Overview
A clinical trial protocol is the foundational document governing every aspect of a study — and regulatory agencies reject protocols that miss required elements, which can delay a trial start by 6-12 months and cost millions. ICH E6(R3) reorganized GCP around a risk-proportionate, quality-by-design framework, while ICH E8(R1) emphasizes fit-for-purpose study design. The challenge isn’t just listing endpoints and procedures — it’s building a protocol where the estimand framework (ICH E9(R1)) aligns with the study objectives, the statistical analysis plan properly powers co-primary or composite endpoints, safety monitoring follows jurisdiction-specific reporting timelines (FDA IND Safety Reports vs. EU SUSARs via EudraVigilance), and the RBQMS identifies critical-to-quality factors before enrollment begins. This skill handles all phases (I-IV), therapeutic areas, and regulatory pathways (FDA IND, EU CTR 536/2014 via CTIS, Health Canada CTA), adapting to your study’s specific design requirements.
What It Covers
- Study objectives & endpoints — Primary, secondary, and exploratory endpoints with clinical justification, including estimand framework (population, treatment, intercurrent events, summary measure) per ICH E9(R1)
- Study design & population — Phase-appropriate design (dose-escalation, randomized, adaptive, platform, basket/umbrella), eligibility criteria, stratification, sample size rationale
- Treatment & procedures — Dosing, schedule, blinding, randomization, concomitant therapy, discontinuation rules, dose modifications, schedule of assessments
- Safety monitoring — AE/SAE definitions, SAE reporting timelines by jurisdiction (FDA 15-day IND Safety Reports per 21 CFR 312.32 vs. EU 7/15-day SUSARs via EudraVigilance), IDMC charter, stopping rules, pregnancy reporting
- Statistical considerations — Hypothesis testing, sample size calculation, analysis populations (ITT, mITT, PP), multiplicity adjustment for co-primary/secondary endpoints, interim analysis with alpha spending, missing data strategy
- Risk-Based Quality Management (RBQMS) — Critical-to-quality factors, risk assessment, centralized monitoring, key risk indicators per ICH E6(R3)
- Regulatory requirements — FDA IND content (21 CFR 312), EU CTR 536/2014 via CTIS, ICH E6(R3) protocol content requirements, investigator brochure cross-reference
Example Conversation
You: I’m the medical director at a mid-size biotech. We’re developing a PD-1/TIGIT bispecific antibody (RKN-4820) for first-line treatment of advanced non-small cell lung cancer. We’re planning a Phase III randomized trial comparing RKN-4820 + carboplatin/pemetrexed vs. pembrolizumab + carboplatin/pemetrexed in patients with non-squamous NSCLC, PD-L1 TPS ≥ 1%, no EGFR/ALK alterations. Co-primary endpoints of PFS and OS. We need the full protocol for FDA IND submission.
Agent: This is a registrational Phase III trial in a competitive space — the protocol needs to be bulletproof. Let me confirm several critical design elements:
- Randomization ratio — 1:1 or 2:1 (experimental:control)? 2:1 favors enrollment speed but reduces power per patient on the control arm.
- Stratification factors — PD-L1 (1-49% vs. ≥50%), ECOG PS (0 vs. 1), and geographic region are standard. Any others?
- PFS assessment — blinded independent central review (BICR) per RECIST 1.1, or investigator assessment? FDA strongly prefers BICR for PFS-based approvals.
- Crossover — will you allow crossover from the control arm to RKN-4820 at progression? This significantly affects OS analysis.
- Biomarker strategy — are you co-developing a companion diagnostic, or using the existing PD-L1 22C3 pharmDx assay?
- Line of therapy — strictly 1L, or will you allow patients with prior adjuvant/neoadjuvant therapy that progressed ≥ 6 months after completion?
You: 1:1 randomization. Standard stratification factors you listed are fine. BICR for PFS using RECIST 1.1. No crossover — we want a clean OS signal. Using the existing 22C3 assay for PD-L1 scoring. Strictly 1L — no prior systemic therapy for advanced disease, but prior adjuvant/neoadjuvant allowed if ≥ 12 months since completion.
