectomy-related weight loss, complications, neuropathy, and deconditioning often limit its delivery.

FLOT is administered every 14 days: docetaxel 50 mg/m², oxaliplatin 85 mg/m², leucovorin 200 mg/m², and fluorouracil 2,600 mg/m² as a 24-hour infusion. The conventional course is four doses before surgery and four afterward. Docetaxel disrupts microtubules; oxaliplatin creates DNA cross-links; fluorouracil inhibits thymidylate synthase and disrupts RNA processing; leucovorin stabilizes fluorouracil’s interaction with thymidylate synthase. Their nonidentical mechanisms improve cytoreduction but also generate overlapping marrow, gastrointestinal, neurologic, and nutritional toxicity.

Teaching Point: The main rationale for perioperative systemic treatment is not merely “shrinking the primary.” Recurrence after an apparently complete gastrectomy often represents micrometastatic disease present at diagnosis. The preoperative interval is therefore an opportunity to treat the lethal component of the cancer while resectability is preserved.

Durvalumab is a monoclonal antibody against PD-L1. By blocking PD-L1 interactions with PD-1 and CD80, it releases inhibitory signaling that restrains activated T cells. Chemotherapy can complement this effect through tumor-antigen release, immunogenic cell death, depletion of suppressive immune populations, and remodeling of the tumor microenvironment. Oxaliplatin can promote calreticulin exposure and danger-signal release; taxanes can enhance antigen presentation and modify myeloid suppression. Administering checkpoint blockade while the primary tumor and draining lymphatic tissue remain present may expose the immune system to a broader antigen repertoire than purely postoperative treatment.

That biological argument is plausible but insufficient by itself. KEYNOTE-585 showed that increasing pathologic response with perioperative pembrolizumab-based therapy did not automatically translate into a statistically definitive event-free-survival result at its reported analysis. DANTE demonstrated greater pathologic regression with atezolizumab plus FLOT, especially in immune-enriched subgroups, but was not practice-defining for survival. These experiences underscore why MATTERHORN’s event-free and overall survival outcomes—not pCR alone—changed practice.

Framework: Before recommending FLOT plus durvalumab, assess oncologic indication and treatment fitness separately. Oncologic factors include cT2 or greater disease, nodal involvement, resectability, absence of metastasis, and a realistic curative operation. Fitness includes ECOG status, renal and hepatic function, marrow reserve, neuropathy, cardiac history, autoimmune disease, transplant history, chronic immunosuppression, nutrition, frailty, and patient goals. A regimen tested largely in ECOG 0–1 patients should not be reflexively extrapolated to an ECOG 2 patient with sarcopenia and borderline organ function.

Biomarker interpretation requires discipline. MMR or MSI testing should be obtained in all newly diagnosed gastric/GEJ adenocarcinomas. HER2 should be assessed early because it informs trials and metastatic contingencies. PD-L1 is commonly reported as combined positive score in advanced gastric cancer, but MATTERHORN stratified patients using tumor area positivity, or TAP, with the SP263 assay. TAP and CPS are not interchangeable.

Nuance: The FDA indication for perioperative durvalumab does not require PD-L1 positivity. Directionally favorable MATTERHORN estimates were seen across TAP strata, but only about 10% of participants had TAP below 1%, so the PD-L1-low estimate was imprecise. “Biomarker-unselected approval” does not mean that every biological subgroup has an equally certain magnitude of benefit.

MSI-high/dMMR tumors are highly immunogenic and may be less dependent on conventional chemotherapy. Small phase II studies of chemotherapy-free immunotherapy have produced striking pCR rates, but neither omission of FLOT nor omission of surgery is yet a routine standard. Conversely, microsatellite-stable status is not a reason to withhold the approved MATTERHORN regimen.

Decision Point: In a fit patient with resectable cT3N1 gastric adenocarcinoma, the practical default is perioperative FLOT plus durvalumab. In a frail patient, a patient with uncontrolled autoimmune disease, or one with severe baseline neuropathy, the question is not simply whether durvalumab “works”; it is whether the complete treatment pathway can be delivered safely enough to preserve the chance of curative surgery.

