FDA 对眼科 503B 企业 Imprimis NJOF 签发 6 条 Form 483，含重复观察项

眼科针剂也敢"无杂质数据就放行"？Imprimis NJOF 6 条 483 揭开 503B 的质量裂缝

2026 年 8 月 18 日，FDA 三名检查员结束对 Imprimis NJOF LLC 为期十天的现场检查，签发了一份只有6 条观察项的 Form 483。条数不多，分量极重：观察项 1 和观察项 2 的部分内容，与 2024 年 4 月那次检查一字不差地重复。换句话说：两年过去了，窟窿还在原地。

互动 一家给眼科手术供药的 503B，竟有 33 个产品"没做杂质检测就放行"——这要是进到眼睛里，你敢想后果吗？

一、背景：给几百万只眼睛供药的"眼科 503B 头部"，带着旧账迎检

这家公司是做什么的。Imprimis NJOF LLC，注册地址在新泽西州 Ledgewood 1705 Route 46 Ste 6B，FDA 登记类别是"外包设施"（Outsourcing Facility）——也就是按《联邦食品药品化妆品法》503B 条款运营的无菌配制企业，可以批量生产、跨州销售"实质上是仿制制剂"的配制药品。它是上市公司 Harrow Health（NASDAQ: HROW）的全资子公司，自我定位为"全美领先的眼科专用 503B"，官方口径服务超过 300 万只眼睛。产品线几乎覆盖眼科手术全流程：莫西沙星、地塞米松/莫西沙星复方、TriMoxi+（曲安奈德/莫西沙星）、地塞米松/莫西沙星/酮咯酸"DMK"复方、阿托品、复方托吡卡胺/去氧肾上腺素，以及仿 Vigamox 等畅销眼药水的单剂量术中注射液。这类产品直接进入前房或玻璃体腔，质量容差天然比普通注射剂更窄。

签字的是谁。483 发给制造运营高级副总裁兼总经理 Sanjay S. Samudre。签署人共三名检查员，来自新泽西 Parsippany 办公室（10 Waterview Blvd, 3rd Floor）：Annet R. Rajan、Simone E. Pitts、Lucila B. Nwatu，均为 FDA 调查员（Investigator）。三人在 21 页纸上的落款签发日期均为 2026 年 8 月 18 日。从行文看，这支队伍的核心打法是"实验室与验证深挖"——放行标准、稳定性方案、灭菌装载鉴定、烟雾研究逐页翻证据，还大量引用企业自己的 SOP 条款号（如 VV-QUAL-00019 第 6.14、6.25、6.27 步）来证明"你连自己写的都没做到"。

旧账没还。上一次 FDA 检查是 2024 年 4 月，同样签发了 483，案件至今未关闭。而这次 6 条里，观察项 1（杂质控制缺失）和观察项 2 的部分内容（干预后不擦拭关闭）被明确标注"This is a repeat observation from the April 2024 inspection"。503B 领域对重复观察项的处理向来最严——它直接证明 CAPA 没有触达根因。

这次检查怎么跑的。检查日期为 2026 年 8 月 4 日至 8 月 18 日，共 10 个工作日（4/5/6/7/10/11/12/13/14/18 日），几乎两个整周。检查员在现场看的是真枪实弹：8 月 4 日观看了人员培养基灌装资质确认（Lot #26JUL009），8 月 10 日观看了莫西沙星单剂量术中注射液（批号 26JUN091）的整线灌装。也就是说，行为层面的缺陷（动作过快、干预后不擦拭、袋装胶塞遮挡 first air）都是实时目击记录，不是翻文件翻出来的。

二、逐条拆解：6 条，英文原文 + 中文还原

（以下每条中文表述均以 483 原文为据，并按上下文补全原文未言明之处以便阅读；(b)(4) 为 FDA 官方脱敏处，照原文保留；依上下文推测的内容以「（推断）」标出，非原文内容。）

