GMP辦公室翻譯組

Data Processing and Peak Integration

數據處理和峰的積分

Chromatographic software uses dynamic peak detection algorithms and automatic peak detection algorithms. Integrating the chromatograms using software to apply the integration events, such as peak width, threshold, height, and area, etc., and then visually observing the peak integration for its correctness is generally recommended. The inherent or analytical variations and combination of various and drifting baselines in which the auto-integration will either underreport or overintegrate the peak areas.

色譜軟件使用動態(tài)尋峰算法和自動尋峰算法。通常建議先用軟件積分色譜來處理像峰寬、閾值、峰高及峰面積等需要積分的情形，再人工確定積分的正確性。因為一些固有的或分析變化或兩者皆有以及基線漂移的原因，自動積分將會使峰面積偏大或者偏小。

Once acquired, the electronic raw data of the measurements (measured values and metadata) are stored and available for processing. Some acquired electronic raw data already represent usable results (e.g., weight, temperature, and humidity). Other acquired electronic raw data, such as intensity values correlated with time or wavelength and generated by chromatography or spectroscopy, require further processing to obtain usable results (e.g., retention times, peak areas, and amounts). These processes, such as integration and calibration, are defined by processing parameters or calibration factors and affect only the resultant data after processing, but not the acquired electronic raw data. In contrast to the acquired electronic raw data, the processing parameters, such as integration events and calibration curves, may be changed during data evaluation. The changed processing parameters, methods, and processed data should be identified by versioning (i.e., number of times reprocessed) either on the results or from the audit trail data (25).

實驗的電子原始數據（測試結果和元數據）一經（色譜軟件）獲取，將會被（軟件）存儲并可用來進行處理。那些諸如重量、溫度、濕度等電子原始數據在被獲取后即可直接作為結果；而其他由色譜或光譜產生的與時間或波長相關的電子原始數據，則需要經過進一步處理才能夠得到如保留時間、峰面積以及物質的量等可以被使用的結果。像積分和校正這些僅由處理參數或校準因子所定義的處理過程，它們只影響處理后的最終結果，而對電子原始數據沒有影響。與獲得的（確定的）電子原始數據相反，在數據評估過程中，積分事件和校正曲線等處理參數可能發(fā)生改變。改變了的處理參數，方法以及處理后的數據應該標示版本控制的方法標記在結果或在審計追蹤中。

Integration events should be defined through standard algorithms, and metadata and should be associated with respective raw data files. In such cases, the analyst adjusts the integration events to obtain proper peak integration. As chromatography is a comparative technique, the consistent integration events should be applied for the entire set of chromatograms, as much as possible, or published along with the chromatograms. Figure 6.3.7-1 represents an overlaid chromatogram of the standard and a sample (of the same concentration) integrated with the same integration events. Integrating small peaks, closer peaks, negative peaks, drifting peaks, and peak-to-valley requires time and skill. Integrating peaks manually is not recommended and should be avoided to the extent possible.

積分事件應該通過標準算法和元數據進行定義，且處理后的結果應與各自的原始文檔進行關聯(lián)。只有這樣，分析師才可以通過調整積分事件從而獲得適當的峰面積積分。因為色譜是一種比對技術，所以應盡可能的將同一積分事件應用于整個色譜圖集或和整個色譜圖一起發(fā)布。圖 6.3.7-1表示標準品和某濃度下的樣品通過同樣的積分事件進行積分，并一同呈現(xiàn)在同一色譜圖中。對小峰、接近峰、負峰、漂移峰以及峰谷進行積分需要時間和技巧。不推薦手動積分，且應盡可能避免進行手動積分。

QC laboratories should have procedures in place that require authorization to perform manual integration and for procedures to track such events to avoid unnoticed or unevaluated cases that may affect the accuracy of the results. PDA defines manual integration as a process used by a person to manually integrate the peak height or area by modifying the baseline of the chromatogram with use of chromatographic software. The conditions and circumstances when manual integration would be allowed should be predefined (e.g., complexity of the sample matrix). Generally, a good chromatographic data system would be able to render consistent and reliable baselines for an overwhelming majority of injections within a chromatographic run. When consistently bad chromatographic peaks and baseline issues are encountered, having good documentation should not be the only way to address the issue. In this circumstance, the goal should be to improve the system and ensure that the data generated is reliable and consistent.