Audience Poll: Would you offer the same perioperative regimen to a fit 45-year-old with cT3N2 disease, a fit 78-year-old with cT2N0 disease, and a patient receiving prednisone 20 mg daily for active autoimmune pneumonitis?

Detailed Analysis of the MATTERHORN Trial

%%FIG2%% MATTERHORN, NCT04592913, was a global phase III, double-blind, placebo-controlled trial conducted at 147 centers in 20 countries. It randomized 948 adults with previously untreated, resectable AJCC eighth-edition stage II–IVA gastric or GEJ adenocarcinoma, ECOG 0–1, and no peritoneal or distant metastases. Approximately two thirds had gastric primaries, 70% were clinically node-positive, and most had stage III disease. Randomization was stratified by Asian versus non-Asian region, clinical nodal status, and PD-L1 TAP of at least 1% versus below 1%.

Participants received durvalumab 1,500 mg IV or placebo on day 1 every four weeks. FLOT was given on days 1 and 15 of each 28-day cycle for two cycles before surgery and two after surgery—four FLOT administrations in each phase. Surgery was planned four to eight weeks after neoadjuvant therapy; postoperative therapy began four to twelve weeks after surgery. After postoperative FLOT plus durvalumab or placebo, participants received ten additional every-four-week doses of durvalumab or placebo. The FDA-labeled durvalumab dose for adults weighing under 30 kg is 20 mg/kg every four weeks.

MUST ACT: Avoid the common dosing error created by the word “cycle.” In MATTERHORN, two 28-day neoadjuvant cycles contained four every-two-week FLOT administrations but only two durvalumab doses. The complete plan contained eight FLOT administrations, four perioperative durvalumab doses given with chemotherapy, and ten additional postoperative durvalumab doses.

The primary endpoint was blinded-independent-central-review event-free survival: progression precluding surgery or requiring off-protocol treatment, recurrence or progression after surgery, or death. Key secondary endpoints were overall survival and centrally reviewed pCR, defined as no viable tumor in either the primary site or sampled lymph nodes—ypT0N0—using modified Ryan criteria.

At a median follow-up of 31.5 months, median EFS was not reached with durvalumab plus FLOT and was 32.8 months with placebo plus FLOT. The hazard ratio was 0.71 (95% CI, 0.58–0.86; P<0.001). Two-year EFS was 67.4% versus 58.5%, an absolute difference of 8.9 percentage points. pCR was 19.2% versus 7.2%, with a relative risk of 2.69 (95% CI, 1.86–3.90). The peer-reviewed primary report is Janjigian et al., NEJM 2025 (PMID: 40454643).

Nuance: The initial publication’s interim OS comparison did not cross its unusually stringent prespecified boundary: two-year OS was 75.7% versus 70.4%, but P=0.03 exceeded the interim threshold of P<0.0001. The curves were nonproportional, with no early separation during months 0–12 and a later hazard ratio favoring durvalumab. At the prespecified final analysis, after approximately 43 months’ median follow-up, there were 160 versus 192 deaths. Median OS remained unreached; the hazard ratio was 0.78 (95% CI, 0.63–0.96; P=0.021), crossing the final threshold of P<0.0499. Three-year OS was 68.6% versus 61.9%.

Safety was dominated by FLOT but not identical between groups. Maximum grade 3–4 adverse events occurred in 71.6% with durvalumab and 71.2% with placebo. Immune-mediated events occurred in 23.2% versus 7.2%, with grade 3–4 immune-mediated events in 7.2% versus 3.6%. Any treatment component was discontinued because of toxicity in 29.9% versus 22.8%. Common events included diarrhea, nausea, neutropenia, anorexia, fatigue, alopecia, anemia, and neuropathy.

Surgery was completed in 86.9% versus 84.4% of randomized participants. Among those completing surgery, R0 resection occurred in 91.5% versus 92.3%. Thus, durvalumab did not increase R0 resection; it improved systemic and pathologic outcomes without materially compromising surgical delivery. Surgical delay rates were 10.1% versus 10.8%, and adverse-event-related delays were 2.3% versus 2.6%.

Decision Point: The regimen received FDA approval on November 25, 2025, for adults with resectable gastric or GEJ adenocarcinoma without an MSI, HER2, or PD-L1 requirement (FDA approval summary). However, MATTERHORN cannot identify whether benefit came from neoadjuvant durvalumab, postoperative chemoimmunotherapy, maintenance durvalumab, or their combination. Planned deletion of one phase is not evidence-based.