观察项 1｜放行与稳定性标准缺乏科学依据，33 个产品无杂质数据放行【2024 年重复项】

Laboratory controls do not include the establishment of scientifically sound and appropriate specifications and test procedures designed to assure that components and drug products conform to appropriate standards of identity, strength, quality and purity. Specifically, Release and stability specifications for (b)(4), potency, particle size distribution, uniformity of dose, and impurities for multiple sterile ophthalmic drug products are inadequate in the following respects: A. (b)(4): 1. (b)(4) Lack scientific basis for Finished Product Release Specifications. Your firm was unable to provide scientific data or justification demonstrating that a (b)(4) release specification is appropriate for the sterile ophthalmic drug products listed below. The applicable compendial standards for the individual active ingredients contained in these products establish (b)(4) ranges that reflect the established chemical stability and ocular safety profiles of those ingredients. Your firm's specification extends beyond those established ranges on both the lower and/or upper ends, permitting release of drug product at (b)(4) values that would not conform to applicable compendial standards. For example: DL003- Moxifloxacin Preservative-Free Solution Single-Use Intraocular Injection (4 mg/0.8 mL (5 mg/mL)); DL006- Dexamethasone Sodium Phosphate/Moxifloxacin Preservative-Free Solution Single-Use Intraocular Injection (1 mg/mL and 5 mg/mL); DL011, DL017 (Vigamox copy), LD063/LD064- Tropicamide/Phenylephrine HCl (M2) Sterile Ophthalmic Solution; LD065, LD066, LD067- Atropine Sulfate Preservative-Free Sterile Ophthalmic Solution (0.01%) (0.025%) (0.05%); LD075- Dorzolamide Preservative-Free Sterile Ophthalmic Solution (2%). 2. (b)(4) Stability specifications contain no acceptance criteria. Section 6.2.3.9 of SOP VV-QUAL-00475, Stability Program, Version 6.0, Effective Date 03/13/2026 states that "Stability Studies protocols shall contain the following, at a minimum: acceptance criteria (with or without OOT/OOE limits)." Your firm was unable to provide scientific justification for the absence of (b)(4) acceptance criteria in the stability program for all commercially available sterile drug products produced by your facility. No acceptance criterion, upper limit, lower limit, or range, is defined for (b)(4) at any stability timepoint. Because no acceptance criterion exists, no (b)(4) result obtained during stability testing can constitute a stability failure regardless of the value measured. As a result, your firm's stability program does not provide assurance that (b)(4) remains within an established specification range throughout the assigned BUD for any of the (b)(4) commercial drug products and cannot demonstrate that these drug products maintain their quality attributes with respect to chemical stability and product safety for the duration of their labeled storage period. B. Potency. Your firm's potency acceptance criteria for Proparacaine Hydrochloride in LD085 Povidone Iodine/Proparacaine HCl Multi-Use Ophthalmic Solution (1.25%, 0.5%) is (b)(4). The applicable compendial standard for Proparacaine Hydrochloride Ophthalmic Solution establishes a narrower potency range. Your firm's specification permits release of drug product at active ingredient concentrations that would not conform to the applicable compendial standard without triggering an out-of-specification investigation or batch rejection. Review of the annual stability study for LD085 revealed Proparacaine Hydrochloride potency values of 85.2–91.1% through (b)(4) of stability testing. These results fall below the applicable compendial lower limit for Proparacaine Hydrochloride Ophthalmic Solution and indicate that the drug product does not conform to the compendial standard throughout its assigned beyond-use date. Because your firm's stability acceptance criterion of (b)(4) is wider than the applicable compendial range, these failing results were not identified as out-of-specification, no OOS investigation was initiated, and the product continued to be distributed without evaluation of the impact of sub-compendial potency on patient safety and therapeutic adequacy. There are currently (b)(4) batches of distributed Povidone Iodine/Proparacaine HCl Multi-Use Ophthalmic Solution (1.25%, 0.5%) within expiry on the market, Lots 26FEB015 and 26JAN035. C. Particle Size Distribution - No Release or Stability Specifications Established. Your firm does not have written stability acceptance criteria for particle size distribution (PSD) for the following sterile ophthalmic suspension and injectable suspension drug products at any stability timepoint: DL016- TriMoxi+ (Triamcinolone Acetonide/Moxifloxacin Preservative Free Suspension Single-Use Intraocular Injection (9 mg/0.6 mL, 0.6 mg/0.6 mL)); LD050/LD051- Prednisolone Acetate/Moxifloxacin/Nepafenac; LD052; LD055; LD056; LD060/LD061; LD082. D. Uniformity of Dosage Unit. Your firm has not established acceptance criteria and does not perform Uniformity of Dosage Unit testing as part of release testing or stability studies for TriMoxi+. E. Impurities. There is a failure to establish an impurity profile for sterile injectable and topical ophthalmic drug products manufactured since the last inspection conducted in April 2024. Your firm currently does not test the majority of your drug products for impurities or degradants at release or during stability and has continued to release and distribute drug products without this data. Your firm has failed to identify, characterize, qualify, and establish acceptance criteria for process related impurities, degradation products, and/or elemental impurities potentially present in 33 of (b)(4) finished drug products manufactured and released for distribution. To date, your firm has not completed and implemented method validation for impurities and degradants for thirty-three (33) drug products manufactured and distributed from your facility.

实验室控制没有建立科学可靠、恰当的质量标准与检验程序，无法保证组分与药品符合鉴别、规格、质量与纯度的适当标准。缺陷分五块：

放行标准缺乏科学依据。企业拿不出任何科学数据或理由，证明它给多个无菌眼科产品设定的 (b)(4) 放行标准是恰当的——药典对各活性成分已有既定范围，反映其化学稳定性与眼部安全性，而企业的标准在上下限两端都超出了药典范围，等于允许“不符合药典”的产品出厂。涉及的产品包括：DL003 莫西沙星无防腐剂单剂量眼内注射液（4mg/0.8mL）、DL006 地塞米松磷酸钠/莫西沙星、DL011 与 DL017（仿 Vigamox）、LD063/LD064 复方托吡卡胺/去氧肾上腺素、LD065–067 阿托品（0.01%/0.025%/0.05%）、LD075 多佐胺 2% 等。

稳定性方案没有验收标准。企业自己的 SOP VV-QUAL-00475《稳定性程序》6.0 版第 6.2.3.9 节白纸黑字要求“稳定性研究方案至少应包含验收标准”；但所有在售无菌产品的稳定性方案里，(b)(4) 项目在任何时间点都没有定义验收标准、上限、下限或区间。结果是：稳定性测出什么值都不算不合格——程序根本无法证明产品在 BUD 内保持在既定范围内。

效价标准宽于药典，实测已经不合格。LD085 碘/丙美卡因复方案里，丙美卡因 HCl 的效价验收标准比药典宽。年度稳定性研究实测值 85.2%–91.1%，低于药典下限——正因为标准宽，这些结果没被判为 OOS，没启动调查，产品在未评估“效价不足对麻醉效果的影响”的情况下继续分销。目前市场上仍在效期内流通的有 (b)(4) 批（Lot 26FEB015 与 26JAN035）。

