← Eligibility-Criteria Authoring comparison · model: Claude Opus 4.8
LLM-judge score: 4.8 / 5 · strong
safety_exclusions: 5 lab_thresholds: 5 enrollability: 5 precedent_completeness: 5 faithfulness_clarity: 4
The agent authored a comprehensive, mechanism-aware I/E list that specifically protects against the DLL3 TCE class liabilities — CRS (cardiac/pulmonary reserve, infection, inpatient cycle-1 monitoring), ICANS (CNS mets/LMD/seizure exclusions, baseline ICE), cytopenias (ANC/platelet/Hgb cutoffs with marrow-involvement relaxations), and infection/HBV/HCV screening — while explicitly reasoning through the safety-vs-enrollability trade-offs (allowing treated brain mets, ≥1 prior line, no DLL3 cutoff). Lab thresholds are clinically appropriate and tied to precedent (tarlatamab DeLLphi). Minor gaps: tumor flare/pyrexia is not distinctly addressed, hypogammaglobulinemia/Ig monitoring is implicit rather than explicit, and the 'precedent-grounded' framing relies on stated tarlatamab analogy without visible tool citations, slightly weakening evidence faithfulness.
Clinical / Medical · Medical monitor

Eligibility-Criteria Authoring — agent trajectory for expert review

claude-opus-4-8 · 2026-06-06T10:24:36.712046+00:00 · 3 tool calls · 36.3s · 7811 tokens
run id: eligibility_authoring_20260606_031816_eligibility-authoring__claude-opus-4-8

Query

Author the inclusion/exclusion criteria for a Phase 1 first-in-human trial of a DLL3 T-cell engager (BiTE) in relapsed/refractory extensive-stage small cell lung cancer (ES-SCLC), accounting for cytokine release syndrome and neurotoxicity (ICANS) risk. Balance safety against enrollability.

Trajectory

Agent · turn 1

I'll pull the relevant safety profile, precedent criteria, and lab thresholds to ground this in precedent.

