Abstract
T-cell immunoreceptor with Ig and ITIM domains (TIGIT) blockade emerged as one of the most anticipated next-generation checkpoint strategies of the early 2020s, yet the 2021–2026 period has been defined by a succession of Phase 3 failures in non-small cell lung cancer (NSCLC) and small-cell lung cancer (SCLC), alongside Phase 3 failures for tiragolumab, and program discontinuations for vibostolimab, ociperlimab, and belrestotug. Amid this setback landscape, the Fc-silent anti-TIGIT antibody domvanalimab and the bispecific PD-1 × TIGIT construct rilvegostomig (AZD-2936) continue to generate encouraging, if unconfirmed, data in biomarker-selected and gastrointestinal (GI) cancer populations. This review synthesizes Phase 2 and Phase 3 clinical evidence through July 2026, identifies mechanistic and trial-design lessons, and presents a framework for understanding TIGIT not as a definitively failed checkpoint target but as a context-dependent, incompletely optimized therapeutic strategy. 89
Background: TIGIT Biology and Therapeutic Rationale
TIGIT is an inhibitory checkpoint receptor expressed on CD4+ and CD8+ T cells, regulatory T cells (Tregs), and natural killer (NK) cells. Its principal ligands in humans are CD155 (poliovirus receptor, PVR) and CD112 (nectin-2), which are frequently overexpressed on tumor cells and immunosuppressive stromal cells. Upon ligand engagement, TIGIT delivers inhibitory signals that suppress cytotoxic T-cell activation and NK cell–mediated tumor killing, while simultaneously promoting Treg-mediated immunosuppression. It competes for shared ligands with the co-stimulatory receptor CD226 (DNAM-1), and a high TIGIT/CD226 ratio on tumor-infiltrating lymphocytes (TILs) is considered a hallmark of T-cell exhaustion. The dual rationale for combining TIGIT blockade with PD-1/PD-L1 inhibition rests on complementary reinvigoration of exhausted T cells and NK cells, supported by robust preclinical synergy in murine tumor models and early Phase 2 human signals. 89
2021–2026 TIGIT Clinical Trial Landscape
Table 1. Major TIGIT Clinical Trial Readouts, 2021–2026
| Agent | Trial / Sponsor | Indication / Setting | Phase / Design | Regimen | N | Key Efficacy Findings | Safety | Development Status |
|---|---|---|---|---|---|---|---|---|
| Tiragolumab | CITYSCAPE / Roche | 1L metastatic NSCLC, PD-L1+ | Phase II, R, blinded, PC | Tiragolumab + atezolizumab vs placebo + atezolizumab | — | ITT ORR 38.8% vs 20.6%; PFS HR 0.62; OS HR 0.69. PD-L1-high: ORR 69% vs 24%, PFS HR 0.29 | G3–4 TRAEs 22.4% vs 25%; no new signals | Positive Phase II; drove Phase III |
| Tiragolumab | SKYSCRAPER-01 / Roche | 1L PD-L1-high advanced NSCLC | Phase III, R, DB, PC | Tiragolumab + atezolizumab vs placebo + atezolizumab | 534 | PFS co-primary not met; final OS HR 0.81 (0.63–1.03); primary OS endpoint not met | Consistent profile; no new signals | Negative (primary OS endpoint failed) 34 |
| Tiragolumab | SKYSCRAPER-06 / Roche | 1L advanced non-squamous NSCLC | Phase II/III, R, DB, PC | Tiragolumab + atezolizumab + chemo vs pembrolizumab + chemo | 542 | PFS HR 1.27 (1.02–1.57); OS HR 1.33 (1.02–1.73); experimental arm inferior | Similar between arms; no new signals | Discontinued (futility / inferiority) 11 |
| Tiragolumab | SKYSCRAPER-02 / Roche | 1L extensive-stage SCLC | Phase III, R, DB, PC | Tiragolumab + atezolizumab + carboplatin/etoposide vs placebo + atezolizumab + carboplatin/etoposide | — | No PFS or OS benefit in PAS or FAS; OS unlikely to reach significance | G3/4 TRAEs 52.3% vs 55.7%; no new signals | Negative 1 |