混悬剂没有粒径标准。TriMoxi+ 与 7 个泼尼松龙类混悬/注射混悬产品，在任何稳定性时间点都没有书面的粒径分布（PSD）验收标准。

剂量单位均匀度不做。TriMoxi+ 既没有验收标准，放行与稳定性研究中也不做该项检测，且给不出省略理由。

杂质谱整体缺失。自 2024 年 4 月上次检查以来，一直没有建立无菌注射与外用眼科产品的杂质谱；多数产品放行与稳定性都不测杂质、降解物，却在无数据的情况下持续放行分销；(b)(4) 个成品中，33 个未鉴别、表征、限定工艺相关杂质/降解物/元素杂质，33 个产品至今未完成杂质与降解物的方法验证。

FDA 的真实关切：本条是"实验室控制"的全景式失守——标准比药典宽就永远不会 OOS，稳定性"只监测不判合格"，33 个产品连杂质长什么样都不知道就敢放行。眼科注射剂入眼即生效，这些数据缺口没有一个有退路。法规：21 CFR 211.160(b)、211.165、211.166。

互动 "稳定性只监测、不判合格"——这不就等于永远不会不合格吗？

观察项 2｜无菌/灭菌工艺验证：最差点被排除、装载超限、干预不模拟、烟雾研究拍不全

Procedures designed to prevent microbiological contamination of drug products purporting to be sterile did not include adequate validation of the aseptic and sterilization process. Specifically, Part 1- (b)(4) Qualification: A. (b)(4), used to (b)(4) glass containers, bowls, and small metal tools for sterile compounding operations, has not been qualified for the (b)(4) shelf load configuration used during routine production. Your firm's Performance Qualification (PQ), QSR-VAL-2000-1766, was conducted using a (b)(4) shelf load configuration. Your firm's temperature mapping study identified the (b)(4) corner of the (b)(4) shelf on the (b)(4) door as the coldest location, the worst-case position least likely to achieve the required (b)(4) time and temperature. The PQ did not include this shelf position and therefore did not challenge the (b)(4) at the confirmed worst-case location under production-representative conditions. There is no data demonstrating (b)(4) achieves the required (b)(4) time and temperature at the coldest location when operated with (b)(4) shelf load. Glass containers (b)(4) in (b)(4) under the (b)(4) shelf configuration have been used in the manufacture and distribution of compounded sterile drug products, including LD050 - Prednisolone Acetate/Moxifloxacin/Nepafenac Sterile Ophthalmic Suspension (1% 0.5% 0.1%), (5ml Fill Volume), Lots 26MAR033, 26JUN076. B. Your firm's (b)(4) (Model (b)(4)) is routinely operated with load quantities that exceed the maximum load established during qualification. (b)(4) Performance Qualification QVP-VAL-1000-1939 qualified a maximum load of (b)(4) glass bottles; however, a review of the (b)(4) cycle logbook identified the following production cycles exceeding the qualified maximum: 03/24/2026 #03242601, 04/01/2026 #04012601, 04/15/2026 #04152602, 06/01/2026 #06012603, 06/11/2026 #06112602, 06/23/2026 #06232603. Operating (b)(4) beyond the qualified load increases the (b)(4) mass within the (b)(4), reduces airflow, and may prevent the coldest location from achieving the minimum validated (b)(4) conditions of (b)(4) required to demonstrate a (b)(4)-log reduction in endotoxin. There is no data demonstrating that the (b)(4) cycle achieves the required endotoxin reduction under these overloaded conditions. Part 2- Media Fill and (b)(4) Assessment: A. Your firm's aseptic process simulation (media fill) for the (b)(4) filling line does not simulate the most challenging and stressful conditions of actual drug production. When filling operations on the (b)(4) extend (b)(4), your firm's production procedure requires that the (b)(4), the bulk drug solution glass bottle and the (b)(4) for (b)(4) prior to resuming filling. This (b)(4) step, performed after (b)(4) hold of product in the filling line, is a routine component of your firm's actual production process. Your firm's media fill protocol, as documented in Media Fill on (b)(4) lot #26MAR037, dated 3/13/2026, did not include the (b)(4) step. As a result, the media fill does not simulate the conditions under which drug products are routinely aseptically filled. B. Your firm has not evaluated the risk of microbial contamination associated with product remaining in the (b)(4) of the (b)(4) filling line during production pauses or (b)(4) holds and has not established controls to address that risk prior to resuming filling operations. During the Media Fill on 03/13/2026, the filling operation carried over to (b)(4), during which the filling operations were paused for approximately (b)(4). The filling continued without (b)(4) or (b)(4) the product remaining in the (b)(4) of the (b)(4), nor did it reflect the (b)(4) step. Your firm was unable to provide a risk assessment, validation data, or scientific justification. The (b)(4) filling machine is used in the manufacture of TriMoxi+ (Triamcinolone Acetonide/Moxifloxacin Preservative Free Injectable Suspension (9 mg/0.6 mL, 0.6 mg/0.6 mL)). Part 3: Airflow Visualization and Aseptic Processes: A. Airflow visualization (smoke) studies, conducted in the (b)(4) ISO 5/Class 100 cleanroom dated 30 SEP 2025, did not demonstrate adequate evaluation of airflow behavior at all critical locations. The airflow study recording of the Plunger, Rails, and Bowls Set-up failed to include the complete range of camera angles necessary to evaluate whether the smoke pattern generated indicates turbulence, eddies, or non-unidirectional airflow at the HEPA filter faces. B. During filling operations for Moxifloxacin Preservative-Free Solution Single-Use Intraocular Injection, Lot #26JUN091 on the (b)(4) vial filling line on 08/10/2026, we observed: 1. The sterile and support technician did not adhere to the principle of slow and deliberate movements required inside the cleanroom as per the firm's procedure # VV-QUAL-00019 "Aseptic Behavior Inside the Cleanrooms" Step 6.14 which states that "the principle of slow, careful movement should be followed at all times inside the cleanroom". We observed the technicians make rapid movements in the cleanroom and while performing interventions such as loading 2 mL vials into the infeed tray under the ISO 5 (b)(4). 2. We observed instances in which the sterile technician did not wipe down the ISO 5 (b)(4) after opening them to perform interventions, exposing the interior to the ISO 7 filling room environment, as required by Step 6.27 which states "When the intervention is complete, wipe the (b)(4) door before closing with (b)(4) wipes". Specifically: a. At approximately 3:55 P.M., the sterile technician opened the (b)(4) close to the stopper bowl and introduced a bag of stoppers into the ISO 5 environment. After placing the bag inside, the operator did not wipe the interior surface prior to closing. b. At approximately 3:58 P.M., the sterile technician opened (b)(4) simultaneously close to the stopper bowl and inserted both (b)(4) into the ISO 5 enclosure. Using sterile scissors, the technician opened the bag of stoppers and filled the stopper bowl. Upon completing the intervention and exiting, the technician did not wipe the interior surfaces prior to closing. ***This is a repeat observation from the April 2024 inspection*** 3. The interior surface of the vial exit tray door of the ISO 5 (b)(4) was observed to have scratches, rendering the surface non-smooth and not easily cleanable. C. During the personnel media fill qualification (Lot# 26JUL009) on the (b)(4) bottle filling machine on 08/04/2026, on two separate occasions during the media fill operations, we observed an operator manually loading a bag of sterile caps into the (b)(4) bowl located within the ISO 5 (b)(4), during which the bag was held directly above the caps already present in the bowl, blocking first air to those components. This practice is not in accordance with Step 6.25 of the firm's procedure VV-QUAL-00019 "Aseptic Behavior Inside the Cleanrooms" which states "Proper aseptic manipulation should be approached from the side and not above the product or sterile components". Part 4: Deficient Environmental Monitoring Program. The firm's environmental monitoring (EM) program for the (b)(4) bottle filling line and (b)(4) vial filling line is inadequate. Specifically: A. The frequency of Active Air Sampling on the (b)(4) bottle filling line is insufficient. B. The frequency of Active Air Sampling on the (b)(4) Vial Filling Line during filling operations is insufficient (e.g., Lot# 26JUN114). This frequency is not scientifically justified to provide adequate assurance of environmental control or to optimize detection of contamination events throughout the full duration of a filling operation. C. Insufficient Surface Sampling: Per Step 6.8.1.1 of procedure VV-QUAL-00252 "Environmental Monitoring Program-New Jersey Facility," surface samples are collected (b)(4) from the direct compounding areas of the filling lines after each product fill. However, the following surfaces are not sampled despite being locations with which sterile technicians routinely interact during filling operations: (b)(4) Bottle Filling Machine (ISO 5): Bottle tip bowl and cap bowl; (b)(4) Vial Filling Machine: Vial crimp bowl and stopper bowl. Although the firm completed Environmental Monitoring Performance Qualification (EMPQ) which included recommendations to increase active air sampling frequency and expand sampling locations, the firm failed to conduct a risk assessment to evaluate the impact of the current, inadequate sampling plan on drug products filled on the respective lines.