⚙ calls get_safety_profile(drug_class=DLL3 T-cell engager)
⚙ calls get_precedent_criteria(indication=ES-SCLC)
⚙ calls get_lab_thresholds()
↳ get_safety_profile result
show data
{
  "drug_class": "DLL3 T-cell engager",
  "class_effect_aes": [
    "Cytokine release syndrome (CRS) \u2014 common, mostly cycle-1, can be high-grade",
    "Immune effector cell-associated neurotoxicity syndrome (ICANS) / neurotoxicity",
    "Cytopenias (neutropenia, anemia, thrombocytopenia)",
    "Hypogammaglobulinemia / infection risk from on-target B/neuroendocrine effects",
    "Tumor flare / pyrexia"
  ],
  "monitoring": [
    "Inpatient monitoring for CRS during step-up/priming doses (cycle 1)",
    "Frequent vitals, ferritin/CRP trend during CRS window",
    "Baseline + on-treatment neuro exam (ICE score) for ICANS",
    "Baseline ECG/echo if cardiac reserve is a concern during CRS",
    "CBC with differential before each dose"
  ],
  "contraindicated_comorbidities": [
    "Active CNS metastases / leptomeningeal disease (raises ICANS danger)",
    "Active autoimmune CNS disease or seizure disorder",
    "Inadequate cardiac reserve (CRS hemodynamic stress) \u2014 recent MI, NYHA III/IV CHF",
    "Active uncontrolled infection (CRS masks/worsens sepsis)",
    "Significant baseline hypoxia / poor pulmonary reserve"
  ],
  "ddi_flags": [
    "Live vaccines contraindicated (immune activation)",
    "Chronic systemic immunosuppression / high-dose steroids blunt efficacy and confound CRS grading",
    "Tocilizumab/steroids must be available on-site as CRS rescue"
  ],
  "notes": "First-in-human TCEs need a step-up/priming dose schedule and mandated CRS/ICANS rescue medication on hand; cycle-1 inpatient observation is standard."
}
↳ get_precedent_criteria result
show data
{
  "indication": "ES-SCLC",
  "precedent_trials": [
    "Tarlatamab DeLLphi-301/304 (DLL3 TCE)",
    "IMpower133 (1L chemo-IO)"
  ],
  "inclusion": [
    "Histologically/cytologically confirmed extensive-stage SCLC",
    "Relapsed/refractory after >=1 prior platinum-based regimen (line per protocol)",
    "Measurable disease per RECIST v1.1",
    "ECOG performance status 0-1",
    "Adequate hematologic, hepatic, renal organ function (see lab thresholds)",
    "Life expectancy >=12 weeks"
  ],
  "exclusion": [
    "Untreated/active/symptomatic CNS metastases (treated, stable may be allowed)",
    "Leptomeningeal disease / carcinomatous meningitis",
    "Active autoimmune disease or condition needing systemic immunosuppression (for IO/TCE)",
    "Insufficient washout from prior anticancer therapy (e.g. ~3-4 wks or 5 half-lives)",
    "Uncontrolled intercurrent illness / active infection",
    "Second malignancy requiring active treatment"
  ],
  "notes": "DLL3 TCE precedents often allow treated/stable brain mets but exclude active CNS disease specifically because of ICANS/neurotoxicity risk."
}
↳ get_lab_thresholds result
show data
{
  "hematologic": [
    {
      "parameter": "Absolute neutrophil count (ANC)",
      "typical_threshold": ">= 1.5 x 10^9/L",
      "rationale": "Adequate marrow reserve before a myelosuppressive agent; lower (>=1.0) sometimes allowed if marrow involvement."
    },
    {
      "parameter": "Platelets",
      "typical_threshold": ">= 100 x 10^9/L (>=75 if marrow involvement)",
      "rationale": "Bleeding risk and tolerance of thrombocytopenia."
    },
    {
      "parameter": "Hemoglobin",
      "typical_threshold": ">= 9.0 g/dL (transfusion-independent)",
      "rationale": "Baseline reserve; many protocols require independence from recent transfusion."
    }
  ],
  "hepatic": [
    {
      "parameter": "Total bilirubin",
      "typical_threshold": "<= 1.5 x ULN (<=3 x ULN if Gilbert's/liver mets)",
      "rationale": "Hepatic clearance reserve; relaxed for documented Gilbert's or hepatic metastases."
    },
    {
      "parameter": "AST / ALT",
      "typical_threshold": "<= 3 x ULN (<=5 x ULN if liver metastases)",
      "rationale": "Baseline hepatocellular integrity; key for hepatotoxic agents (PI3Ki, JAKi)."
    },
    {
      "parameter": "Albumin",
      "typical_threshold": ">= 2.8-3.0 g/dL (some protocols)",
      "rationale": "Synthetic liver function / nutritional status, esp. GI cancers."
    }
  ],
  "renal": [
    {
      "parameter": "Creatinine clearance (Cockcroft-Gault)",
      "typical_threshold": ">= 50-60 mL/min",
      "rationale": "Renal elimination reserve; threshold depends on renal clearance of the drug."
    },
    {
      "parameter": "Serum creatinine",
      "typical_threshold": "<= 1.5 x ULN",
      "rationale": "Simpler surrogate when CrCl not used."
    }
  ],
  "cardiac": [
    {
      "parameter": "Left-ventricular ejection fraction (LVEF)",
      "typical_threshold": ">= 50% (institutional LLN)",
      "rationale": "Required for cardiotoxic agents (anti-HER2, anthracycline exposure)."
    },
    {
      "parameter": "QTc (Fridericia)",
      "typical_threshold": "<= 470 ms (women) / <= 450 ms (men)",
      "rationale": "Exclude baseline prolongation for QT-liable agents; pair with congenital long-QT exclusion."
    }
  ],
  "coagulation_metabolic": [
    {
      "parameter": "INR / aPTT (if on anticoagulation off-study)",
      "typical_threshold": "within therapeutic range / <= 1.5 x ULN",
      "rationale": "Bleeding risk, esp. for agents with hemorrhage signal (BTKi)."
    },
    {
      "parameter": "Fasting glucose / HbA1c",
      "typical_threshold": "HbA1c <= 6.5-8.0% per protocol",
      "rationale": "Baseline glycemic control for hyperglycemia-liable agents (PI3K-alpha)."
    }
  ],
  "notes": "Teaching-snapshot typical oncology cutoffs. Exact thresholds and relaxations (marrow involvement, liver mets, Gilbert's) are drug- and protocol-specific and must be set against the investigator's brochure and standard of care."
}
Agent · turn 2

