Colorectal & lung surface targets, prioritised for ADC, TCE and CAR‑T, split by maturity.

INVEST. FGFR2 is a clinically and commercially validated target. Although the prevalence in NSCLC is low, it represents a clear, de-risked opportunity for a precision medicine approach. The path for development is well-trodden by other kinase inhibitors, and the potential for a best-in-class small molecule or a novel ADC provides a strong strategic rationale.

INVEST. CCR8 is a highly promising, next-generation immuno-oncology target with a very strong biological rationale, high tractability for depleting ADCs, and a clear path to market as a combination therapy with checkpoint inhibitors in NSCLC. The timing is optimal to enter as a fast-follower with a best-in-class asset. The Treg-depletion rationale is sound, but this is already an active space, not white-space: BMS-986340 and the ITeos/BMS CCR8 depleting antibody carry Phase 1/2 data with on-target depletion. Given the INVEST verdict this is a fast-follower position; systemic Treg-depletion autoimmunity is the first-order risk, and CAR‑T faces fratricide from effector-cell CCR8 expression.

WATCH. LAIR1 is a promising, tractable, next-generation immuno-oncology checkpoint with a strong rationale in the fibrotic NSCLC TME. However, the NSCLC-specific preclinical validation is still emerging, and on-target toxicity is a key risk. This would become INVEST-grade with in vivo proof-of-concept in NSCLC models and clearer safety data.

WATCH because CDCP1 is a high-tractability surface nomination aligned with ADC/TCE, but SCLC-specific prevalence and dependency are not established in retrieved 2024-2026 evidence. The most decision-changing data are tumour/normal expression atlases and functional knockdown/targeted killing in SCLC models.

WATCH because NGFR/p75NTR is a druggable surface nomination with a plausible SCLC lineage/plasticity thesis, but SCLC-specific prevalence, dependency, and safety window are not yet established. The most decision-changing data are SCLC tumour/normal expression atlases and functional depletion studies in SCLC models.

WATCH, leaning toward INVEST. LAIR1 is a classic, tractable IO surface target with a strong mechanistic rationale analogous to PD-1. The main gate is the lack of direct SCLC evidence, but its potential as a next-generation checkpoint makes it a high-priority for validation.

INVEST because this nomination is focus-modality compatible and could represent antigen white-space; the key next data are tumour-selective expression and internalisation.

WATCH: strong antigen-class tractability plus recent in vivo CAR‑T proof-of-concept, but NSCLC-specific expression, efficacy, and safety data are insufficient. Upgrade requires NSCLC IHC/omics prevalence, internalisation data for ADC, and in vivo NSCLC models.

WATCH because FOLR3 is a surface receptor nomination with plausible myeloid/TME targeting potential and payload-delivery tractability, but SCLC-specific compartment expression and causal relevance are unproven. The most decision-changing data are SCLC single-cell/spatial maps and intervention studies showing improved antitumor immunity.

WATCH because ITGA5 is tractable and mechanistically plausible for metastasis biology, but CRC-specific causal evidence and tumour-selective expression are not established in this run. A near-term de-risking plan is scRNA/proteomics tumour-selectivity plus in vivo metastasis models with pathway engagement.

INVEST because DLL3 has the deepest SCLC-specific clinical validation, clear modality fit (especially TCE), and a maturing biomarker strategy via CTC DLL3 quantitation. The main swing factors are resistance management and competitive differentiation rather than biology credibility. SCLC second-line is a single, contested slot: DLL3 T-cell engagers now compete head-to-head with SEZ6 and B7-H3 ADCs for the same relapsed patients. Only one is likely to become standard of care, which directly discounts the timing and commercial scores. Target validity is settled; the contest is competitive, not scientific.

INVEST because HER2 is a validated, highly tractable surface receptor with strong lung-cancer clinical efficacy for ADCs and mature biomarker pathways. In SCLC, the key uncertainty is prevalence and expression biology; prioritise work on HER2-amplified/transformed subsets and differentiated ADC designs to compete. HER2 is the best-known ADC target, but that franchise is in breast; HER2-expressing lung is a small subset. Differentiation now comes from bispecific ADCs, not another naked ADC. The HER2 TKIs (zongertinib and peers) are a different mechanism used in combination, not direct ADC competition.