Teaching Point: Pathologic response is prognostic, not a validated individual surrogate for cure. A patient with residual ypN-positive disease remains at high risk; a patient with pCR still requires surveillance and should not automatically omit protocol-intended postoperative treatment.

Audience Poll: Which result is most practice-changing: the EFS hazard ratio, the final OS result, the 12-point increase in pCR, or the absence of a surgical-safety penalty?

Clinical Implications for Surgical Procedures Post-Immunotherapy

%%FIG3%% The operation remains indispensable. Neither radiographic response, negative repeat biopsies, nor apparent clinical complete response can reliably exclude viable tumor in the gastric wall or regional nodes. Outside a prospective organ-preservation study, response to chemoimmunotherapy should lead to definitive resection—not watchful waiting.

Restaging CT should be performed after neoadjuvant therapy and before surgery. The purpose is primarily to identify interval metastatic disease or clear progression, not to demand a dramatic RECIST response from a gastric-wall lesion that is difficult to measure. Diffuse-type tumors and peritoneal disease may remain poorly visualized; repeat laparoscopy can be considered when initial risk was high or new ascites, serosal thickening, or unexplained symptoms appear. Apparent enlargement also requires judgment: true progression is more common than immune pseudoprogression, but an equivocal isolated finding should be confirmed by targeted imaging, biopsy, or laparoscopy when cancellation of curative surgery is at stake.

Framework: Surgical readiness after chemoimmunotherapy has four domains: absence of prohibitive progression, recovery from cytotoxic toxicity, control of immune-mediated toxicity, and physiologic capacity for the proposed operation. Review CBC, renal and hepatic function, albumin and weight trajectory, neuropathy, cardiopulmonary symptoms, corticosteroid exposure, and all suspected immune toxicities. Prehabilitation should address aerobic conditioning, inspiratory training, smoking cessation, glycemic control, anemia, and protein-calorie depletion.

MATTERHORN planned surgery four to eight weeks after the last neoadjuvant dose. That interval permits marrow and mucosal recovery without an unnecessarily prolonged treatment gap. There is no evidence-based requirement to delay surgery solely to “wash out” durvalumab if the patient is otherwise ready. Conversely, a fixed calendar date should not override unresolved pneumonitis, myocarditis, hepatitis, severe colitis, adrenal insufficiency, infection, or persistent cytopenia.

MUST ACT: New dyspnea before gastrectomy is not automatically deconditioning. Evaluate pulmonary embolism, infection, anemia, heart failure, immune pneumonitis, and myocarditis. New hypotension, hyponatremia, nausea, or unexplained fatigue should prompt assessment for adrenal insufficiency. Patients with established adrenal failure require perioperative stress-dose glucocorticoids; untreated adrenal crisis can mimic sepsis and cause cardiovascular collapse.

Operation selection remains anatomical. A distal or subtotal gastrectomy is appropriate when an adequate proximal margin and oncologic nodal clearance are achievable; otherwise, total gastrectomy is required. GEJ tumors may require total gastrectomy with distal esophagectomy or transthoracic gastroesophagectomy depending on Siewert type, esophageal extension, nodal basins, and reconstruction. D2 lymphadenectomy should preserve the pancreas and spleen unless directly involved. Multivisceral resection is justified only when an en-bloc R0 operation remains technically and physiologically realistic.

The 2026 MATTERHORN surgical analysis is reassuring but should be stated precisely. Serious events considered possibly surgery-related occurred in 12.8% with durvalumab and 13.0% with placebo. Among completed operations, R0 resection was 91.5% versus 92.3%, and D2/D3 lymphadenectomy was performed in 91.0% versus 93.3%. Thirty-day mortality was 1.2% versus 1.5%; 90-day mortality was 3.1% versus 2.0% (ASCO GI surgical analysis). The trial therefore supports feasibility and no clear excess in serious surgical morbidity; it does not prove lower postoperative mortality.