Phase 1 First-in-Human Trial — DLL3 T-Cell Engager (BiTE) in R/R ES-SCLC

Inclusion / Exclusion Criteria

Authored against DLL3 TCE precedent (tarlatamab DeLLphi-301/304) and TCE class-effect safety (CRS, ICANS, on-target neuroendocrine effects). Design intent: protect against the cycle-1 CRS/ICANS window while keeping a heavily pretreated ES-SCLC population enrollable.


INCLUSION CRITERIA

Disease
1. Histologically or cytologically confirmed extensive-stage small cell lung cancer (ES-SCLC).
2. Relapsed or refractory disease after ≥1 prior platinum-based regimen. (FIH dose-escalation: do not over-restrict prior lines — allow ≥1 to maximize the eligible pool; a sensitive- vs. platinum-refractory stratification factor is preferable to an exclusion.)
3. At least one measurable lesion per RECIST v1.1.
4. Available archival or fresh tumor tissue (DLL3 expression may be collected as a biomarker but not required for eligibility in a FIH study, to preserve enrollability and characterize the expression–response relationship).

Performance / reserve
5. ECOG performance status 0–1.
6. Life expectancy ≥12 weeks.

Organ function (within 7–14 days of first dose)
7. ANC ≥1.5 ×10⁹/L (≥1.0 acceptable if documented marrow involvement).
8. Platelets ≥100 ×10⁹/L (≥75 if marrow involvement).
9. Hemoglobin ≥9.0 g/dL, transfusion-independent for ≥2 weeks.
10. Total bilirubin ≤1.5× ULN (≤3× ULN if Gilbert's or hepatic metastases).
11. AST/ALT ≤3× ULN (≤5× ULN if hepatic metastases).
12. Creatinine clearance ≥50 mL/min (Cockcroft-Gault) or serum creatinine ≤1.5× ULN.
13. Adequate coagulation if on anticoagulation: INR/aPTT within therapeutic range.

Safety-context (TCE-specific)
14. Adequate cardiac reserve to tolerate CRS hemodynamic stress: LVEF ≥50%; no QTc prolongation (>470 ms women / >450 ms men).
15. Adequate pulmonary reserve: baseline O₂ saturation ≥92% on room air, no supplemental-oxygen dependence — CRS-associated hypoxia is a key cycle-1 hazard.
16. Baseline normal/near-normal neurologic exam with documented baseline ICE score, to enable ICANS detection.
17. Willing/able to be hospitalized for inpatient monitoring during step-up/priming doses (cycle 1).
18. Recovered to ≤Grade 1 from prior-therapy toxicities (excluding alopecia, stable neuropathy).


EXCLUSION CRITERIA

CNS / neurotoxicity risk (heightened priority for a TCE — ICANS)
1. Active, untreated, or symptomatic CNS metastases. (Treated, stable, asymptomatic brain mets — radiographically stable ≥4 weeks, off steroids ≥7 days — may be ALLOWED; precedent permits this and excluding all brain mets would gut SCLC enrollability.)
2. Leptomeningeal disease / carcinomatous meningitis (absolute exclusion — disproportionate ICANS danger).
3. History of seizure disorder, or active CNS autoimmune/inflammatory disease (confounds and amplifies ICANS).