为防止“声称无菌”的药品受微生物污染而设计的程序，未包含对无菌与灭菌工艺的充分验证。缺陷分四部分：

第 1 部分——(b)(4) 灭菌设备的鉴定。A：处理玻璃容器、器皿与小金属工具的 (b)(4)，从未按生产实际使用的 (b)(4) 层架装载状态做过鉴定。性能确认 PQ（QSR-VAL-2000-1766）用的是另一种装载；而企业自己的温度分布研究已确认，靠 (b)(4) 门那层架的 (b)(4) 角是最冷点、最难达标。PQ 恰恰跳过了这个位置——最差点从未在生产代表性条件下被挑战过。该状态下处理的玻璃容器已用于生产分销（如 LD050，批号 26MAR033、26JUN076）。B：设备日常装载量超出鉴定确认的最大值（PQ QVP-VAL-1000-1939 确认上限 (b)(4) 瓶），周期日志里至少 6 个生产周期超限（2026/3/24、4/1、4/15、6/1、6/11、6/23）。超载会增加负荷、减少气流，最冷点可能达不到证明内毒素 (b)(4) 个对数级下降所需的条件——而超载状态下没有任何数据。

第 2 部分——培养基灌装与 (b)(4) 评估。实际生产中，灌装超过 (b)(4) 时要对灌装线内的药液瓶与 (b)(4) 做 (b)(4) 处理再继续，这是常态工序；但 2026 年 3 月 13 日的培养基灌装方案（Lot #26MAR037）没有包含这一步。同样，产品滞留在灌装线 (b)(4) 中的污染风险、以及恢复灌装前的控制措施，企业既没评估也没验证——3 月 13 日那次介质填充跨入 (b)(4)、暂停约 (b)(4) 后直接继续灌装，既未处理滞留产品也未复现 (b)(4) 步骤，还拿不出风险评估或验证数据。这台 (b)(4) 灌装机用于生产 TriMoxi+。