INVEST. TROP2 is a clinically and commercially validated surface antigen in NSCLC with strong efficacy data for ADCs in late-line settings. The high expression across subtypes, accelerating momentum, and emerging potential for CAR‑T to address drug resistance make it a top-tier target for a portfolio focused on ADC/CAR‑T modalities. The target is settled; the contest is at the asset layer. The ASCO 2026 readout for Kelun/Merck's sac-TMT (SKB264/MK-2870) positions it as current best-in-class on payload, linker and tolerability, ahead of Dato-DXd (Daiichi/AZ) and Trodelvy (Gilead). The play is a superior ADC, not de novo targeting. Note the high CAR‑T score rests on a 2025 Cancer Discovery report of TROP2 CAR‑T eradicating drug-tolerant persister cells specifically in EGFR-mutant NSCLC on osimertinib, a single-indication and largely preclinical rationale, not pan-NSCLC.

Nectin-4 has FDA-approved ADC precedent and validated NSCLC overexpression. WATCH pending NSCLC-specific clinical trial data for ADC efficacy. Phase I/II data in NSCLC would upgrade to INVEST. Nectin-4 ADCs are in Phase 1 in lung, but the target has not yet delivered outside urothelial cancer; treat NSCLC as unproven.

INVEST because SEZ6 has strong recent primary clinical efficacy in relapsed/refractory SCLC and is highly tractable as an ADC antigen with a defined RP2D. Key data that could change the view are durability, safety in larger cohorts, and confirmation of benefit versus standard second-line options. ABBV-706 posted ORR 52% (65/124) in R/R SCLC at RP2D 1.8 mg/kg Q3W, but with Grade 3+ TRAEs in 61% (cytopenias, ILD/pneumonitis) as the dose-limiting constraint, and a next-generation biparatopic SEZ6 ADC (HWK-206, superior preclinical internalisation) is emerging. Normal-tissue CNS expression is documented, not unknown (SEZ6 is a neuronal synaptic protein; ABBV-706 is in CNS-tumour cohorts), so neurotoxicity is a real constraint for engager or CAR‑T formats; the relapsed-SCLC slot is contested with DLL3 and B7-H3.

HER3 is among the strongest clinically validated NSCLC targets in this list due to consistent patritumab deruxtecan efficacy in EGFR-mutant post-resistance disease and ongoing phase 3 development. New data that would change this verdict would be clear evidence of superior differentiation (safety/CNS/combination efficacy) for next-gen HER3 assets. HER3 underwhelmed as a monotherapy target historically, but the picture has changed. HER3-DXd (patritumab deruxtecan) posted ORR 29.8%, mPFS 5.5mo and mOS 11.9mo in HERTHENA-Lung01 (Phase 2), and its pivotal HERTHENA-Lung02 (Phase 3, Dec 2025) reported a statistically significant PFS benefit vs standard of care in EGFR-mutant, post-TKI NSCLC. The older 'disappointed' framing is outdated; treat the target as Phase 3-validated, and the science score as now well-supported rather than generous.

CDH17 has unusually strong convergence across mechanism, multiple in vivo therapeutic modalities, and active clinical programmes in GI cancers including CRC. The main swing factor is defining a safe therapeutic window in humans; convincing early clinical ORR with acceptable GI toxicity would upgrade to sustained priority. Near-universal CRC expression with a diverse pipeline (ADCs, CAR‑T, bispecifics). A 2025 Science Translational Medicine study showed systemic CDH17 CAR‑T suppresses CRC liver xenografts, with a mechanistic safety basis: healthy colon CDH17 is restricted to lateral junctions (inaccessible to systemic CAR‑T) while tumour cells expose it surface-wide. Counter-intuitively, locoregional delivery produced harsher cytokine release than systemic in the humanised model. The active Chimeric Therapeutics Phase 1 is neuroendocrine, not CRC-specific, so CRC efficacy proof-of-concept remains the gap (see also BI 905711; 7MW4911 ADC, 71-99% tumour-growth inhibition preclinically).