Immune / infection (CRS confounders)
4. Active autoimmune disease requiring systemic immunosuppression within the prior 2 years (well-controlled endocrinopathies on replacement, e.g., thyroid, are permitted).
5. Chronic systemic corticosteroids >10 mg/day prednisone-equivalent or other immunosuppression at screening (blunts efficacy, confounds CRS grading) — taper required pre-dose.
6. Active, uncontrolled infection (CRS masks/worsens sepsis); known active HIV (unless controlled per protocol), active HBV/HCV.

Cardiopulmonary (CRS tolerance)
7. Myocardial infarction or unstable angina within 6 months; NYHA Class III/IV heart failure; clinically significant uncontrolled arrhythmia.
8. Interstitial lung disease/pneumonitis requiring steroids, or clinically significant baseline hypoxia.

General oncology
9. Insufficient washout from prior anticancer therapy (≥3–4 weeks or 5 half-lives, whichever shorter; ≥2 weeks for palliative RT).
10. Second malignancy requiring active treatment (excluding adequately treated in-situ/non-melanoma skin cancers).
11. Uncontrolled intercurrent illness.
12. Live vaccine within 4 weeks of first dose (contraindicated with immune activation).
13. Pregnancy/lactation; refusal of effective contraception.


Operational safeguards (protocol-level, supporting the criteria)


Key balance decisions (rationale)

Tension Decision Why
Brain mets Exclude active; allow treated/stable ES-SCLC has high CNS-met prevalence; full exclusion would cripple accrual. ICANS risk managed via "active/symptomatic" and LMD carve-outs.
Prior lines ≥1 only, stratify rather than cap FIH should access the broadest R/R pool; refractory status captured as covariate.
DLL3 expression Not required for eligibility Preserves enrollability and lets the study define the expression–response relationship in dose-escalation.
Cardiopulmonary bars Added explicit O₂ sat + LVEF/QTc + recent-MI/CHF exclusions These are TCE-specific additions beyond generic SCLC criteria — CRS imposes hemodynamic/hypoxic stress not present with chemo.
Steroids Threshold (≤10 mg) rather than absolute ban Allows physiologic replacement; protects efficacy and clean CRS grading.

Teaching/precedent-grounded draft. Exact thresholds, washouts, and brain-met stability windows must be finalized against the Investigator's Brochure, the specific step-up schedule, and regulatory input.

Want me to draft the matching CRS/ICANS management and dose-modification tables, or the stratification/dose-escalation (e.g., BOIN/3+3) design to pair with these criteria?

Final output

Phase 1 First-in-Human Trial — DLL3 T-Cell Engager (BiTE) in R/R ES-SCLC

Inclusion / Exclusion Criteria

Authored against DLL3 TCE precedent (tarlatamab DeLLphi-301/304) and TCE class-effect safety (CRS, ICANS, on-target neuroendocrine effects). Design intent: protect against the cycle-1 CRS/ICANS window while keeping a heavily pretreated ES-SCLC population enrollable.


INCLUSION CRITERIA

Disease
1. Histologically or cytologically confirmed extensive-stage small cell lung cancer (ES-SCLC).
2. Relapsed or refractory disease after ≥1 prior platinum-based regimen. (FIH dose-escalation: do not over-restrict prior lines — allow ≥1 to maximize the eligible pool; a sensitive- vs. platinum-refractory stratification factor is preferable to an exclusion.)
3. At least one measurable lesion per RECIST v1.1.
4. Available archival or fresh tumor tissue (DLL3 expression may be collected as a biomarker but not required for eligibility in a FIH study, to preserve enrollability and characterize the expression–response relationship).

Performance / reserve
5. ECOG performance status 0–1.
6. Life expectancy ≥12 weeks.