| Vibostolimab | KeyVibe-003 / Merck | 1L PD-L1+ metastatic NSCLC | Phase III, R, DB | Vibostolimab/pembrolizumab FDC vs pembrolizumab | 1,264 | Met pre-specified OS futility criteria (TPS ≥50%) | Higher irAEs with dual checkpoint; no new signals | Discontinued (OS futility) 6 |
| Vibostolimab | KeyVibe-007 / Merck | 1L treatment-naïve metastatic NSCLC | Phase III, R, DB | Vibostolimab/pembrolizumab FDC + chemo vs pembrolizumab + chemo | 739 | Met pre-specified OS futility criteria (TPS ≥1%) | Higher irAEs with dual checkpoint; no new signals | Discontinued (OS futility) 6 |
| Vibostolimab | KeyVibe-006 / Merck | Stage III NSCLC | Phase III, R, OL | Vibostolimab/pembrolizumab + CCRT + maintenance vs durvalumab | ~580 | Not safety-driven; program-level discontinuation | Not safety-driven | Discontinued 6 |
| Domvanalimab | ARC-7 / Arcus/Gilead | 1L metastatic NSCLC, PD-L1 TPS ≥50% | Phase II, R, OL, 3-arm | Domvanalimab + zimberelimab ± etrumadenant vs zimberelimab | 150 | ORR 41% vs 27% (doublet vs mono); DOR NR vs 13.2 mo; PFS 33% risk reduction for doublet | G≥3 TEAEs 47% vs 58%; consistent with known profiles | Positive Phase II 5 |
| Domvanalimab | ARC-10 Part 1 / Arcus/Gilead/Taiho | 1L LA/metastatic NSCLC, TPS ≥50% | Phase III, R, OL | Domvanalimab + zimberelimab vs zimberelimab vs platinum chemo | ~98 treated | OS HR 0.64 (doublet vs zimberelimab); 12-mo OS 68% vs 57% vs 50% | G≥3 TRAEs 21.1% vs 15.0% vs 47.1% | Positive OS signal 1 |
| Domvanalimab | EDGE-Gastric (Arm A1) / Arcus | Gastric/GEJ/esophageal adenocarcinoma | Phase II | Domvanalimab + zimberelimab + FOLFOX | 41 | Median OS 26.7 mo; 24-mo OS rate 50.2%; PFS 12.9 mo; ORR 59% | Immune-mediated TEAEs 22%; no unexpected signals | Positive; advancing to Phase 3 STAR-221 10 |
| Ociperlimab | NCT04952597 / BeiGene | Untreated limited-stage SCLC + cCRT | Phase II, R, OL, 3-arm | Ociperlimab + tislelizumab + cCRT vs tislelizumab + cCRT vs cCRT | — | PFS 12.6 vs 13.2 vs 9.5 mo (Arm A vs C HR 0.84); no clear added TIGIT benefit | G≥3 TRAEs 73.2% vs 78.6% vs 65.1% | Mixed; negative for incremental TIGIT contribution 1 |
| Ociperlimab | BeiGene | NSCLC (advanced) | Phase III | Anti-TIGIT antibody details not disclosed | — | Disappointing efficacy signals | — | Discontinued 12 |
| Belrestotug | GALAXIES Lung-201 / GALAXIES H&N-202 / GSK/iTeos | Lung cancer; head and neck cancer | Phase II | Anti-TIGIT antibody details not disclosed | — | Did not meet pre-specified efficacy criteria | — | Discontinued (May 2025) 13 |
| Etigilimab | ACTIVATE / Mereo BioPharma | Select advanced solid tumors (CPI-naive) | Phase Ib/II, OL | Etigilimab + nivolumab | — | Cervical: ORR 43%, DCR 71%; uveal melanoma: ORR 17%; additional signals in ovarian, endometrial, liposarcoma | Manageable; no major new signals | Exploratory signals; hypothesis-generating 1 |
| Rilvegostomig | ARTEMIDE-01 / AstraZeneca | NSCLC, gastric, GEJ, biliary tract | Phase I/II; 10 Phase III ongoing | Bispecific PD-1 × TIGIT antibody | — | Early promise; Phase 3 results pending | Acceptable safety profile | Ongoing Phase 3 enrollment 14 |
Abbreviations: 1L = first-line; R = randomized; DB = double-blind; PC = placebo-controlled; OL = open-label; FDC = fixed-dose combination; cCRT = concurrent chemoradiotherapy; PAS = primary analysis set; FAS = full analysis set; TPS = tumor proportion score; CCRT = concurrent chemoradiotherapy; GEJ = gastroesophageal junction; irAE = immune-related adverse event; TRAE = treatment-related adverse event.