Teaching Point: Postoperative fever, diarrhea, transaminitis, hypoxemia, or hypotension has a broadened differential after checkpoint blockade. Anastomotic leak, abscess, pneumonia, venous thromboembolism, medication injury, and ordinary postoperative physiology remain more common than many immune toxicities and must be excluded urgently. The reverse error is also dangerous: persistent “postoperative” hypoxemia may be pneumonitis, and refractory vasoplegia may be adrenal crisis or myocarditis.

Pathology should document ypTNM stage, margins, number of examined and involved nodes, treatment-regression score, lymphovascular and perineural invasion, and histologic subtype. Adjuvant treatment was started four to twelve weeks after surgery in MATTERHORN. Recovery, wound integrity, nutrition, organ function, neuropathy, and residual immune toxicity determine the exact date. Durvalumab monotherapy could continue if postoperative FLOT was discontinued for toxicity, but this flexibility should support individualized salvage—not preplanned chemotherapy omission.

Audience Poll: Which would make you postpone surgery: grade 1 thyroid dysfunction on replacement, an ANC of 0.8 × 10⁹/L, resolving grade 2 pneumonitis on prednisone, or an indeterminate new peritoneal nodule?

Toxicity Management and Supportive Care Strategies

%%FIG4%% Successful perioperative therapy depends on preserving curability, not achieving perfect dose intensity at any cost. Before treatment, document CBC, electrolytes, renal and hepatic function, glucose, TSH and free T4, weight, nutritional intake, bowel pattern, neuropathy, cardiopulmonary history, autoimmune disease, and medications. Consider baseline ECG and troponin in patients with cardiovascular disease or according to institutional immunotherapy pathways. Evaluate fluoropyrimidine risk, including DPYD testing where routinely implemented; unexpectedly severe early mucositis, diarrhea, cytopenia, neurotoxicity, or cardiotoxicity should trigger immediate cessation of fluorouracil and assessment for DPD deficiency.

FLOT requires deliberate prophylaxis. Use a serotonin-3 antagonist, dexamethasone, and an NK1 antagonist for emesis prevention; add olanzapine for patients with substantial nausea risk. Docetaxel requires steroid premedication unless contraindicated. Primary G-CSF is appropriate when the estimated febrile-neutropenia risk is at least 20%, or at intermediate risk when age, frailty, malnutrition, comorbidity, prior neutropenia, or curative intent makes an avoidable delay consequential. Many centers use prophylactic pegfilgrastim with FLOT, but the decision should be individualized.

MUST ACT: Fever during FLOT is neutropenic sepsis until proven otherwise. A single temperature of 38.3°C or at least 38.0°C sustained for an hour warrants immediate assessment, CBC with differential, cultures, lactate and organ-function evaluation, and timely empiric antipseudomonal antibiotics when neutropenia is present or expected. Do not attribute fever to durvalumab before excluding infection.

Oxaliplatin causes acute cold-triggered dysesthesia and cumulative sensory neuropathy. Counsel patients to avoid cold exposure immediately after treatment, assess function before every dose, and reduce or omit oxaliplatin when persistent neuropathy interferes with instrumental activities. Docetaxel contributes neutropenia, mucositis, edema, nail changes, and neuropathy. Fluorouracil can cause mucositis, diarrhea, hand-foot toxicity, encephalopathy, and coronary vasospasm. Chest pain during infusion requires immediate discontinuation, ECG and troponin assessment, and cardiology-guided evaluation; casually rechallenging without a risk-mitigation plan is unsafe.

Nutrition is treatment, not an afterthought. Obtain early oncology-dietitian input; treat nausea, pain, obstruction, constipation, oral candidiasis, and pancreatic or bile-acid-related malabsorption when present. Favor frequent energy-dense, high-protein meals and enteral support when the gut is usable. Monitor iron, B12, folate, vitamin D, calcium, and postgastrectomy dumping or hypoglycemia. Routine parenteral nutrition is not benign and should be reserved for patients in whom enteral intake cannot meet needs.

Checkpoint toxicity can involve any organ and may appear after the final dose. Common syndromes include rash, thyroiditis, hepatitis, colitis, and pneumonitis; less common but immediately dangerous events include myocarditis, myositis, myasthenic syndromes, nephritis, hypophysitis, adrenal insufficiency, encephalitis, and severe cutaneous reactions.