Organ function (within 7–14 days of first dose)
7. ANC ≥1.5 ×10⁹/L (≥1.0 acceptable if documented marrow involvement).
8. Platelets ≥100 ×10⁹/L (≥75 if marrow involvement).
9. Hemoglobin ≥9.0 g/dL, transfusion-independent for ≥2 weeks.
10. Total bilirubin ≤1.5× ULN (≤3× ULN if Gilbert's or hepatic metastases).
11. AST/ALT ≤3× ULN (≤5× ULN if hepatic metastases).
12. Creatinine clearance ≥50 mL/min (Cockcroft-Gault) or serum creatinine ≤1.5× ULN.
13. Adequate coagulation if on anticoagulation: INR/aPTT within therapeutic range.

Safety-context (TCE-specific)
14. Adequate cardiac reserve to tolerate CRS hemodynamic stress: LVEF ≥50%; no QTc prolongation (>470 ms women / >450 ms men).
15. Adequate pulmonary reserve: baseline O₂ saturation ≥92% on room air, no supplemental-oxygen dependence — CRS-associated hypoxia is a key cycle-1 hazard.
16. Baseline normal/near-normal neurologic exam with documented baseline ICE score, to enable ICANS detection.
17. Willing/able to be hospitalized for inpatient monitoring during step-up/priming doses (cycle 1).
18. Recovered to ≤Grade 1 from prior-therapy toxicities (excluding alopecia, stable neuropathy).


EXCLUSION CRITERIA

CNS / neurotoxicity risk (heightened priority for a TCE — ICANS)
1. Active, untreated, or symptomatic CNS metastases. (Treated, stable, asymptomatic brain mets — radiographically stable ≥4 weeks, off steroids ≥7 days — may be ALLOWED; precedent permits this and excluding all brain mets would gut SCLC enrollability.)
2. Leptomeningeal disease / carcinomatous meningitis (absolute exclusion — disproportionate ICANS danger).
3. History of seizure disorder, or active CNS autoimmune/inflammatory disease (confounds and amplifies ICANS).

Immune / infection (CRS confounders)
4. Active autoimmune disease requiring systemic immunosuppression within the prior 2 years (well-controlled endocrinopathies on replacement, e.g., thyroid, are permitted).
5. Chronic systemic corticosteroids >10 mg/day prednisone-equivalent or other immunosuppression at screening (blunts efficacy, confounds CRS grading) — taper required pre-dose.
6. Active, uncontrolled infection (CRS masks/worsens sepsis); known active HIV (unless controlled per protocol), active HBV/HCV.

Cardiopulmonary (CRS tolerance)
7. Myocardial infarction or unstable angina within 6 months; NYHA Class III/IV heart failure; clinically significant uncontrolled arrhythmia.
8. Interstitial lung disease/pneumonitis requiring steroids, or clinically significant baseline hypoxia.

General oncology
9. Insufficient washout from prior anticancer therapy (≥3–4 weeks or 5 half-lives, whichever shorter; ≥2 weeks for palliative RT).
10. Second malignancy requiring active treatment (excluding adequately treated in-situ/non-melanoma skin cancers).
11. Uncontrolled intercurrent illness.
12. Live vaccine within 4 weeks of first dose (contraindicated with immune activation).
13. Pregnancy/lactation; refusal of effective contraception.


Operational safeguards (protocol-level, supporting the criteria)


Key balance decisions (rationale)

Tension Decision Why
Brain mets Exclude active; allow treated/stable ES-SCLC has high CNS-met prevalence; full exclusion would cripple accrual. ICANS risk managed via "active/symptomatic" and LMD carve-outs.
Prior lines ≥1 only, stratify rather than cap FIH should access the broadest R/R pool; refractory status captured as covariate.
DLL3 expression Not required for eligibility Preserves enrollability and lets the study define the expression–response relationship in dose-escalation.
Cardiopulmonary bars Added explicit O₂ sat + LVEF/QTc + recent-MI/CHF exclusions These are TCE-specific additions beyond generic SCLC criteria — CRS imposes hemodynamic/hypoxic stress not present with chemo.
Steroids Threshold (≤10 mg) rather than absolute ban Allows physiologic replacement; protects efficacy and clean CRS grading.

