毒理学关注阈值怎么用

结论：关注阈值是处理数据缺口的工具，不是免检通道

化学表征之后常遇到这个问题：检出了某种物质，但查不到它的毒理学数据，无法直接判断风险。

关注阈值提供的思路是：如果某种物质的暴露量低到一定程度，即使不知道它具体是什么，风险也可以认为是可接受的。 这个思路建立在对大量已知物质的统计分析之上。

但它有明确的适用前提和排除情形。用在不适用的场合，结论不成立，而这类问题在审查时会被直接指出。

适用的前提

使用关注阈值的基本前提包括：

物质已被检出并定量。 阈值比较的是暴露量，没有量就无从比较。

物质不属于排除类别。 某些高毒性或特殊机理的物质类别被明确排除在阈值适用范围之外。

暴露途径匹配。 阈值的推导基于特定的暴露途径，用在其他途径上要论证。

暴露量的计算合理。 从浸提浓度到人体暴露量的换算，假设要明确且保守。

第二条是使用时最需要核对的。 排除类别通常包括具有特定结构警示的物质、已知的高毒性类别等。检出物质如果落在排除类别里，必须查找具体的毒理学数据，不能用阈值处理。

分级的依据

关注阈值通常按物质的结构特征分级，不同级别对应不同的限值：

暴露时间这一项在器械评价中很关键。 短期接触的产品与长期植入的产品，适用的限值差别很大。判定接触时间时要按实际使用方式，并留有保守余量。

使用的步骤

第一步，确认物质身份。 至少要知道它属于哪一类，才能判断是否落在排除类别里。完全未知结构的物质，使用阈值要更谨慎。

第二步，检查是否属于排除类别。 落在排除类别的，必须找具体数据。

第三步，确定适用的限值。 按接触时间和途径选择。

第四步，计算暴露量。 从检出浓度出发，按产品的接触方式、接触面积、接触时间换算成日暴露量。这一步的假设要写清楚。

第五步，比较并得出判断。 暴露量低于限值的，可以认为风险可接受；高于限值的，需要进一步的数据或措施。

第六步，记录论证过程。 每一步的依据和假设都要可追溯。

暴露量计算的常见错误

这一步出错最多，列几个常见问题：

直接用浸提浓度当暴露量。 浸提条件与实际使用条件不同，必须换算。

忽略接触面积的差异。 试样的面积与实际接触面积不同。

接触时间估计过短。 应当按实际使用的累计时间，并考虑重复使用。

未考虑多个部件的叠加。 同一种物质可能来自多个部件，暴露量要累加。

体重假设不合理。 使用者包括儿童时，体重假设要相应调整。

第四项在结构复杂的产品上容易遗漏，尤其是同一种添加剂用在多个部件上的情况。

不适用的情形

以下情形不宜用关注阈值处理：

物质属于明确的排除类别；暴露量高于适用限值；产品用于特殊人群且限值未作相应调整；物质身份完全未知且无法归类；以及监管方明确要求提供具体数据的情形。

遇到这些情形，可行的路径是： 查找或委托做该物质的具体毒理学数据；或者通过配方或工艺调整降低该物质的释放量；或者改变材料。

「降低释放量」往往是最实际的路径。 比如增加清洗工序、调整固化条件、更换添加剂，都可能把暴露量降到阈值以下。

与整体评价的关系

关注阈值只是生物学评价中的一个工具，它处理的是「有物质检出但无数据」这个特定问题。整体评价还需要：

完整的材料信息；接触性质与时间的判定；已有数据的利用；必要的生物学试验；以及综合的风险判断。

不能因为所有检出物质都低于阈值，就认为生物学评价完成了。 化学层面无风险，不等于材料的整体生物学反应无问题——比如材料的物理形态、表面特性也可能引起反应。

记录与可追溯

使用关注阈值的论证过程应当完整记录，包括：

检出物质清单与浓度；每种物质的分类判断及依据；所选限值及选择理由；暴露量计算过程与全部假设；比较结果；以及对超过限值物质的处理。

「全部假设」这一点值得强调。 审查方要复核的是假设是否合理，如果假设没写出来，就无法复核，这本身就是问题。

保守假设的把握

暴露量计算中要用到不少假设，假设的保守程度需要把握：

过于保守的后果。 假设定得极端，算出的暴露量远高于实际，可能显示出本来不存在的风险，导致不必要的整改。

不够保守的后果。 假设偏乐观，结论站不住，审查时会被质疑。

合理的做法是：采用有依据的保守假设，并把依据写出来。 比如假设产品中的物质全部释放，这是一个明确的上限假设，容易论证也容易被接受；如果按部分释放计算，就要说明释放比例的依据。

