FDA 2024 Dose Optimization Guidance: Regulatory Language Becomes Requirement

FDA 2024 劑量最佳化指引：監管語言成為正式要求

English

There is a meaningful difference between a regulator expressing a preference and a regulator issuing a guidance document. Preferences can be noted and deprioritized; guidance documents inform FDA review decisions, shape the questions reviewers ask, and — in practice — determine whether a marketing application will be approved without conditions or with postmarketing requirements attached. When FDA published “Optimizing the Dosage of Human Prescription Drugs and Biological Products for the Treatment of Oncologic Diseases” in August 2024, the shift from preference to formal expectation became official. The concepts that Project Optimus had been promoting since the early 2020s — that dose selection should be justified by a multi-attribute evidence package, not just the absence of dose-limiting toxicity at the chosen level — were now embedded in the regulatory language that governs oncology drug approval.

The guidance applies to human prescription drugs and biological products for oncologic diseases, including the full range of modern drug types: small molecule targeted therapies, monoclonal antibodies, ADCs (antibody-drug conjugates), bispecific antibodies, and immunotherapies. Notably, it explicitly does not cover radiopharmaceuticals, cellular and gene therapy products, oncolytics, microbiota-based therapies, or cancer vaccines — these require separate frameworks. It also does not directly regulate how to choose a FIH (first-in-human) starting dose; that topic is addressed in the FDA-AACR 2025 publication series. The guidance focuses specifically on the process of identifying an optimized dosage (or dosages) during clinical development, before submitting a new indication application. This is a critical distinction that clinicians reading about dose optimization should internalize: the 2024 guidance is about the development-phase optimization process, not the first-in-human safety calculation.

What does the guidance concretely expect? The central requirement is that sponsors identify an optimized dosage (or set of dosages) during clinical development — before the marketing application is filed. Optimized does not mean lowest possible; it means a dosage where the evidence demonstrates a reasonable benefit-risk profile compared to alternative dosages tested in the program. This requires generating data at multiple dose levels — not just the highest tolerated — and analyzing those data through the lenses of exposure-response for efficacy and exposure-safety for toxicity. Randomized dose comparisons are not always required, but for drugs where multiple candidate doses show similar efficacy signals but different safety profiles, the guidance makes clear that a non-randomized, observational comparison of dose levels is a weak basis for dose selection.

The 2024 guidance also introduced formal language around dose-optimization plans, which should be included in early-phase protocols to specify: which dose levels will be explored, what data will be collected at each level, how those data will inform dose selection for the next stage, and what decision criteria define “optimized.” This formalization matters because it turns dose optimization from an aspirational concept into an auditable protocol element. When a reviewer reads a phase 1 protocol, they can now check whether the sponsor has a pre-specified plan for how they will compare doses and select among them — not just a plan for escalating until toxicity is reached.

The impact on real-world protocol design has been measurable. The JCO Oncology Advances study (Bhamidipati et al., 2025) showed that dose-optimization plans appeared in 30% of phase 1 oncology protocols initiated in 2021-2024, with Bayesian methodology adoption rising from 48% to 75% over this period. But the gap between statistical sophistication and patient-centered evidence remained: patient-reported outcomes were present in only about 12% of protocols. The guidance does emphasize tolerability as a dimension of optimization — patients should be able to sustain treatment — but converting this principle into concrete PRO (patient-reported outcome) endpoints, with pre-specified analysis plans, remains an implementation challenge that the field has not yet solved.

For clinicians who read FIH and early-phase trial publications, the 2024 guidance creates a useful mental checklist: Does the paper present data at multiple dose levels, or only at the selected RP2D (recommended phase 2 dose)? Is there an exposure-response analysis for both efficacy and safety, or only a summary of adverse events and response rates at the chosen dose? Is there evidence of chronic tolerability — beyond the first-cycle DLT window — including dose interruptions, dose reductions, discontinuation rates, and patient-reported symptoms? If these elements are absent or superficially addressed, the paper is presenting a dose hypothesis, not a dose conclusion — and the 2024 FDA guidance would expect the marketing application to contain substantially more evidence than what the paper shows.

中文

監管機構表達偏好和發布指引文件之間，存在有意義的差異。偏好可以被記錄但被降低優先順序；指引文件則指導 FDA 的審查決定，塑造審查員提出的問題，並在實踐中決定上市申請是否無條件核准，或附帶上市後要求。當 FDA 於 2024 年 8 月發布《Optimizing the Dosage of Human Prescription Drugs and Biological Products for the Treatment of Oncologic Diseases》時，從偏好到正式期望的轉變正式生效。Project Optimus 自 2020 年代初一直在推動的概念——劑量選擇應由多屬性證據包來證明合理，而不只是所選水平上沒有劑量限制毒性——現在嵌入了管理腫瘤藥物核准的監管語言中。