Teaching/precedent-grounded draft. Exact thresholds, washouts, and brain-met stability windows must be finalized against the Investigator's Brochure, the specific step-up schedule, and regulatory input.

Want me to draft the matching CRS/ICANS management and dose-modification tables, or the stratification/dose-escalation (e.g., BOIN/3+3) design to pair with these criteria?

Reference answer key (deterministic) — open after grading

Expected-coverage checklist for drug class 'DLL3 T-cell engager' — a grading aid, NOT ground truth. A strong I/E list visibly protects against each liability (exclusion / lab cutoff / monitoring) while keeping the target population enrollable.

Class-effect liability to addressWhy it constrains eligibility
Cytokine release syndrome (CRS) — common, mostly cycle-1, can be high-gradeDrives cycle-1 inpatient monitoring + exclude poor cardiac/pulmonary reserve and active infection.
Immune effector cell-associated neurotoxicity syndrome (ICANS) / neurotoxicityExclude active CNS mets / leptomeningeal / seizure disorder; baseline neuro exam.
Cytopenias (neutropenia, anemia, thrombocytopenia)Set ANC/platelet/hemoglobin thresholds; CBC monitoring.
Hypogammaglobulinemia / infection risk from on-target B/neuroendocrine effectsBaseline TB/HBV/HCV screening; exclude active/latent untreated infection.
Tumor flare / pyrexiaShould be reflected in an exclusion, lab threshold, or monitoring requirement.

Expert grading

Safety-driven exclusions 1 (poor) 5 (excellent)
5: Authors exclusions that specifically protect against THIS drug's liabilities, grounded in its pulled safety profile — e.g. for a DLL3 TCE, excludes active CNS mets and poor cardiac/pulmonary reserve for CRS/ICANS; for a HER2 ADC, hard-excludes ILD history; for a JAK inhibitor, excludes untreated TB/active infection and prior VTE.
1: Generic exclusions that ignore the drug's mechanism, or misses the headline class risk (e.g. no ILD exclusion for a deruxtecan-class ADC; no infection screen for a JAK inhibitor).
Organ-function / lab thresholds 1 (poor) 5 (excellent)
5: Specifies appropriate, justified organ-function cutoffs (ANC, platelets, hemoglobin, bilirubin, AST/ALT, creatinine clearance, LVEF, QTc, HbA1c) tied to the drug's toxicities and the precedent thresholds — e.g. tighter LFT/glucose cutoffs for a hepatotoxic/hyperglycemic PI3Ki, LVEF for a cardiotoxic ADC.
1: Arbitrary or missing thresholds, copies generic numbers without tying them to the drug's risks, or sets cutoffs that are clinically implausible.
Enrollability balance 1 (poor) 5 (excellent)
5: Keeps the target population realistically enrollable — relaxes thresholds where precedent allows (marrow involvement, liver mets, Gilbert's; treated/stable brain mets), and does not stack so many exclusions that the trial cannot accrue. Explicitly reasons about the safety-vs-enrollability trade-off.
1: Over-restrictive: excludes the realistic target population into infeasibility (e.g. bars all prior lines, all brain mets, demands pristine organ function) with no enrollability reasoning; or the opposite, ignores safety to maximize accrual.
Precedent alignment & completeness 1 (poor) 5 (excellent)
5: Covers the standard I/E domains for the indication (histologic confirmation, ECOG, measurable disease per RECIST, washout, prior-line definition, life expectancy) and is consistent with the analogous-trial precedent it pulled.
1: Omits whole standard domains (no performance status, no washout, no disease confirmation), or contradicts the precedent criteria for the indication.
Evidence faithfulness & clarity 1 (poor) 5 (excellent)
5: Grounds each criterion in the pulled safety/precedent/threshold data rather than memory; criteria are unambiguous and operational (measurable, with explicit values, timeframes, and units a site could screen against).
1: Fabricates risks or thresholds not in the data, contradicts the tool outputs, or writes vague, non-operational criteria a coordinator could not apply.
Overall verdict
Your name / ID