B7-H3 has strong 2024–2026 translational depth with a clear ADC modality path and demonstrated NSCLC clinical activity. The key value creation lever is biomarker and safety optimisation; a validated predictive biomarker or superior therapeutic index would further strengthen the case. The highest-signal B7-H3 programme, ifinatamab deruxtecan (I-DXd), reported ORR 48.2% in previously treated ES-SCLC and is now in Phase 3 (IDeate-Lung02, ~540 patients), validating the target and intensifying the relapsed-SCLC competition, and absent from the auto-retrieved evidence here. YL201 (ORR 28.6% in lung adenocarcinoma, ILD 1.3%) does not correlate expression with response, so patient selection is unsolved. The WATCH verdict predates the Phase 3 context and reads conservative against it.

CEACAM5 is among the most clinically validated CRC tumour antigens, with modern ADC data and multiple immune-engaging strategies. The value-creation lever is not proving relevance, but achieving a superior therapeutic index and durability versus competing CEACAM5 constructs; definitive early signals in MSS metastatic CRC would further strengthen conviction. The most advanced CEACAM5 T-cell engager, cibisatamab (RG6123), showed only modest Phase 1 CRC efficacy with significant cytokine-release toxicity requiring careful dose escalation; the M9140 ADC is in Phase 1 CRC. The TCE-fit reads optimistic against that Phase 1 reality, and antigen shedding remains a real risk.

Watch (not INVEST) because EGFR is actionable in SCLC only for a low-frequency, biomarker-defined subset, despite strong tractability and recent cohort-level outcome signals. Upgrade would require prospective SCLC trials confirming benefit and defining optimal combination/sequence. Relevant in both lung (mutant, TKI-driven) and colorectal (RAS-wild-type, cetuximab/panitumumab). Its many approved agents are TKIs and antibodies, not ADCs, so there is genuine whitespace for an EGFR ADC; TKIs are also limited by universal acquired resistance and lack a conjugate's bystander effect.

GUCY2C is one of the most CRC-specific immunotherapy antigens with credible clinical-stage momentum and a newly clarified, actionable resistance mechanism. The most value-creating next data would be prospective clinical demonstration that combining/engineering around IFNγ-JAK stress signalling improves durability without unacceptable intestinal toxicity. First human clinical data land squarely behind the INVEST call and post-date the auto-retrieved evidence: a Phase 1 GUCY2C CAR‑T in metastatic CRC (JCO, Jun 2026) showed ORR 26.3% overall and 40% at DL3 (mPFS 7.0mo), rising to 50% ORR and 9.0mo mPFS in medium-to-high GUCY2C expressors. Grade 3 diarrhoea was the main DLT (55%); CRS was manageable (1 patient Grade 3). This is the strongest clinical support in the shortlist for a CAR‑T-lead CRC target.

GPA33 has strong biological and technical credentials, but translation risk is dominated by intratumour heterogeneity and invasive-margin antigen loss. This is worth WATCH with a clear plan to induce/expand antigen expression (e.g., WNT inhibition) and to demonstrate tumour coverage and safety in early trials.

MSLN is a credible subset target in CRC with recent strong preclinical CAR-NK and regional CAR‑T data, but development is gated by well-documented on-target/off-tumour toxicity risk in other indications. Upgrading requires a clinically credible safety-optimised construct (logic gating, CAR-NK) and a robust MSLN-high selection threshold in CRC. Validated across CAR‑T and ADC programmes, but MSLN carries a specific, documented safety precedent: an early autologous MSLN CAR‑T death from on-target pulmonary/anaphylactic toxicity (Penn, 2016). Treat off-tumour pulmonary and pleural expression as a first-order risk for any cell-depleting format, not a generic caveat.

MET is the strongest INVEST target in this cohort. Recent SCLC-specific data demonstrate therapeutic benefit of MET inhibition in combination with CIT, leveraging an established and well-tolerated drug class. The key next step is a clinical trial of MET inhibitor + CIT in ES-SCLC with HGF/MET biomarker stratification. Value is likely in combination, e.g. the AbbVie MET ADC (telisotuzumab vedotin) plus osimertinib, rather than monotherapy.

WATCH: antigen is attractive and modality-diverse with recent mechanistic de-risking for soluble CA125 concerns, but NSCLC-specific prevalence and clinical efficacy remain to be established. Decision-changing data: NSCLC IHC prevalence/thresholding, and early clinical activity in MUC16-high NSCLC.