Agent: Solid design choices. No crossover strengthens the OS analysis considerably but will need careful justification in the protocol for ethical review — you’ll need clear language about why crossover is not offered and the scientific rationale. I’ll create
output/clinical-trial-protocol.mdand start building.For the estimand framework (ICH E9(R1), which FDA now expects in Phase III protocols), I’m defining:
- Population: adults with confirmed Stage IV non-squamous NSCLC, PD-L1 TPS ≥ 1%, no EGFR/ALK, ECOG 0-1
- Variable (endpoint): PFS per BICR (RECIST 1.1); OS
- Intercurrent events: treatment discontinuation due to toxicity → treatment policy strategy (include regardless of discontinuation); initiation of new anti-cancer therapy → composite strategy (count as PFS event)
- Population-level summary: hazard ratio and difference in median PFS/OS
What’s your target enrollment? For PFS HR 0.70 and OS HR 0.75, at 80% power with one interim analysis for OS (O’Brien-Fleming alpha spending), you’re looking at approximately 600 patients.
You: 640 patients, 320 per arm. We’ve budgeted for one formal interim analysis for OS at 60% information fraction, with O’Brien-Fleming spending function. Final PFS analysis when ~400 PFS events have occurred.
Agent: I’ve completed the statistical section with co-primary endpoints (PFS and OS), hierarchical testing (PFS tested first at full alpha, OS tested at remaining alpha after PFS boundary crossing), and the interim analysis at 60% information fraction using O’Brien-Fleming. The multiplicity strategy preserves the overall Type I error at 0.025 (one-sided).
For the safety section — since this is an immuno-oncology combination, I’m including specific immune-related adverse event (irAE) management guidelines, infusion reaction protocols, and pneumonitis monitoring. SAE reporting follows FDA IND requirements: 15 calendar days for unexpected serious adverse reactions (IND Safety Reports per 21 CFR 312.32), 7 days for fatal/life-threatening events. I’ve also included the IDMC charter framework with pre-specified stopping boundaries for both futility and efficacy at the interim analysis. Want me to include specific dose modification tables for the chemotherapy backbone?
Output Excerpt
After the agent completes all sections, the output includes the complete protocol narrative ready for FDA IND submission. Here is a representative excerpt showing the eligibility criteria, SAE reporting, and statistical considerations:
# Clinical Trial Protocol
**Protocol Title:** A Phase III, Randomized, Open-Label Study of RKN-4820
(PD-1/TIGIT Bispecific Antibody) Plus Carboplatin/Pemetrexed vs.
Pembrolizumab Plus Carboplatin/Pemetrexed in Participants with Previously
Untreated Advanced Non-Squamous Non-Small Cell Lung Cancer
**Protocol Number:** RKN-4820-301
**Phase:** III
**Sponsor:** [Sponsor]
**IND Number:** [To be assigned]
**Version:** 1.0 / March 2026
---
## 5. Study Population
### 5.1 Inclusion Criteria
1. Age ≥ 18 years at time of informed consent
2. Histologically or cytologically confirmed Stage IV non-squamous NSCLC
(per AJCC 8th edition)
3. PD-L1 TPS ≥ 1% by the PD-L1 IHC 22C3 pharmDx assay (Dako/Agilent)
4. No prior systemic therapy for advanced/metastatic NSCLC. Prior
adjuvant/neoadjuvant chemotherapy or chemoradiation is permitted if
completed ≥ 12 months before first dose of study treatment.
