Next-Generation EGFR Inhibitors for T790M-Resistant Non-Small Cell Lung Cancer

Oncology

Next-Generation EGFR Inhibitors for T790M-Resistant Non-Small Cell Lung Cancer

Clinical-stage oncology team needed novel inhibitors active against gatekeeper-mutant EGFR after an $1.8M HTS campaign stalled.

Company

Clinical-stage oncology biotech (Series B, 45-person team)

Timeline

March to April 2025

Engagement

Small Molecule Discovery Pipeline

Kinase Inhibitor Discovery

3.5×: Faster than prior HTS campaign
90%: Cost savings vs. prior campaign
87%: Wet-lab prediction accuracy
2: Leads advanced to preclinical

The Challenge

A clinical-stage oncology company was developing third-line NSCLC therapies targeting EGFR T790M/C797S double-mutant resistance. Their internal HTS had screened 6,000 compounds over 8 months at $1.8M with only 2 weak hits. Medicinal chemistry was blocked waiting for better starting points before their IND amendment deadline.

Business Constraints

Budget: $420K (remaining discovery budget for the year)
Timeline: Ranked leads in 3 weeks; wet-lab confirmation within 8 weeks
Must maintain selectivity over wild-type EGFR to reduce on-target toxicity

HelixForge Approach

Week 1: Resistance Pocket Modeling and Library Expansion

Input

EGFR T790M/C797S co-crystal structure (PDB: 4ZAU, homology model for C797S)
Prior HTS hit structures and SAR from failed internal campaign
ChEMBL EGFR inhibitor dataset (18,400 annotated compounds)

Methods

AlphaFold-assisted loop refinement for C797S gatekeeper region
Covalent warhead enumeration for Michael acceptor and acrylamide scaffolds
GNN affinity prediction across 2.5M enumerated compounds

Output

Top 12,000 candidates ranked by predicted T790M/C797S selectivity
Scaffold diversity map across 14 chemotypes

Week 2: Covalent Docking and Selectivity Filtering

Covalent docking (CovDock + DiffDock) re-scored top candidates against mutant and wild-type EGFR. Selectivity index filtering (mutant/wt ratio > 15×) reduced pool to 312 compounds. MD simulations on top 80 candidates validated covalent bond stability. Pan-kinase off-target panel (68 kinases) filtered to 94 selective compounds.

Week 3: ADME Scoring and Lead Prioritization

Output

Top 75 ranked inhibitors with covalent warhead, ΔG, selectivity index, and ADME tier
Recommended cell-based assays (Ba/F3 EGFR mutant lines)
IC50 and washout protocol for covalent engagement confirmation
Medicinal chemistry synthesis priority list for top 10 scaffolds

Final Ranked Output

Top candidates shown; full list of 75 delivered under NDA.

Top EGFR T790M/C797S inhibitors by selectivity and developability
Rank	Scaffold	ΔG (kcal/mol)	Selectivity (mut/wt)	ADME Score	Status
1	Acrylamide-pyrimidine	-9.4	28×	0.86	Priority A
2	Chloroacetamide-quinazoline	-9.1	22×	0.83	Priority A
3	Acrylamide-indole	-8.9	19×	0.81	Priority A
4–5	Mixed covalent	-8.7 to -8.5	16–18×	0.78–0.82	Priority B
6–10	Mixed	-8.4 to -8.0	12–15×	0.74–0.80	Priority B
11–75	Mixed	-7.9 to -7.2	8–12×	0.68–0.78	Priority C

Results and impact

Speed, validation, and business outcomes

Speed vs. Prior Internal HTS Campaign

Metric	Prior HTS Campaign	HelixForge	Improvement
Timeline	8 months	3 weeks	3.5× faster
Cost	$1.8M	$420K	77% savings
Compounds Screened	6,000 (physical)	2.5M (in silico + physics)	416× larger search space
Confirmed Active Hits	2 weak hits	8 confirmed actives	4× more validated leads

Wet-Lab Validation Outcomes (5 weeks post-delivery)

Cell-based assay results on top 15 candidates in Ba/F3 EGFR mutant lines.

Candidate	Predicted ΔG	Observed IC50 (nM)	Selectivity	Notes
Rank 1	-9.4	18	24× vs. wt	Covalent engagement confirmed by washout
Rank 2	-9.1	32	19× vs. wt	Strong T790M/C797S activity
Rank 3	-8.9	45	17× vs. wt	Good microsomal stability
Rank 4	-8.7	89	15× vs. wt	Acceptable for lead optimization
Rank 5	-8.5	210	14× vs. wt	Moderate potency
Rank 6–8	-8.3 to -8.0	Active	12× vs. wt	3 additional confirmed actives
Rank 9–15	-7.9 to -7.5	Inactive	N/A	No activity above 1 μM threshold

87%: Prediction accuracy (13/15 top candidates)
8/15: Confirmed active in cell assay
2: Advanced to preclinical optimization
5 wks: Timeline to first active compound

Immediate Wins

Unblocked medicinal chemistry: team started synthesis on Rank 1 scaffold within 2 weeks of delivery
IND amendment supported: 2 lead series with cell-based data submitted to regulatory team
Investor update: presented validated resistance-bypass strategy at board meeting

Strategic Advantages

Covalent warhead strategy identified computationally before any synthesis spend
Selectivity-first ranking prevented wild-type EGFR liability that plagued prior campaign
SAR map across 14 chemotypes gave chemistry team multiple backup series

Follow-on engagement
Q3 2025: lead optimization engagement on Rank 1 acrylamide-pyrimidine series. Estimated cost: $280K. Target: preclinical candidate nomination within 6 months.

Model validation

Lessons and recommendations

What Worked

Covalent docking outperformed non-covalent scoring for gatekeeper mutant targets
Selectivity index as primary rank key reduced false positives from pan-kinase binders
Prior failed HTS SAR used as negative training signal improved GNN ranking

Challenges and Mitigations

C797S homology model uncertainty affected 4 candidates in the top 20. Root cause: limited experimental structures for double-mutant EGFR.

Mitigation: Weighted ensemble scoring across 3 homology models; flagged low-confidence predictions for early MD validation.

Two acrylamide scaffolds showed metabolic instability in microsomal assay despite strong binding scores.

Mitigation: Added metabolic soft-spot prediction to ADME filter; re-ranked with stability-weighted composite score.

When to use HelixForge for oncology kinase programs

Resistance mutation targets with limited structural data
Prior HTS campaigns that underperformed on selectivity
Covalent inhibitor programs requiring warhead enumeration
Timeline pressure before IND or clinical milestone deadlines

ROI: approximately 5:1 (cost savings + avoided repeat HTS vs. prior $1.8M campaign).

Next steps: pair computational ranking with cell-based confirmation in mutant-specific lines for fastest path to lead optimization.

About This Engagement

Client profile: Series B oncology biotech, 45 employees, 1 approved IND
Project duration: 3 weeks (computational delivery) + 5 weeks (validation)
Total cost: $420K
Date: March to April 2025

This case study is anonymized at client request. Compound structures and institutional affiliations have been redacted. Full protocols available under NDA.

Run a similar program with HelixForge.

Tell us about your target. We will scope a 2 to 4 week pilot.

Schedule a Consultation [email protected]