RP2D Determination in Early-Phase Clinical Trials: A Practical Overview

Early-phase clinical trials are where the trajectory of a drug program is set. Dose selection in Phase I directly determines whether a candidate advances with regulatory confidence or struggles through Phase II and Phase III with unresolved questions. Despite its impact, this process is frequently underplanned.

At the center of this decision is the Recommended Phase II Dose (RP2D). According to a systematic analysis, trials of molecularly targeted agents (MTAs) that used non-toxicity or multi-endpoint approaches to define RP2D showed a significantly higher rate of FDA drug approval, with an odds ratio of 5.03 compared with trials that relied solely on toxicity endpoints. This makes the methodology behind RP2D selection not just a scientific question, but a strategic one with direct consequences for regulatory outcomes.

For clinical development teams navigating this complexity, a structured understanding of RP2D clinical trial methodology, from escalation design to pharmacokinetic (PK)/pharmacodynamic (PD) integration, expansion cohorts, and documentation, is essential before Phase II begins.

Why is RP2D Determination Critical in Early-Phase Clinical Trials?

RP2D determination is one of the most consequential decisions made in Phase I development. Once a dose is carried forward into Phase II, reversing that decision becomes operationally complex and regulatorily visible.

The importance of RP2D selection lies in its downstream impact on both development efficiency and regulatory confidence.

Specifically, RP2D determination influences:

• Efficacy signal integrity, doses set too low may fail to demonstrate activity, while doses set too high can obscure benefit due to poor tolerability.
• Safety and patient retention are compromised by excessive toxicity, which leads to early discontinuations, dose reductions, and incomplete exposure that compromise data interpretability.
• Phase II and III program stability, where an unsupported RP2D increases the likelihood of protocol amendments, additional dose-finding studies, or delayed enrollment.
• Regulatory scrutiny at submission, given FDA and EMA expectations that multiple dose levels are evaluated early to justify the selected dose for later-phase studies.
• Post-marketing risk, where insufficient early-dose exploration has historically triggered the need for dose-optimization studies after approval.

For clinical operations and development teams, a well-supported RP2D reduces avoidable rework later in the program and strengthens the credibility of the overall evidence package entering Phase II.

Key Concepts Used in RP2D Determination

RP2D determination relies on integrating multiple dimensions of early-phase data rather than applying a single escalation endpoint. Understanding how traditional dose-finding concepts relate to modern RP2D selection is essential before evaluating escalation methods, PK/PD inputs, and expansion cohort design.

The distinction between Maximum Tolerated Dose (MTD) and Recommended Phase II Dose (RP2D) illustrates this shift in approach.

Differences Between MTD and RP2D

Parameter	Maximum Tolerated Dose (MTD)	Recommended Phase II Dose (RP2D)
Basis	DLT rate during dose escalation.	Integrated safety, PK, PD, and early efficacy data.
Position	Identified during escalation.	Confirmed after escalation, often post-expansion cohort.
Equivalence	Starting reference point.	May be set at, or below, the MTD.

For cytotoxic agents, the MTD remains a practical upper boundary. For MTAs and immunotherapies, dose-response relationships are frequently non-linear, and the biologically active dose may be well below the MTD. The RP2D synthesizes this distinction and must be explicitly justified on the basis of all available evidence.

Dose Escalation Methods Used to Support RP2D Decisions

The escalation design determines the quality of the MTD estimate that feeds RP2D selection. Three primary categories are in use:

• Rule-based: The 3+3 design enrolls cohorts of three patients per dose level, escalating when fewer than two dose-limiting toxicities (DLTs) are observed. It is operationally straightforward and globally applicable, but statistical simulations show it accurately identifies the true MTD in as few as 30% of trials.
• Model-based: The Continual Reassessment Method (CRM) uses Bayesian modeling to estimate the dose-toxicity curve after each cohort, assigning the next patients to the dose closest to the target DLT probability. It produces more accurate MTD estimates but requires real-time biostatistical support.
• Model-assisted: Designs such as BOIN (Bayesian Optimal Interval design), Keyboard, and mTPI-2 retain the transparency of rule-based methods while incorporating the statistical accuracy of model-based approaches. These have become the default in many US oncology Phase I programs.

Design selection should reflect the agent’s mechanism of action, the expected shape of the dose-toxicity curve, and the target regulatory pathway.

Role of Pharmacokinetics and Pharmacodynamics in RP2D Selection

For MTAs and immunotherapies, PK and PD analyses are primary inputs to RP2D decision-making, not supplementary data.

PK questions that must be answered at the candidate RP2D:

• Does the dose achieve adequate systemic exposure, measured by AUC (area under the curve) and Cmax, to produce biological activity?
• Is exposure linear and predictable across dose levels?
• Are there signs of accumulation or autoinduction affecting schedule selection?

PD questions:

• Does the drug engage its target at the selected dose? Target protein modulation, receptor occupancy, and pathway suppression are standard PD indicators.
• What is the minimum dose that achieves near-maximal PD effect? This defines the Optimal Biological Dose (OBD), which may be lower than the MTD.

Where target saturation occurs below the MTD, the RP2D should reflect the OBD, supported by PD biomarker data and exposure-response modeling. Selecting a higher dose in this scenario increases toxicity without increasing biological activity, a pattern that has driven post-marketing dose-optimization requirements across multiple oncology programs.