5. Documented absence of EGFR sensitizing mutations and ALK
rearrangements (local or central testing accepted)
6. ECOG Performance Status 0 or 1
7. At least one measurable lesion per RECIST 1.1
8. Adequate organ function as defined in Table 5.1
9. Women of childbearing potential: negative serum pregnancy test at
screening; agreement to use protocol-specified contraception
10. Willing and able to provide written informed consent
### 5.2 Exclusion Criteria
1. Squamous or mixed histology with predominant squamous component
2. Known EGFR sensitizing mutations (exon 19 del, L858R, T790M, exon 20
insertion) or ALK rearrangement/translocation
3. Active autoimmune disease requiring systemic immunosuppressive therapy
within 2 years (exceptions: vitiligo, resolved childhood asthma,
type 1 diabetes, hypothyroidism on stable replacement)
4. Prior therapy with anti-PD-1, anti-PD-L1, anti-TIGIT, or any immune
checkpoint inhibitor
5. Known active CNS metastases (treated brain metastases stable ≥ 4 weeks
without steroids are eligible)
6. Active interstitial lung disease or history of pneumonitis requiring
systemic corticosteroids
7. Receipt of live vaccine within 30 days before first dose
---
## 10. Safety Reporting
### 10.3 Serious Adverse Event Reporting
| Reporting Obligation | Timeline | Recipient |
|---------------------|----------|-----------|
| SAE initial report (Sponsor) | Within 24 hours of site awareness | Sponsor Safety |
| IND Safety Report — fatal/life-threatening unexpected | 7 calendar days (21 CFR 312.32(c)(2)) | FDA (via MedWatch 3500A) |
| IND Safety Report — other unexpected serious | 15 calendar days (21 CFR 312.32(c)(1)) | FDA (via MedWatch 3500A) |
| Follow-up IND Safety Report | 15 calendar days of new information | FDA |
| Annual IND Safety Report (21 CFR 312.33) | Within 60 days of IND anniversary | FDA |
| IDMC unblinded safety review | Per IDMC charter schedule | IDMC |
Note: IND Safety Reports are required only for suspected adverse
reactions that are both serious AND unexpected (not listed in the
Investigator's Brochure). Expected serious adverse reactions are
tracked in aggregate for the annual IND Safety Report (DSURs may
also be submitted to fulfill ICH E2F requirements).
---
## 11. Statistical Considerations
### 11.1 Co-Primary Endpoints
| Endpoint | Definition | Analysis Population |
|----------|-----------|-------------------|
| Progression-Free Survival (PFS) | Time from randomization to disease progression (BICR per RECIST 1.1) or death from any cause, whichever occurs first | ITT (all randomized) |
| Overall Survival (OS) | Time from randomization to death from any cause | ITT (all randomized) |
### 11.2 Estimand Framework (ICH E9(R1))
**PFS Estimand:**
- **Population:** Adults with Stage IV non-squamous NSCLC, PD-L1 ≥ 1%,
no EGFR/ALK, ECOG 0-1
- **Treatment:** RKN-4820 + carbo/pem vs. pembrolizumab + carbo/pem
- **Intercurrent events:**
- Treatment discontinuation due to AE → treatment policy strategy
- Initiation of new anti-cancer therapy before progression →
composite strategy (counted as PFS event at start of new therapy)
- Death without documented progression → event at date of death
- **Population-level summary:** Hazard ratio (stratified Cox model)
### 11.3 Sample Size
Total enrollment: 640 participants (320 per arm).
| Parameter | PFS | OS |
|-----------|-----|-----|
| Target HR | 0.70 | 0.75 |
| Alpha (one-sided) | 0.025 | 0.025 |
| Power | 80% | 80% |
| Required events | ~400 | ~410 |
| Interim analyses | None | 1 (at 60% information) |
| Alpha spending | — | O'Brien-Fleming |
### 11.4 Multiplicity Adjustment
Co-primary endpoints tested using a hierarchical (fixed-sequence)
procedure: PFS tested first at full α = 0.025 (one-sided). If PFS
is statistically significant, OS is tested at α = 0.025. If PFS
does not cross the significance boundary, OS is tested in a
descriptive/supportive capacity only. This procedure controls the
overall family-wise Type I error rate at 0.025 (one-sided).
The protocol includes all content elements required by ICH E6(R3), with the estimand framework per ICH E9(R1), risk-based quality management considerations per ICH E6(R3), jurisdiction-specific safety reporting timelines, and a rigorous statistical analysis plan with multiplicity control for co-primary endpoints.
Getting Started
Add your compound information, investigator’s brochure, target indication, and any existing study design documents to the project workspace, then activate the Clinical Trial Protocol (ICH E6 R3) skill. The agent will determine your study phase, collect design parameters, and produce a complete protocol narrative — ready for IND submission or CTIS application.