Narrative Analysis: Clinical Setbacks and Mixed Signals
Early Promise (2021–2022)
The TIGIT era opened with genuine Phase 2 optimism. The CITYSCAPE trial established a compelling proof-of-concept: tiragolumab added to atezolizumab improved objective response rate (ORR) and progression-free survival (PFS) in chemotherapy-naive PD-L1-positive metastatic NSCLC, with a striking PD-L1–high subgroup (tumor proportion score [TPS] ≥50%) showing an ORR of 69% versus 24.1% and a PFS hazard ratio (HR) of 0.29. Similarly, the ARC-7 Phase 2 study reported that domvanalimab combined with the anti-PD-1 antibody zimberelimab reduced the risk of progression or death by 33% versus zimberelimab monotherapy in PD-L1-high first-line NSCLC, with durable responses. 51 These signals energized an already competitive field, with 16 distinct TIGIT-directed assets or combination regimens entering development globally, spanning conventional Fc-competent anti-TIGIT monoclonal antibodies (mAbs), Fc-silent mAbs, and emerging PD-1 × TIGIT bispecific constructs. 2
Phase 3 Failures and Program Discontinuations (2023–2025)
The transition to Phase 3 delivered a series of major setbacks. Roche's SKYSCRAPER-01, which enrolled 534 patients with PD-L1-high advanced NSCLC, failed to meet its co-primary PFS endpoint and subsequently failed the primary overall survival (OS) endpoint at final analysis (HR 0.81; 95% CI 0.63–1.03), with no new safety signals. 34 SKYSCRAPER-02 in extensive-stage SCLC was equally discouraging, showing no incremental PFS or OS benefit for tiragolumab added to atezolizumab plus carboplatin/etoposide. 1 Most damaging was SKYSCRAPER-06, in which the tiragolumab arm was not merely non-superior but actively inferior to pembrolizumab plus chemotherapy, with a PFS HR of 1.27 and an OS HR of 1.33, leading to study discontinuation. 11
Merck's KeyVibe program — comprising three Phase 3 trials of vibostolimab (MK-7684) co-formulated with pembrolizumab — was voluntarily discontinued in December 2024. KeyVibe-003 (n=1,264) and KeyVibe-007 (n=739) both met pre-specified OS futility criteria based on independent Data Monitoring Committee assessments in first-line PD-L1-positive and treatment-naive metastatic NSCLC, respectively; KeyVibe-006 in stage III NSCLC was discontinued as part of the same program-level decision. 6 The GSK/iTeos collaboration on belrestotug (EOS-448/GSK-4428859) similarly ended in May 2025 following interim analyses from the Phase 2 GALAXIES Lung-201 and GALAXIES H&N-202 studies that failed to meet pre-established efficacy criteria. 13 BeiGene discontinued its ociperlimab NSCLC program following disappointing efficacy signals. 12
Where Signals Persist
Against this broadly negative backdrop, domvanalimab and rilvegostomig represent the clearest areas of residual promise. In the Phase 2 EDGE-Gastric study (Arm A1, n=41), domvanalimab combined with zimberelimab and FOLFOX chemotherapy in patients with gastric, gastroesophageal junction (GEJ), or esophageal adenocarcinoma produced a median OS of 26.7 months, a 24-month OS rate of 50.2%, a median PFS of 12.9 months, and a confirmed ORR of 59% per RECIST v1.1, with favorable efficacy signals observed across all PD-L1 subgroups. 10 These results supported initiation of the Phase 3 STAR-221 trial comparing domvanalimab/zimberelimab ± chemotherapy versus nivolumab ± chemotherapy in first-line GEJ adenocarcinoma. Rilvegostomig, a bispecific antibody simultaneously targeting PD-1 and TIGIT with reduced Fc-effector functions, is currently enrolled in 10 Phase 3 studies across NSCLC, biliary tract cancer, gastric, and GEJ adenocarcinoma, with mature efficacy readouts pending. 14 In NSCLC, the Fc-silent domvanalimab plus zimberelimab doublet achieved a 36% reduction in the risk of death versus zimberelimab monotherapy in the ARC-10 Part 1 analysis (HR 0.64; 12-month OS 68% vs. 57%). 1
The etigilimab (OMP-313M32) plus nivolumab program (ACTIVATE) generated exploratory Phase 1b/2 signals in cervical cancer (ORR 43%, disease control rate 71%) and uveal melanoma (ORR 17%), with isolated responses in ovarian, endometrial, and liposarcoma cohorts, but these data are not yet confirmatory given small cohort sizes. 1
Biomarker and Trial-Design Lessons
Table 2. Key Biomarker and Trial-Design Lessons from TIGIT Trials, 2021–2026
| Theme | Evidence from Trial Data | Practical Implication |
|---|---|---|
| PD-L1 enrichment is necessary but insufficient | PD-L1 TPS ≥50% enrichment in SKYSCRAPER-01 and KeyVibe-003 did not translate to OS benefit; futility signals emerged across all subgroups | Richer biomarker frameworks needed: TIGIT/CD155 axis expression, TIL density, T-cell exhaustion markers, tumor mutational burden |
| Phase 2 signals can overestimate Phase 3 benefit | CITYSCAPE was strongly positive; SKYSCRAPER-01 subsequently failed both co-primary endpoints | Early Phase 2 success should not be overinterpreted; confirmatory design and power must account for control-arm performance |
| Comparator selection is critical | SKYSCRAPER-06 compared tiragolumab/atezolizumab/chemo to pembrolizumab/chemo; experimental arm was inferior | TIGIT regimens must be benchmarked against contemporaneous, validated standards—not historical data |