第 3 部分——气流可视化与无菌行为。2025 年 9 月 30 日在 ISO 5/百级洁净间做的烟雾研究，没有覆盖判断 HEPA 面湍流、涡流或非单向流所需的完整机位。2026 年 8 月 10 日莫西沙星单剂量眼内注射液（批号 26JUN091）灌装现场，检查员看到：技术员动作飞快，违反自家规程 VV-QUAL-00019 第 6.14 步“任何时候都应缓慢小心”；打开 (b)(4) 干预后不擦拭内表面就关闭，把 ISO 5 内部直接暴露给 ISO 7 环境（下午 3:55 放胶塞袋、3:58 剪开倒入胶塞碗，两次都没擦）——这一条是 2024 年 4 月检查的重复项；出口托盘门内表面还有划痕，不再平滑、不易清洁。8 月 4 日的人员培养基灌装资质确认（Lot #26JUL009）中，又两次看到操作员把整袋无菌盖悬在碗里已有盖子的正上方倾倒，挡住了这些组件的 first air（第一股层流空气），违反第 6.25 步“应从侧面接近，不能在产品或无菌组件正上方操作”。

第 4 部分——环境监测。(b)(4) 塑瓶线与 (b)(4) 西林瓶线的主动空气采样频次都不足（西林线如批号 26JUN114），频次没有科学依据，覆盖不了一场完整灌装。表面采样也有漏洞：按企业自己的规程 VV-QUAL-00252，每次灌装后要采直接配制区，但操作员每次都要碰的瓶尖碗、瓶盖碗、轧盖碗、胶塞碗从来不采。企业做完 EMPQ、报告里明确建议提高频次、扩展点位，却始终没做风险评估。

FDA 的真实关切：这是一条完整的验证失效链——PQ 排除自己找到的最冷点、装载超限照样跑、介质填充跳过常态步骤、烟雾研究缺机位、现场行为违规、EM 采样避重就轻，而且"干预后不擦拭"还是 2024 年的重复项。法规：21 CFR 211.113(b)、211.111；USP <797>/<1207>/<1116>。

观察项 3｜调查不彻底：纤维进了针剂找不到根因，无菌失败说作废就作废

There is a failure to thoroughly review any unexplained discrepancy and the failure of a batch or any of its components to meet any of its specifications whether or not the batch has been already distributed. Specifically, A. Your firm failed to conduct an adequate investigation to determine the root cause of cellulose, non-cellulose, and other fiber-like material that have been observed in the sterile injectable finished drug products containing API Moxifloxacin Hydrochloride bulk powder from 2023 until present day. The firm's root cause report VV-QUAL-02496 "Report for the presence of natural cellulose fibers in DM, DMK and Moxi 5" (Approved: 05/06/2025) was initiated to address a trend of an increase in the level of fibers detected in several injection products. The firm attributed the most probable root cause of the presence of natural cellulose fibers to the API Moxifloxacin, specifically the use of the bulk drug substance; however, the exact mechanism of fiber generation from use of Moxifloxacin bulk drug substance has not been identified. The firm has not initiated an investigation with the current Moxifloxacin supplier to assess any changes to the manufacturing process that could potentially contribute to the fibers. The firm has initiated change controls to use a commercially available ophthalmic solution containing Moxifloxacin from a new supplier; however, your firm continues to manufacture ophthalmic drug products using the Moxifloxacin API bulk powder from the existing supplier. The firm continues to identify fibers in sterile ophthalmic drug products; for example: 1. QE-000873 (Occurred: 11/11/2025): Moxifloxacin PF Intraocular Injection Lot# 25OCT045 (Rejected) failed for exceeding the critical defect limit, critical defects included (b)(4) vials with fibers. 2. QE-000908 (Occurred: 12/09/2025): Lot# 25NOV001 (Rejected), critical defects included (b)(4) vials with fibers. 3. QE-000912 (Occurred: 12/11/2025): Lot# 25NOV012 (Rejected). 4. QE-000932 (Occurred: 12/30/2025): Lot# 25DEC002 (Rejected). B. Deviation investigation (QE-000458) into sterility failure of Dexamethasone Sodium Phosphate/Moxifloxacin/Ketorolac Tromethamine (DMK), Stability Batch #23NOV035, Time Point: Day (b)(4) (Date Occurred: 12/24/2024) is deficient. The third-party laboratory reported that all (b)(4) media enrichments from USP (b)(4) sterility test had mold in it during an intermediary read of DMK batch 23NOV035. The third-party investigation concluded that (b)(4) used to wash the (b)(4) was the source of contamination. However, the specific lot of (b)(4) used during the original test was no longer available for investigation, and (b)(4) and the negative control, both of which were rinsed with the same batch of (b)(4), did not have mold growth. Upon the firm's request, the third-party conducted (b)(4) microbiological test on retain vials of this product and no fungal contaminants were noted and invalidated the original sterility test result. The investigation did not identify an obvious, documented root cause assignable to laboratory error or equipment failure before proceeding to invalidate the original sterility test results and no documented scientific justification was provided for the use of (b)(4) results as a basis to invalidate the results of the compendial USP (b)(4) Sterility Test. C. The Corrective and Preventative Actions (CAPA) implemented are not commensurate with the scope of the identified root causes: 1. (b)(4) separate (b)(4) test failure investigations were conducted for (b)(4) of the same lot number, each of which returned an "Error Leak Test Failure" message across multiple repeat test attempts. The manufacturer's investigation determined that the (b)(4) were non-integral and exhibited damage consistent with exposure to excessive forward differential pressure. The root cause was identified as method due to the firm's internal procedure allowing (b)(4) supply up to (b)(4) psi while the (b)(4) maximum rate forward differential pressure is (b)(4) psi; root cause for the deviation was due to over-pressurization during the post use (b)(4) test. The CAPA did not include an assessment to update the internal procedure to ensure that pressure applied during (b)(4) testing does not exceed the manufacturer's validated maximum. 2. Deviation Investigation QE-000985 was conducted on 02/13/2026 following a failed (b)(4) test used to (b)(4) sterilize Povidone Iodine/Proparacaine Sterile Ophthalmic Solution (1.25%, 0.5%), Lot 26JAN035. Five (b)(4) tests yielded a result of "Leak Test Failure". The manufacturer concluded the observed failures were attributable to inadequate (b)(4) due to (b)(4) rather than a loss of (b)(4), and recommended validating product (b)(4) test parameters for the product. However, the firm's CAPA was limited to a vendor-conducted training session on (b)(4) use and handling requirements. D. Your firm failed to conduct an adequate investigation to determine the root cause of the recovery of Bacillus cabrialesii from a passive settle plate located near the unscrambler of the (b)(4) filling machine in the ISO 5 environment. The action level specification is greater than or equal to (b)(4) CFU, thereby exceeding the specification limit. Deviation report QE-000684 described the recovery of (b)(4) CFU of Bacillus cabrialesii, a spore-forming organism, from the passive plate during the filling of Bimatoprost 0.01% Preservative-Free Sterile Ophthalmic Solution, Lot #25MAY040 on July 25, 2025. The investigation failed to establish a definitive root cause, assess the potential source and the significance of the organism to adequately justify the release of the batch.