Use of Dose Expansion Cohorts in RP2D Determination

Dose expansion cohorts (DECs) are added after escalation to generate additional safety, efficacy, and PK/PD data at the candidate RP2D before Phase II begins. A 2025 systematic review covering 479 Phase I oncology trials with expansion cohorts found that 26.5% of trials contributed novel safety findings during the expansion phase, and 5.2% resulted in RP2D modification after the expansion.

Key design principles for expansion cohorts:

• Define the patient population and disease indication before enrollment begins.
• Pre-specify primary endpoints (safety) and secondary objectives (biomarker, efficacy).
• Establish protocol-specified stopping rules for excess toxicity or futility.
• Avoid ad hoc subgroup selection based on small-sample efficacy patterns, which increases the risk of false-positive signals.

Safety and Tolerability Factors Influencing RP2D Decisions

DLT data from an escalation establishes the initial safety boundary, but it is insufficient on its own. The RP2D selection should also incorporate cumulative toxicities observed across multiple cycles, dose modification rates (reductions, holds, and discontinuations) at the proposed RP2D, and relative dose intensity (RDI), the proportion of planned dose actually delivered over a defined treatment period.

Best practices for collecting and evaluating DLT data:

• DLT windows must be pre-specified in the protocol, typically the first cycle, but extended for agents with late-onset or cumulative toxicity profiles.
• Grade and causality assessments must follow CTCAE (Common Terminology Criteria for Adverse Events) standards, with clear attribution guidance for multi-drug regimens.
• Separate DLT definitions for cytotoxic and non-cytotoxic agents to avoid applying inappropriate thresholds.
• Capture quality-of-life-affecting toxicities that may not reach Grade 3/4 thresholds but affect sustained dose delivery.

A dose at which a substantial proportion of patients require early reductions is operationally unsuitable as an RP2D, even if the formal DLT rate falls within the pre-specified threshold.

Regulatory and Documentation Considerations for RP2D Selection

Both the FDA and EMA (European Medicines Agency) evaluate RP2D justification as part of IND (Investigational New Drug) proceedings and NDA (New Drug Application)/BLA (Biologics License Application) submissions.

Documentation That Strengthens RP2D Justification

Key protocol and analysis elements reviewers expect:

• A pre-specified dose escalation algorithm with defined stopping rules and DLT criteria.
• PK summaries by dose level, including individual profiles and population-level exposure-response analyses.
• PD biomarker summaries demonstrating target engagement at the proposed RP2D.
• A clinical dose selection summary document integrating safety, PK, PD, and preliminary efficacy into a unified evidence narrative.
• A written rationale for any downward dose adjustment relative to the MTD.
• Appendices for the expansion cohort protocol, including cohort definitions, stopping criteria, and sample size justification.

Practical Steps to Confirm RP2D Before Phase II

Before advancing into Phase II, the proposed RP2D must be supported by a complete and internally consistent evidence package. This step is not about generating new signals, but about confirming that the selected dose can be carried forward with scientific and regulatory confidence.

In practice, confirmation of RP2D involves the following sequence of activities:

1. Complete dose escalation using a pre-specified stopping rule, target DLT probability, and escalation algorithm.
2. Evaluate PK exposure and PD biomarker activity at the candidate RP2D to determine if the OBD and MTD converge or diverge.
3. Open and complete the expansion cohort with protocol-specified objectives, stopping rules, and a pre-defined patient population.
4. Compile a clinical dose selection summary covering DLTs at all dose levels, PK parameters, PD data, dose modification rates, and any preliminary efficacy signals.
5. Confirm the proposed RP2D and treatment schedule map directly to the Phase II protocol, and re-evaluate PK/PD assumptions if any schedule parameters change.

How RP2D Determination Varies by Therapeutic Approach?

The principles of RP2D determination apply broadly, but implementation differs by agent class:

• Cytotoxic agents: MTD-based approaches remain appropriate. Dose-response relationships are steeper and more predictable. RP2D commonly equals MTD.
• Molecularly targeted agents: PK/PD endpoints and OBD carry primary weight. The MTD may substantially exceed the biologically necessary dose.
• Immunotherapies: DLT-based escalation alone may underestimate the incidence of irAEs (immune-related adverse events), which are often delayed and cumulative. Extended safety follow-up beyond cycle one is required, along with immune biomarker evaluation.
• Antibody-drug conjugates (ADCs): Dual toxicity profiles from the antibody and cytotoxic payload components require comprehensive safety data from expanded cohorts before RP2D confirmation.

Conclusion

RP2D determination in early-phase clinical trials is a multi-domain process. It combines dose escalation methodology, PK/PD science, DLT data analysis, and regulatory documentation into a single evidence package that must hold up under Phase II scrutiny and regulatory review.

The shift toward multi-endpoint RP2D definitions, driven by FDA guidance updates and accumulated clinical evidence, reflects an operational reality: the highest tolerated dose is not always the right dose. Building an RP2D dossier that reflects this, with adaptive escalation designs, pre-specified biomarker strategies, well-structured expansion cohorts, and integrated clinical pharmacology analyses, is now a baseline expectation for competitive oncology programs in the US market.

Teams entering early-phase development should treat RP2D determination as a primary endpoint from the earliest protocol design, not as a milestone to be achieved quickly before Phase II begins.