Framework: For a new symptom, ask: Is this infection, cytotoxic toxicity, immune toxicity, progression, a surgical complication, or an unrelated disease? Diarrhea may reflect fluorouracil, enteric infection, laxatives, pancreatic insufficiency, dumping, or immune colitis. Dyspnea may be pneumonia, embolism, anemia, fluid overload, pneumonitis, or myocarditis. Weakness may represent deconditioning, neuropathy, electrolyte loss, adrenal failure, myositis, or a neuromuscular junction disorder.

For suspected immune toxicity, grade severity and evaluate the organ involved. Grade 1 toxicity can often be monitored while durvalumab continues, except when cardiac, neurologic, hematologic, or certain pulmonary toxicity is suspected. For most grade 2 events, hold durvalumab and begin prednisone 0.5–1 mg/kg/day if an inflammatory immune toxicity is confirmed or strongly suspected. Grade 3 events generally require hospitalization or close specialty management, durvalumab interruption, and methylprednisolone or prednisone 1–2 mg/kg/day. Grade 4 toxicity usually requires permanent discontinuation, except controlled endocrinopathies managed with hormone replacement. Durvalumab is held or stopped; it is not dose-reduced.

Teaching Point: Steroids should follow appropriate diagnostic sampling when the patient is stable, but treatment must not be dangerously delayed in myocarditis, severe pneumonitis, adrenal crisis, neurologic toxicity, or fulminant colitis. Taper corticosteroids over at least four to six weeks after improvement. Provide Pneumocystis prophylaxis, bone protection, glucose monitoring, and gastric protection when prolonged high-dose therapy makes them appropriate.

Immune colitis requires stool pathogen testing and assessment for dehydration; colonoscopy or flexible sigmoidoscopy is useful in persistent or severe disease. Steroid-refractory cases may require infliximab or vedolizumab. Avoid infliximab in immune hepatitis, where mycophenolate is a preferred second-line agent. Pneumonitis requires CT imaging and exclusion of infection and embolism. Suspected myocarditis warrants ECG, troponin, natriuretic peptide, echocardiography, urgent cardio-oncology consultation, and high-dose corticosteroids; a normal ejection fraction does not exclude it. Isolated hypothyroidism is treated with replacement rather than high-dose steroids. Suspected adrenal crisis requires immediate IV hydrocortisone and fluids before confirmatory testing is complete.

Decision Point: Rechallenge is an organ- and severity-specific decision. It may be reasonable after complete resolution of selected grade 2 toxicities and steroid taper, but is usually inappropriate after myocarditis, severe neurologic toxicity, grade 4 events, or recurrent severe pneumonitis. The curative setting raises both sides of the calculation: durable immune control is valuable, but compromising surgery with uncontrolled toxicity is unacceptable.

Audience Poll: A patient has six watery stools daily after FLOT plus durvalumab. Which action comes first: loperamide alone, empiric infliximab, stool and laboratory evaluation with treatment held, or permanent discontinuation of all therapy?

Future Directions in Gastric/GEJ Cancer Therapeutics

%%FIG5%% The next advance will not simply add another drug to every patient’s regimen. It will identify which tumors require intensification, which are uniquely immune-sensitive, and which can safely avoid ineffective components. That requires pretreatment tissue adequate for MMR/MSI, HER2, and PD-L1 assessment, with CLDN18.2 and broader genomic testing considered for trials and future metastatic planning.

MSI-high/dMMR localized gastric cancer is the clearest candidate for biological de-escalation. In the phase II NEONIPIGA study, neoadjuvant nivolumab plus low-dose ipilimumab produced pCR in 17 of 29 resected patients, or 58.6% (PMID: 35969830). INFINITY has also explored durvalumab plus tremelimumab and a nonoperative strategy. These are provocative results, but small single-arm cohorts cannot establish that chemotherapy or gastrectomy can be safely omitted. Endoscopy and CT may miss residual nodal or mural disease, and salvage surgery after delayed regrowth is not yet a proven equivalent.

Decision Point: For a fit patient with resectable MSI-high disease, reasonable discussion includes the FDA-approved FLOT–durvalumab pathway, an MSI-directed clinical trial, and the uncertainties surrounding conventional chemotherapy benefit. Routine immunotherapy alone or watch-and-wait should remain trial-based.