无论批次是否已分销，企业都未能彻底审查任何无法解释的差异，以及批次或其组分不符合质量标准的情况。四个具体例子：

A. 纤维查了两年没查出根因。从 2023 年至今，含莫西沙星 HCl 原料粉的无菌注射剂里反复发现纤维素、非纤维素及其他纤维状物质。根因报告 VV-QUAL-02496（2025/5/6 批准）把最可能根因归给莫西沙星原料药本身，但纤维产生的确切机制从未查明；企业也没和现行供应商启动过调查，评估对方工艺变化是否可能带来纤维。企业倒是启动了“换用新供应商成品眼用溶液”的变更控制，但同时仍继续用老供应商的原料粉生产。纤维仍在检出：QE-000873（2025/11/11，25OCT045）、QE-000908（12/9，25NOV001）、QE-000912（12/11，25NOV012）、QE-000932（12/30，25DEC002）——四个批次全部因纤维超关键缺陷限度拒收，根因一个都没查清。

B. 无菌失败说作废就作废。DMK 稳定性批次 #23NOV035 在中途读板时发现：USP (b)(4) 无菌检查的全部 (b)(4) 培养基增菌物里都有霉菌。第三方调查认定是洗涤 (b)(4) 用的 (b)(4) 被污染——但原始测试那批 (b)(4) 已不在现场，而用同一批 (b)(4) 冲洗的 (b)(4) 与阴性对照都没长霉。企业随后请第三方对留样瓶做 (b)(4) 微生物检测，未见真菌，于是作废了原始无菌结果。问题在于：作废之前并没有找到可归因于实验室错误或设备故障的明确根因，也没有任何书面科学依据支持用 (b)(4) 结果去推翻药典 USP (b)(4) 无菌检查。

C. CAPA 与根因不相称。其一：同一批号产品的 (b)(4) 泄漏测试反复报“Error Leak Test Failure”，三次调查都把根因认定为“方法”——内部规程允许的供气压力 (b)(4) psi 超过了器件的最大正向压差 (b)(4) psi，属使用后测试超压；但 CAPA 里没有任何“修改规程、确保压力不超厂商验证上限”的动作。其二：2026/2/13 的 QE-000985，碘/丙美卡因（Lot 26JAN035）的 (b)(4) 测试五次报“Leak Test Failure”，厂商认定是产品导致的 (b)(4) 问题而非器件失效，并建议为该产品验证专用测试参数——而企业的 CAPA 只安排了一场供应商培训。

D. ISO 5 长芽孢菌，没根因就放行。2025 年 7 月 25 日灌装比马前列素 0.01%（Lot #25MAY040）时，ISO 5 环境里 (b)(4) 灌装机理瓶机旁的被动沉降碟检出芽孢形成菌 Bacillus cabrialesii (b)(4) CFU，超出行动限。调查既没确定根因，也没评估来源和该菌的意义，就这样支持了批次的放行。

FDA 的真实关切：纤维反复进针剂、无菌失败轻易作废、泄漏测试超压三年不改规程、ISO 5 长芽孢菌照样放行——调查体系失效的直接后果是同一类问题反复发生。法规：21 CFR 211.192、211.180(e)。