建议先用最保守的假设算一遍 ——如果最保守的情况下都低于限值，结论就很稳，不必再细算。只有超过限值时，才需要用更贴近实际的假设重新计算。

多物质的处理

表征结果通常检出多种物质，处理方式：

逐个比较。 每种物质分别与适用限值比较。

同类物质考虑合并。 结构相似、毒性机理相同的物质，可能需要合并考虑暴露量。

总量的考虑。 某些评价体系对总暴露量也有要求。

主要贡献者识别。 找出暴露量占比较大的物质，这些是改进的重点。

第四项对产品改进有直接价值 ——知道是哪种物质贡献最大，就知道该调整哪个添加剂或哪道工序。

我们的做法

做化学表征时，我们会把检出物质的浓度、检出限、鉴定依据完整列出，并把检出限与常用的阈值放在一起对照 ——如果检出限高于阈值，那么「未检出」这个结果支撑不了结论，方法需要调整。

这个核对在方案设计阶段就应当做，而不是等数据出来才发现灵敏度不够。我们在讨论方案时会主动提出这一点。

如果你在做生物学评价，遇到检出物质没有毒理学数据的情况，可以把表征结果和产品使用方式发过来一起讨论，或者直接联系：132 4819 8029。能力范围见服务介绍，送检要求见送检要求，流程见检测流程。

English version

Conclusion. After chemical characterisation a familiar problem arises: a substance is detected but no toxicological data can be found for it, so its risk cannot be judged directly. The threshold of toxicological concern offers a route: if exposure to a substance falls below a certain level, the risk may be regarded as acceptable even without knowing exactly what the substance is. The approach rests on statistical analysis of a large body of substances with known toxicity. It carries explicit preconditions and exclusions, however. Applied where it does not fit, the conclusion does not hold, and reviewers identify such cases directly.

Preconditions. The substance must have been detected and quantified, since the threshold compares against exposure and without a quantity there is nothing to compare. The substance must not fall within an excluded category, certain highly toxic classes and substances with particular mechanisms being expressly outside the approach. The exposure route must match, since thresholds are derived for particular routes and use on others needs justification. And the exposure calculation must be sound, with the conversion from extraction concentration to human exposure resting on explicit and conservative assumptions. The second is the one most in need of checking. Excluded categories typically cover substances bearing specific structural alerts and known high-concern classes. Where a detected substance falls within them, specific toxicological data must be found and the threshold approach cannot be used.

Basis for tiering. Thresholds are generally tiered by structural characteristics, with different limits per tier. Structural alerts indicate features suggesting potential toxicity. Known toxicity classes indicate membership of a high-concern category. Exposure duration distinguishes short-term from long-term limits. And exposure route carries its own limits. Duration matters greatly in device evaluation: products in brief contact and those implanted long-term attract very different limits. Determine duration from actual use, with conservative margin.

Steps. Establish the substance's identity, at least to the extent of knowing its class, since that is what allows exclusion to be checked; substances of entirely unknown structure demand greater caution. Check whether it falls within an excluded category, in which case specific data are required. Determine the applicable limit from contact duration and route. Calculate exposure, converting from detected concentration through the product's contact mode, area and duration to a daily exposure, with assumptions written down. Compare and judge, treating exposure below the limit as acceptable and exposure above it as requiring further data or action. And record the reasoning so that every step's basis and assumptions are traceable.

Common errors in exposure calculation. Treating extraction concentration directly as exposure, when extraction conditions differ from use and conversion is required. Ignoring differences in contact area between specimen and product. Underestimating contact duration, which should reflect cumulative actual use including repeated use. Failing to sum across components, since the same substance may come from several parts and exposures add. And using an inappropriate body weight assumption where users include children. The fourth is easily missed in structurally complex products, particularly where one additive appears in several components.

When the approach does not apply. Where the substance falls within a clear exclusion. Where exposure exceeds the applicable limit. Where the product serves a special population and the limit has not been adjusted accordingly. Where identity is entirely unknown and cannot be classified. And where the regulator expressly requires specific data. In those cases the available routes are finding or commissioning specific toxicological data for the substance, reducing its release through formulation or process adjustment, or changing the material. Reducing release is often the most practical: adding a cleaning step, adjusting cure conditions or substituting an additive can each bring exposure below the threshold.

Relationship to the overall evaluation. The threshold approach is one tool within biological evaluation, addressing the specific problem of a substance detected without data. The overall evaluation still requires complete material information, determination of contact nature and duration, use of existing data, any necessary biological testing, and an integrated risk judgement. Every detected substance falling below the threshold does not mean the biological evaluation is complete: absence of chemical risk does not establish that the material as a whole provokes no biological response, since physical form and surface characteristics can also provoke one.

Records and traceability. Record the reasoning fully: the list of detected substances and concentrations, the classification judgement and its basis for each, the limit chosen and why, the exposure calculation with all assumptions, the comparison, and the treatment of any substance exceeding its limit. All assumptions deserves emphasis: what a reviewer checks is whether the assumptions are reasonable, and assumptions not written down cannot be checked, which is itself a finding.

How we handle it. In chemical characterisation we set out concentrations, detection limits and the basis of identification in full, and compare detection limits against the thresholds commonly applied. Where a detection limit sits above the threshold, a not-detected result cannot support the conclusion and the method needs adjusting. That check belongs at the protocol design stage rather than after the data arrive, and we raise it when discussing the protocol.

Send us the characterisation results and how the product is used and we will work through it. Phone or WeChat: +86 132 4819 8029.