QC實驗室應建立需要授權才能進行手動積分的程序并追蹤此類事件（注：手動積分）的程序，避免因為人員忽視或者未評估而影響結果的準確性。PDA將手動積分定義為：由人員使用色譜軟件修飾色譜圖的基線來人為對峰高或者峰面積進行積分的過程。應當事先定義允許進行手動積分的條件和情形（如：復雜的樣品基質）。通常，一個良好的CDS系統(tǒng)應該能夠在色譜運行中為絕大多數進針提供連續(xù)且可靠的基線。當色譜峰連續(xù)比較差且發(fā)生基線發(fā)生問題時，將問題完整的記錄下來并不是唯一的解決途徑。在這種情況下，應該以改善系統(tǒng)并確保產生的數據真實可信為目的。

The Quality Unit  should define standard protocols for processing data to include the following, which may be instrument or application-specific:

質量部門應定義處理數據的標準方案，可以是針對儀器或特定應用。方案應包含以下內容：

• Reprocessing peaks

  峰重新處理

• Applyingin strument/application-specific integration events/algorithms

  使用儀器/特定軟件的積分事件/算法

• Fully integrating peaks

  峰完全積分

• Inhibiting ordisregarding any peaks in the test chromatogram (e.g., blank peaks, placebo peaks, solvent peaks) without scientific justification; examples where justification is needed include counter ion and reagent interactions with a sample

  禁止在沒有科學論證的情況下，忽略測試色譜圖中的任何峰值（如：空白峰，安慰劑峰，溶劑峰）。如樣品與反離子或試劑相互反應需要科學論證。

• Applying thesame integration events for all samples in the sample set  or sequence andjustifying any change in integration events.

  對處在同一樣品集或序列的樣品采用同樣的積分事件，并說明積分事件的任何改動。

Peak integration is the process used by a chromatographic system to determine the peak height and width and obtain the quantitation of the peak of interest. Certain USP monograph methods specify inhibiting integration at specified zones in the chromatographic run. USP <621> states that peaks can be disregarded by setting the thresholds in the integration to at least half of those below reporting threshold(32). Thereby, utilizing the built-in capability of the chromatographic data acquisition software to inhibit integration of peaks from solvent, mobile phase, placebo, and counter ions in impurity analysis in a common industry practice. Firms should scientifically evaluate and judiciously determine whether to use the inhibit integration functionality. Peaks in the chromatographic analysis may be excluded in the event of a known abnormality. Unknown peaks should be integrated and investigated according to the firm’s quality procedures.

峰（自動）積分是色譜系統(tǒng)用來確定目標峰峰高和峰寬以及定量的過程。某些USP專論方法指出在色譜運行時禁止（系統(tǒng)）對特定區(qū)域進行積分。USP<621>指出只有在通過對積分中的閾值進行設置（低于報告閾值一半以上）（32），才可以將那些低于報告閾值一半以上的峰忽略。因此，行業(yè)的普遍做法是，在雜質分析時利用色譜數據采集軟件（CDAS）來防止對溶劑、流動相、安慰劑和反離子產生的峰進行積分。公司應科學合理的評估和決定是否使用抑制積分功能。在異常已知的情況下，可以將譜圖分析中的一些峰排除。而未知峰應根據公司的質量程序進行積分和調查。

Manual integration is the process used by the analyst to integrate the peak height or area by modifying the baseline of the chromatograph using software(49). Manual integrations may be required for R&D and biological labs. Though in QC labs, manual interagtion may be necessary or acceptable only under special circumstances (e.g., complex chromatography due to sample matrix interferences, poor resolution, co-elution of peak of interest, problem with the baseline, or  software with limited capabilities); however, manual integration is not generally accepeable for assay. Manual integration should not be left to an analyst’s discretion; it should be performed only according to an approved procedure, with documented approval from a supervisor and results appropriately documented. A mature quality system will review these instances as part of continual improvement of methods and  equipment. Modern chromatographic software identifies and displays manually integrate peaks. Figure 6.3.7-2 represents a model chromatogram with manually intergrate peaks and software integrated peaks.

手動積分是分析師通過使用軟件對色譜圖基線進行修改從而對峰高或峰面積進行積分的過程（49）。研發(fā)和生物實驗室可能需要（對色譜圖）手動積分。盡管對于QC實驗室，手動積分可能僅在特殊情況（如：由于樣品基質的干擾所引起的復雜圖譜；分辨率差；目標峰的共洗脫，基線問題以及系統(tǒng)能力有限）下是必需的或可接受的；但手動積分還是不適合應用于化驗。手動積分只有在得到主管書面批準且結果正確記錄后，根據批準程序才能進行，而不應該由分析師自行決定。一個成熟的質量體系應審查這些實例，作為方法和設備改進的一部分。現(xiàn)在的色譜軟件已可以識別和標示出手動積分峰。Figure 6.3-7.2 是一個模型色譜圖，它同時展現(xiàn)了手動積分峰和軟件積分峰。

Printed chromatographic should be presented in visible scale as per the respective analysis(peak top visible for assay or single analysis, peak base clearly visible for purity analysis). After integration, results may be publish or electronically stored. If results are reprocessed, permission from supervisor is required. Those types of events may be noted in a log (paper of electronic) for quick reference.