观察项 4｜"测到合格"的灯检程序 + 缺失的 (b)(4) 相容性研究

Your firm failed to establish written procedures for production and process controls designed to assure that the drug products have the identity, strength, purity, and quality that they are purported or represented to possess. Specifically, Part A: Visual Inspection Automatic Reinspection. Your firm's procedure VV-QUAL-00606, (b)(4) Visual Inspection, Version 24.0, Effective Date: 05/12/2026, specifies that when a batch fails 100% visual inspection by exceeding action levels for critical or major defects, a deviation is initiated, and a (b)(4) inspection of the remaining accepted units is conducted in tandem with the deviation investigation. Your firm's procedure permits a batch to proceed to (b)(4) inspection and, if passing, to tightened AQL inspection and potential release, before the deviation investigation establishing the root cause of the initial 100% inspection failure has been completed. Proceeding to additional inspection passes prior to completing a thorough investigation of the initial failure does not constitute a thorough review of the batch failure and represents testing into compliance. Part B: Lack of (b)(4) Compatibility Testing. The firm does not have a procedure governing the process for (b)(4) compatibility studies and has not conducted (b)(4) compatibility studies for all finished drug products that are (b)(4) sterilized to evaluate the effect of the sterilizing (b)(4) on the final product. For example, during the Process Performance Qualification (PPQ) of Povidone Iodine/Proparacaine Sterile Ophthalmic Solution (1.25%, 0.5%) (VV-QUAL-02204), bulk solution samples were collected from (b)(4) of the mixing vessel (b)(4) and were tested for potency; however, no (b)(4) compatibility studies such as extractables/leachable studies or chemical compatibility studies were conducted.

企业没有建立“确保药品具备其声称的鉴别、规格、纯度与质量”的书面生产与工艺控制程序，两处具体失守：

A. 灯检失败不查根因就“测到合格”。规程 VV-QUAL-00606（(b)(4) 灯检，24.0 版，2026/5/12 生效）规定：一批做 100% 灯检时若关键或主要缺陷超行动限，就启动偏差，并对其余合格品做 (b)(4) 复检。问题在于——规程允许在“查清初始失败根因”的偏差调查完成之前，就让这批产品进入复检，复检过了就走加严 AQL 并可能放行。在完成彻底调查前追加多轮复检，FDA 直接定性为testing into compliance（测到合格）。

B. 灭菌相容性研究缺失。企业没有管理 (b)(4) 相容性研究的规程，也没有对所有经 (b)(4) 灭菌的成品做过相容性研究，去评估灭菌 (b)(4) 对最终产品的影响。碘/丙美卡因的 PPQ（VV-QUAL-02204）只从混合罐 (b)(4) 取样测了效价，没有任何可提取物/浸出物研究或化学相容性研究。

FDA 的真实关切：灯检失败不查根因就多测几轮直到过线，这是检查里最典型的"程序性作弊"；灭菌方式对产品的影响则从未被评估。法规：21 CFR 211.100、211.94。

观察项 5｜一扇关不上的门，开了十年

Equipment used in the manufacture, processing, packing or holding of drug products is not of appropriate design to facilitate operations for its intended use. Specifically, The firm failed to conduct an airflow visualization study (smoke study), either under static or dynamic conditions with the door in its open position, to demonstrate proper unidirectional airflow is maintained during the filling operations of sterile manufactured ophthalmic drug products in the (b)(4). The exit door on the (b)(4) remains open, the door does not fully close, remaining ajar at an approximate 45-degree angle during filling operations. This design deficiency allows for the receiving tray of vials (capped and sealed) to protrude out, preventing the exit door from fully closing during filling operations thereby, exposing the ISO 5 classified sterile core to potential ingress of ISO 7 classified airflow. In addition, the firm has not conducted a risk assessment to evaluate the impact of personnel exiting the ISO 7 classified area and the potential of ingress/blow back of ISO 7 classified air into the ISO 5 sterile (b)(4) during filling operations. The firm has not evaluated when personnel exit the ISO 7 filling room through the room door if there is a disruption in the room pressure differential that can cause a momentary ingress of unclassified air into the ISO 7 room. Because the (b)(4) exit door remains open during filling, this pressure disruption in the ISO 7 room may propagate into the ISO 5 (b)(4), further compromising the integrity of the critical filling environment. The firm has not evaluated this risk through a smoke study, risk assessment, or pressure differential monitoring under dynamic conditions that include personnel movement. The (b)(4) has been in use since October 2016 with this design and is used to fill multiple sterile drug products specifically the product line of Moxifloxacin and other products such as TriMoxi+.

灌装用 (b)(4) 设备的出口门关不上——灌装作业期间它始终以约 45 度角半开，因为已压塞封口的成品瓶接收托盘向外凸出，顶住了门。企业从未做过“门处于开启位置”的静态或动态烟雾研究，来证明 ISO 5 内仍保持正常单向流；于是 ISO 5 无菌核心区就这样暴露在 ISO 7 气流可能倒灌的风险里。更麻烦的是，企业也没评估过“人员进出 ISO 7 时房间压差被扰动、未分级空气瞬间涌入”的连锁效应——门常开时，这种扰动会一路传导进 ISO 5。这些风险，烟雾研究、风险评估、含人员移动的动态压差监测，一样都没做。而这台 (b)(4) 设备自 2016 年 10 月起就带着这个设计在用，用于灌装莫西沙星产品线与 TriMoxi+ 等多个无菌产品。