該指引適用於腫瘤疾病的人類處方藥和生物製劑，涵蓋全系列現代藥物類型：小分子標靶藥、單株抗體、ADC（抗體藥物複合體）、雙特異性抗體和免疫治療。值得注意的是，它明確不涵蓋放射性藥物、細胞和基因療法產品、溶瘤病毒、微生物群療法或癌症疫苗——這些需要單獨的框架。它也不直接規範如何選擇 FIH 起始劑量；那個主題在 FDA-AACR 2025 系列文章中處理。指引具體聚焦於在提交新適應症申請之前，在臨床開發期間識別最佳化劑量（或多個劑量）的過程。這是閱讀劑量最佳化文獻的臨床醫師應內化的關鍵區別：2024 年指引是關於開發階段的最佳化過程，而非首次人體安全性計算。

指引具體期望什麼？核心要求是申請者在臨床開發期間識別最佳化劑量（或劑量組合）——在提交上市申請之前。最佳化並不意味著盡可能低；它意味著證據證明該劑量與計畫中測試的替代劑量相比，具有合理效益-風險特性的劑量。這需要在多個劑量水平——不只是最高耐受劑量——產生資料，並通過暴露-反應（療效）和暴露-安全性（毒性）的視角分析這些資料。並非總是要求隨機劑量比較，但對於多個候選劑量顯示相似療效訊號但安全性特性不同的藥物，指引明確指出，劑量水平的非隨機觀察性比較是劑量選擇的薄弱基礎。

2024 年指引還引入了關於劑量最佳化計畫的正式語言，這應包含在早期方案中，以明確：將探索哪些劑量水平、在每個水平收集什麼資料、這些資料如何告知下一階段的劑量選擇，以及什麼決策標準定義「最佳化」。這種正式化很重要，因為它將劑量最佳化從一個抱負性概念轉變為可審計的方案元素。當審查員閱讀一期方案時，他們現在可以檢查申請者是否有預先規定的計畫，說明將如何比較劑量並在其中選擇——而不只是升量直到達到毒性的計畫。

對真實世界方案設計的影響是可測量的。JCO Oncology Advances 研究（Bhamidipati 等人，2025）顯示，劑量最佳化計畫出現在 2021-2024 年啟動的 30% 第一期腫瘤方案中，貝氏方法採用率在此期間從 48% 上升到 75%。但統計複雜度和以病人為中心的證據之間的差距依然存在：病人回報結果只出現在約 12% 的方案中。指引確實強調耐受性作為最佳化的一個維度——病人應能持續治療——但將這一原則轉化為具體的 PRO 終點，並帶有預先規定的分析計畫，仍然是該領域尚未解決的落地挑戰。

對閱讀 FIH 和早期試驗文獻的臨床醫師，2024 年指引創建了一個有用的心理清單：論文是否呈現多個劑量水平的資料，還是只呈現選定 RP2D 的資料？是否有療效和安全性兩者的暴露-反應分析，還是只有所選劑量的不良事件和反應率摘要？是否有慢性耐受性的證據——超越第一療程 DLT 視窗——包括劑量暫停、劑量減量、停藥率和病人回報症狀？如果這些元素缺失或膚淺處理，論文呈現的是劑量假說，而非劑量結論——而 2024 年 FDA 指引將期望上市申請包含遠比論文顯示更多的證據。

Key Concepts | 核心概念

• Guidance vs. preference | 指引 vs. 偏好: FDA guidance documents formally shape review decisions; the 2024 document makes dose optimization an auditable expectation, not an informal hope. FDA 指引正式塑造審查決定；2024 年文件使劑量最佳化成為可審計的期望，而非非正式期望。
• Scope boundaries | 適用範圍界限: The guidance covers targeted therapies, mAbs, ADCs, bispecifics, immunotherapies — but explicitly excludes radiopharmaceuticals, cell/gene therapy, oncolytics, cancer vaccines. 涵蓋標靶藥、單株抗體、ADC、雙特異性抗體、免疫治療——但明確排除放射性藥物、細胞/基因療法、溶瘤病毒、癌症疫苗。
• Pre-submission optimization | 提交前最佳化: Optimized dosage must be identified during development, before the marketing application — retrofitting post-approval is what the guidance aims to prevent. 最佳化劑量必須在開發期間識別，在上市申請之前——指引旨在防止核准後才補救。
• Dose-optimization plan | 劑量最佳化計畫: A pre-specified protocol element stating which doses will be explored, what data collected, and how decisions will be made — turning aspiration into auditable action. 預先規定的方案元素，說明將探索哪些劑量、收集什麼資料、如何做決策——將抱負轉化為可審計的行動。
• Gap: PRO underrepresentation | 缺口：PRO 代表性不足: Despite guidance emphasizing tolerability, only ~12% of real-world phase 1 protocols include patient-reported outcomes — the largest implementation gap. 儘管指引強調耐受性，只有約 12% 的真實世界一期方案包含病人回報結果——最大的落地缺口。

Related Pages | 相關頁面

• project-optimus-dose-optimization — The initiative that drove this guidance
• fda-aacr-2025-dose-optimization-trilogy — The companion publications series
• real-world-protocol-impact — How this guidance changed actual trial designs
• postmarketing-dose-optimization-risk-signals — Consequences of non-compliance
• benefit-risk-dose-decision-framework — Practical tools for operationalizing the guidance