| Fc-effector function affects efficacy-safety balance | IgG1 anti-TIGIT (tiragolumab, vibostolimab) failed; Fc-silent domvanalimab and bispecific rilvegostomig show more promising signals | Fc engineering matters; Fc-silent formats may avoid unintended depletion of exhausted T cells and NK cells 9 |
| Tumor-type heterogeneity is substantial | NSCLC: largely negative in Phase 3; GI cancers: encouraging Phase 2 OS with domvanalimab; SCLC: no convincing TIGIT benefit | TIGIT should not be treated as a pan-tumor strategy; biology is indication-specific |
| Incremental TIGIT benefit on top of PD-1/PD-L1 is not guaranteed | SKYSCRAPER-02 and ociperlimab limited-stage SCLC study showed no incremental TIGIT contribution | Combination rationale must be empirically demonstrated in each disease context, not assumed from shared biology |
| Pre-specified futility criteria enable timely discontinuation | KeyVibe-003 and KeyVibe-007 futility analyses allowed early termination, limiting patient exposure and resource expenditure | Adaptive designs with pre-planned futility stopping rules should be incorporated routinely in Phase 3 TIGIT trials 6 |
| CD155 and TIGIT axis remain underexplored biomarkers | Anti-PD-1 response correlates with CD155 but not TIGIT expression; PVR expression may enrich for responders in etigilimab program | Prospective biomarker plans should go beyond PD-L1 and assess CD155/CD112 tumor expression and dynamic immune profiling 8 |
Future Development Strategy and Conclusion
The mechanistic case for TIGIT blockade remains intact despite the clinical setbacks catalogued above. Several non-mutually exclusive explanations have been proposed for the Phase 3 failures of Fc-competent IgG1 anti-TIGIT antibodies: (1) unintended Fc-mediated depletion of exhausted TILs and NK cells that were intended to be reinvigorated; (2) redundant inhibitory checkpoint pathways (LAG-3, TIM-3, BTLA) that remain operative after TIGIT blockade; (3) adaptive tumor resistance through upregulation of alternative immunosuppressive mechanisms; and (4) insufficient enrichment for patients with "hot" tumor microenvironments where TIGIT biology is most operative. 89
The clinical survival of domvanalimab and rilvegostomig in the landscape offers a critical hypothesis: pure TIGIT blockade — without Fc-mediated effector cell destruction — may be the operative mechanism in humans, and this can be achieved either through Fc-silent mAb engineering or through bispecific co-blockade of PD-1 and TIGIT on the same immune cell. 9 The divergent performance of NSCLC (broadly futile) versus gastroesophageal cancer (encouraging Phase 2 OS) further argues that TIGIT biology is tumor-context dependent, likely shaped by baseline immune infiltration, CD155 ligand expression, and the immunogenic potential of each disease. 10
For clinical researchers and translational medicine teams, five practical priorities emerge from the 2021–2026 data:
First, biomarker enrichment must extend beyond PD-L1. Composite immune profiling incorporating CD155 tumor expression, TIL density, TIGIT expression on immune cell subsets, and CD226/TIGIT balance should be integrated prospectively into Phase 2 trial design before Phase 3 commitment. Second, Fc-silent formats and bispecific constructs should be prioritized over Fc-competent IgG1 scaffolds, given the accumulating mechanistic and clinical evidence. Third, control-arm performance must be contemporary: the SKYSCRAPER-06 lesson that a TIGIT arm can be inferior to pembrolizumab plus chemotherapy underscores the need to benchmark against the actual standard of care. Fourth, adaptive trial designs with pre-specified futility criteria should be routine, as demonstrated by the efficient termination of the KeyVibe program. Fifth, tumor-type selection should be hypothesis-driven, focusing early development on indications where TIGIT axis biology is mechanistically supported and early-phase signals are clear.
The broader portfolio landscape through July 2026 identifies 16 distinct TIGIT-directed assets spanning preclinical through Phase 3, with China increasingly prominent both as a clinical trial geography for multinational programs and as a source of early-stage innovation in Fc-silent, bispecific, and multispecific TIGIT-directed constructs (e.g., rilvegostomig, IBI-939, HLX-301, GBD-209). 2
In summary, the 2021–2026 TIGIT experience should not be read as a verdict of target failure, but as a lesson in the complexity of translating mechanistic elegance into clinical efficacy. TIGIT is best characterized as a context-dependent, incompletely optimized target, requiring refined patient selection, Fc-engineered or bispecific antibody formats, rational combination frameworks, and rigorously powered, biomarker-enriched trial designs to achieve durable clinical benefit. 89 The imminent Phase 3 readouts of rilvegostomig and the STAR-221 domvanalimab program in GEJ adenocarcinoma will be decisive in determining whether TIGIT blockade can yet fulfill its immunological promise in precision oncology.