FDA 的真实关切：设备缺陷存在近十年、从未被任何评估工具覆盖——设计问题背后是"没人提出问题"的质量文化。法规：21 CFR 211.63。

互动 一扇关不上的门，竟用了十年——是没想到，还是没人管？

观察项 6｜质量部门：批批放行、笔笔签字，就是不行使否决权

The quality control unit lacks the responsibility and authority to approve and reject all components, in-process materials and drug products. Specifically, Your firm's Quality Control Unit (QCU) has failed to exercise its responsibility to approve or reject drug products, procedures, and specifications, and to ensure that all established written procedures are followed, as evidenced by the following examples: A. The quality control unit approved and released drug products for distribution without impurity and degradant testing having been performed, despite the absence of validated test methods for the vast majority of products manufactured by the firm. With the exception of one product, [PMB Solution, relaunched July 2026], no validated impurity or degradant test method exists for any drug product manufactured and distributed by the firm. The QCU released and approved these batches for distribution without this data, and without documented justification for the omission of a required quality attribute from the release testing program. B. The quality unit had the responsibility to ensure the Performance Qualification of the (b)(4) was scientifically sound and that all acceptance criteria were evaluated at the appropriate locations before approving the study. Approving a PQ that structurally excluded the confirmed worst-case location identified by the firm's own temperature mapping represents a failure of the quality unit's review and approval function. The quality unit approved a qualification that, on its face, did not demonstrate what it purported to demonstrate. C. The quality unit is responsible for reviewing production records and ensuring that manufacturing operations are performed within qualified and approved parameters. The cycle logbook entries showing (b)(4) bottles are facially inconsistent with the (b)(4) bottle qualified maximum - a discrepancy that should have been identified during routine QC review of cycle logs. The quality unit failed to exercise its responsibility and authority to reject drug product containers processed outside of qualified conditions. D. The Quality Unit did not implement Corrective Action and Preventative Actions that were commensurate with the root causes identified in Deviation Investigations QE-000814, QE-000849, QE-000901 related to critical (b)(4) test failures.

质量控制部门（QCU）没有真正行使批准与拒收所有组分、中间物料和药品的职责与权限，四个例子：

A. 无杂质数据照样放行。除一个产品（PMB 溶液，2026 年 7 月重新上市）外，企业生产分销的任何产品都没有经验证的杂质或降解物检测方法。QCU 在既无数据、也无“为何省略这一必备质量属性”的书面理由的情况下，批准并放行了这些批次。

B. 批准了字面上就站不住的 PQ。质量部门有责任在批准前确认 PQ 科学可靠、所有验收标准都在恰当位置评估过。批准一份“结构性排除了自家温度分布研究确认的最冷点”的 PQ，是审核批准职能的失职——它批准的那份确认文件，字面上就证明不了它声称要证明的东西。

C. 超限日志漏审。质量部门负责审核生产记录、确保生产在已确认参数内进行。周期日志里 (b)(4) 瓶的记录与 (b)(4) 瓶的确认最大值明显矛盾——这种不一致本应在日常 QC 审核周期日志时被发现；质量部门没有行使拒收“在超出确认条件下处理的产品容器”的职责。

D. CAPA 不到位。对 QE-000814、QE-000849、QE-000901 三起关键 (b)(4) 测试失败调查所识别的根因，质量部门没有实施与之相称的 CAPA。

FDA 的真实关切：把观察项 1–5 的所有问题串起来的是这一条——放行无数据、批准坏文件、漏审超限日志、CAPA 敷衍，质量部门签字签了，把关没有做。这正是 2024 年缺陷重复出现的根源。法规：21 CFR 211.22。

三、主线收敛

6 条浓缩成一句："质量部门放弃了把关，连 2024 年的旧账都没还。"标准比药典宽（1）、验证漏最差点（2）、调查无根因（3）、灯检测到合格（4）、门常开十年（5）——这五件事每一件都要经过质量部门的签字；而第 6 条 FDA 干脆点名：你们签了字，但没有行使职权。眼科针剂入眼即生效，这种溃散格外危险。

四、严重度一览

五、把这份 483 放回执法链条上看

这份 483 的杀伤力不在条数，而在证据形态：21 页纸里，FDA 大量引用企业自己的 SOP 条款号、PQ 报告编号、偏差编号和批次号，证明"你连自己写的都没做到"——这类证据几乎无法用文书回复化解。更关键的是两条 2024 年重复项：监管逻辑上，重复观察项是"CAPA 无效"的最直接证明，也是从 483 升级到警告信的高频导火索。叠加"33 个产品无杂质数据放行"这种量化事实，以及眼科产品直接入眼的敏感性，这份 483 在执法链条上的位置相当靠前。放在本批检查序列里看，它与同期的 Olympia、RC Outsourcing 同属 503B 序列，共同短板是"无菌保证的验证与执行双缺"；Imprimis 的特殊性在于还叠加了一层实验室控制缺口——不只是"无菌可能不保"，而且是"里面有什么杂质你根本不知道"。

当然，Form 483 本身不是最终判定。企业仍有 15 个工作日的回复窗口与后续申诉程序，FDA 也尚未公布分类结论——以上是风险方向，不是判决书。

互动 两年时间、一份新 483，旧账原样躺在那里——你认为问题出在整改本身，还是出在验收整改的人？

声明：本文基于美国 FDA 公开的执法文书（Form 483 / Warning Letter / Import Alert 等）撰写，仅供医药行业质量管理与合规学习交流之用，不构成对任何企业的法律或商业评价。原始文书以 FDA 官网公布版本为准；文书内容不构成 FDA 的最终判定。