HER2 is another logical perioperative target. Trastuzumab improves outcomes in advanced HER2-positive disease, and pembrolizumab plus trastuzumab and chemotherapy has become important for PD-L1-positive advanced disease. Perioperative studies such as PETRARCA and PHERFLOT have reported high pathologic-response rates using HER2 blockade with FLOT, sometimes with immunotherapy, but small samples, treatment-related diarrhea, postoperative complications, and limited survival follow-up prevent routine adoption. Antibody–drug conjugates such as trastuzumab deruxtecan are moving earlier, but pneumonitis and perioperative safety require careful study.

CLDN18.2 has become a validated target in advanced HER2-negative gastric/GEJ cancer. SPOTLIGHT and GLOW established zolbetuximab with fluoropyrimidine-platinum chemotherapy, leading to FDA approval in 2024 for CLDN18.2-positive unresectable or metastatic disease (FDA approval summary). Nausea and vomiting are prominent on-target practical toxicities. Whether zolbetuximab, CLDN18.2-directed antibody–drug conjugates, bispecific antibodies, or cellular therapies improve cure rates perioperatively remains investigational.

FGFR2b overexpression or amplification defines another subset. Bemarituzumab, an FGFR2b-directed antibody, produced encouraging results with chemotherapy in advanced disease and is undergoing phase III evaluation. Ocular toxicity, biomarker-assay performance, intratumoral heterogeneity, and the optimal expression threshold must be resolved before perioperative use. Other targets—including MET amplification, rare NTRK fusions, and emerging immune checkpoints such as TIGIT—should be pursued through genotype- or phenotype-matched trials rather than empiric off-label combinations.

Nuance: Biomarkers overlap and change. A tumor may be HER2-positive and PD-L1-positive, or CLDN18.2-positive with heterogeneous HER2 expression. Primary and metastatic lesions can differ, and HER2 may be lost after trastuzumab. Repeat biopsy at recurrence is therefore often more informative than indefinitely relying on the diagnostic specimen.

Circulating tumor DNA may ultimately connect response to treatment intensity. Postoperative ctDNA positivity is strongly prognostic, and longitudinal studies such as PLAGAST found that persistent ctDNA during neoadjuvant therapy, after therapy, or after surgery identified patients at markedly higher recurrence risk (PMID: 40707450). The limitations are equally important: diffuse and peritoneal-predominant cancers may shed little DNA; postoperative timing affects background signal; assay sensitivity varies; and prognostic association does not prove that changing therapy improves survival.

MUST ACT: Do not omit adjuvant treatment because ctDNA is negative or add unvalidated therapy because it is positive outside a clinical trial. The necessary next step is randomized ctDNA-guided escalation and de-escalation, not more observational correlation.

Future trials should incorporate molecular subtype, serial ctDNA, standardized pathologic regression, patient-reported outcomes, nutritional recovery, and surgery quality. Adaptive platforms may test HER2-, CLDN18.2-, FGFR2b-, or immune-directed regimens within biologically defined cohorts. They must also answer a practical question MATTERHORN could not: how much benefit comes from neoadjuvant checkpoint blockade, postoperative combination therapy, and maintenance individually?

Framework: Precision medicine is not maximal therapy. It is matching the necessary therapy to the tumor while protecting the patient’s opportunity for complete resection, functional recovery, and durable cure.

Audience Poll: Which strategy is most likely to change practice next: MSI-directed omission of chemotherapy, ctDNA-guided adjuvant therapy, perioperative HER2/CLDN18.2 targeting, or validated omission of surgery after complete immune response?

Clinical Case: A 58-Year-Old Man with Stage II Gastric Cancer

A 58-year-old man presents with early satiety, iron-deficiency anemia, and a 6-kg weight loss. Endoscopy shows a 4.5-cm ulcerated antral mass. Biopsy demonstrates poorly differentiated adenocarcinoma without signet-ring morphology. CT shows gastric-wall thickening and one suspicious perigastric node but no distant disease. Endoscopic ultrasonography suggests cT3N1 disease, clinical stage IIB. Diagnostic laparoscopy shows no implants, and peritoneal cytology is negative. He has ECOG 0 status, normal organ function, no neuropathy or autoimmune disease, and adequate cardiac function.