打印出的色譜圖應以可見比例形式呈現(xiàn)各個分析項需要的內容，如對于含量和單個分析項，峰頂應可見；對于純度分析，峰基應可見。在積分后，結果應能夠輸出或以電子形式存儲。如果結果被重新處理了，那么應得到主管的許可?？梢詫⑦@些時間記錄在日志（紙質或電子形式）中，以供快速參閱。

Figure 6.3.7-3 illustrates a chromatogram in proper scale and visibility to determine the proper integration. Common practice across the industry is to present a visible and clear baseline for multicomponent analysis and the entire peak for single-peak analysis.

Figure 6.3.7-3以一個有著適當比例和可見的色圖譜來說明適當的積分。行業(yè)內通常的做法是為多組分分析提供清晰可見的基線，而為單峰分析提供整個色譜峰。 

Figure 6.3.7-4a represents to the chromatogram printed in large scale, where peak integration is not visible. Figure 6.3.7-4b represents the same chromatogram with printed proper scaling, where peak trimming and missing peak integrations are visible, revealing possible intentional data manipulation.

Figure 6.3.7-4a是一張以最大比例打印的色譜圖，其中峰值積分并不可見。Figure 6.3.7-4b是同樣的色譜圖以適當的比例打印，其中可以看到峰值修正和缺失峰的積分，揭示了可能的有意的數據操作。

Assay analysis, dissolution, and content uniformity involve a single analyte/peak analysis where the content will be calculated against a known standard. Gaussian peaks, with allowed system suitability characteristics of accuracy and precision, are expected in these types of analysis. Normal Gaussian peaks should be integrated exactly where the peak starts and finishes. The peak integragtion represented in these types of analysis is illustrated in Figures 6.3.7-5a-f.

含量，溶出度以及含量均一性包含了對單一分析物/峰按照已知標準進行計算的分析項目。在這些類型的分析項目中，都期望其具有準確和精確的系統(tǒng)適應性特征的高斯峰。正常的高斯峰應該確切地從色譜峰起始位置開始積分到截止為止結束積分。Figure 6.3.7-5a-f 展現(xiàn)了這些類型測試代表性的峰積分。

Figure 6.3.7-5a is the unprocessed signal of  a  single component analysis. Figure 6.3.7-5b represents raw data integrated with proper integration events and displayed in visible scaling. Figure 6.3.7-5c represents the peak integrated by trimming to yield low area. (Associated risk: Reporting fewer peak areas leads to less assay or dissolution or lower content values.) Figures 6.3.7-5d-f represent the raw data integrated with improper integration events, which presents more area. (Associated risk: Reporting more peak areas leads to more assay or dissolution or higher content values.)

Figure 6.3.7-5a是單組分分析的未處理的信號峰。Figure 6.3.7-5b表示的是對原始數據以適當的積分事件進行積分并以可見的比例呈現(xiàn)。Figures 6.3.7-5c 表示的是通過修整積分而產生了低面積的峰。（相關風險：報告峰面積的減少會導致含量或溶解度的降低或含量值的減少）Figures 6.3.7-5d-f 表示的是對原始數據用不適當的積分事件進行積分會表達更多的面積。（相關風險：報告峰面積的增加會導致含量或溶解度的增加，或含量值的增加）

Data manipulation of chromatograms by manually manipulating the integrated peaks is one factor the FDA has cited in Warning Letters(50-53). In these illustrations, all the integrations presented are made through chromatographic software by adjusting integration events. Chromatographic software will indicate which peaks are maually integrated. Laboratory management should have appropriate controls in place for detecting the data compromises made by applying the wrong integration events.

手動操作積分峰對色譜圖進行數據處理是FDA在警告信中引用的一個因素。在這些圖示中，所有的積分都是通過調整色譜軟件的積分事件來完成的。色譜軟件會標示出哪些峰是手動積分的。實驗室應該有適當的控制措施以檢測應用了錯誤的積分事件而造成的數據損毀。

Figures 6.3.7-6a-c and Figures 6.7.7-7a-c show integration errors that occur for multicomponent analysis (peak groups) , such as related substances analysis/chromatographic purity analysis. In both cases,fewer peak areas will reported.

Figure 6.7.7-6a-c 和Figure 6.7.7-7a-c展示的是在如成分分析/色譜純度分析這種多組分分析（峰組）中的發(fā)生的積分錯誤。這兩種情況下，報告的峰面積都會減少。