Pathology demonstrates proficient MMR/microsatellite-stable disease, HER2 IHC 0, and PD-L1 TAP 3%. These results help characterize the tumor but do not create the indication: his resectable, locally advanced disease and fitness for therapy do. The multidisciplinary team recommends perioperative FLOT plus durvalumab followed by gastrectomy and postoperative therapy.

Decision Point: He receives durvalumab 1,500 mg on days 1 and 29, with FLOT every two weeks for four administrations. Before each dose, the team reviews CBC, chemistry, liver tests, weight, intake, bowel function, neuropathy, rash, respiratory symptoms, and endocrine symptoms. He receives docetaxel premedication, multidrug antiemetic prophylaxis, and G-CSF because of curative intent and institutional assessment of neutropenic risk.

After the third FLOT administration, he reports five additional loose stools daily. He is afebrile and hemodynamically stable, with ANC 1.8 × 10⁹/L and no abdominal tenderness. The differential includes fluorouracil toxicity, infection, laxative exposure, and immune colitis. Both treatment components are held while stool C. difficile and multiplex pathogen testing, electrolytes, inflammatory markers, and hydration status are assessed. Infectious testing is negative, and the diarrhea resolves within 48 hours with fluids and antidiarrheal therapy without corticosteroids. The short, cycle-linked course supports chemotherapy-related diarrhea more than immune colitis. Persistent symptoms, blood, pain, fever, or inflammatory findings would instead prompt imaging, gastroenterology evaluation, endoscopy when appropriate, and immune-colitis treatment.

Restaging shows no metastases and modest reduction in wall thickening. Six weeks after neoadjuvant therapy, he undergoes subtotal gastrectomy with D2 lymphadenectomy. Pathology shows an R0 resection, ypT1bN0 disease, treatment response, and 0 of 31 involved nodes. He has not achieved pCR, but substantial downstaging is prognostically favorable.

Teaching Point: Neither radiographic stability before surgery nor residual microscopic disease afterward proves treatment failure. The relevant MATTERHORN benefit was reduced progression, recurrence, or death across the full intention-to-treat population—not pCR alone.

He resumes therapy seven weeks after surgery. Oxaliplatin is reduced when persistent grade 2 neuropathy begins to interfere with fine motor tasks; the remaining FLOT components are adjusted according to recovery and tolerance. Durvalumab continues every four weeks, followed by ten planned monotherapy doses. Adequate D2 surgery and an R0 margin provide no routine indication for adjuvant radiation. Surveillance emphasizes symptoms, examination, nutrition, laboratory deficiencies, and guideline-concordant imaging.

Nuance: If postoperative recovery prevented further FLOT, continuation of durvalumab alone would be consistent with protocol flexibility, but the team should document why chemotherapy was stopped. If he developed immune myocarditis or severe pneumonitis, preserving life and surgical recovery would take precedence over completing durvalumab.

Audience Poll: Would your recommendation change if this patient were MSI-high, had baseline grade 2 neuropathy, had positive peritoneal cytology, or required chronic prednisone for active inflammatory lung disease?

Tonight on Shift

Confirm the curative premise: Verify pathology, CT staging, resectability, performance status, and whether staging laparoscopy with cytology is required before treatment.
Name the regimen precisely: Four every-two-week FLOT administrations plus two every-four-week durvalumab doses before surgery; repeat postoperatively, then ten additional durvalumab doses.
Protect the operation: Restage, correct cytopenias and malnutrition, identify immune toxicities, and target surgery four to eight weeks after neoadjuvant treatment when clinically ready.
Treat fever and dangerous symptoms first: Assume neutropenic sepsis with fever; urgently evaluate dyspnea, chest pain, weakness, hypotension, or severe diarrhea for infectious, thrombotic, cardiac, endocrine, and immune causes.
Interpret MATTERHORN accurately: EFS, pCR, and final OS improved; R0 resection and postoperative mortality did not. PD-L1 TAP was a stratification factor, not an FDA eligibility requirement.
Use precision without overreach: Obtain MSI/MMR and HER2 testing, understand PD-L1 methodology, consider emerging targets and ctDNA trials, and do not omit chemotherapy or surgery on immature biomarker evidence.

Read this seminar as Markdown · All seminars · Lecture library · Question bank