FDAQC试验室检查指引

1、GUIDE TO INSPECTIONS OF PHARMACEUTICAL QUALITY CONTROLLABORATORIESNote: This document is reference material for invescigacors and other FDA personnel. The document does not bind FDA, and docs no confer any rights, privileges, benefits, or immunities fbr or on any pcrson(s)1. introductionThe pharmace

2、utical quality control Iaboratorr setves one of the mosc important funccioiis in pharmaceutical production and control. A sigilificaiic portion of rhe CGMP regulations (21 CFR 211) pertain co the quality concrol laboratory and product testing. Similar concepts apply co bulk drugsThis inspection guid

3、e supplements other iiispccrioiial informarion concained in other agency iiispccrional guidance documents. For example, Compliance Program 7346.832 requiring prc-approval NDA/ANDA inspccrioiis contains general inscrucrions to conduct product specific NPA/ANDA inspeccion audits to measure compliance

4、with chc applications and CGMP requirements. This includes pharmaccurical laboratories used for inprocess and finished produce testing.2. OBJECTIVEThe specific objective will be spelled out prior to the inspection. The laboratory iilspcccioil may be limited to specific issues, or chc iilspccrion may

5、 encompass a comprehensive evaluation of the laboratory compliance with CCtMP As a minimum, each pharmaceutical quality concrol laboratory* should receive a comprehensive GMP evaluation each raT years as pare of the smcuu)r- inspection obligation.In general these iiispccdolls may includechc specific

6、 mcchodologr which will be used to cese a new product一 a complete assessment of laboracorx-s conformance with GM Ps一 a specific aspect of laboratory* operations3. INSPECTION PREPAfLATIONFDA Inspection Guides arc based on the team inspection approach and our iilspccrion of a laboraton* is consiscent

7、with this concept. As pare of our effort to nehievu unifc)nnirr and consistency in lab()rat()nT inspections, we cxpccc chat complex, highly technical and specialized cesring equipment, procedures and data mailipulacions, as well as scienrific laboraton* opcrauons will be evaluated by all experienced

8、 laborat（）r）r analyst 5X*ich specialized knoxkdcdgc in such macrers.District management makes the final decision regarding the assigtimenc of pcrsoiuic） to inspections. NcTcrthclcss, we expect investigators, analysts and others to work as teams and co advise management 5x*hcn additional cxpcrrisc is

9、 required to complete a meaningful inspection.Team members parricipacing in a piuapproval inspection must read and be familiar with Compliance Program 7346.832, Prc-Approval Inspcctioiis/liivcsugauons. Rclcvaxic sections of the NDA or ANDA should be revised prior to the inspection； but if the applic

10、arion is not available from any other source, this rcviuw will have to be conducted using the companys copy of the application.Team members should meet, if possible, prior to rhe inspection co discuss the approach to chc inspection, to define the roles of the team members, and to establish goals for

11、 completion of the assignment. Rcsponsibilirics for development of all reports should also be established prior to chc inspccuon. This includes rhe preparation of the FDA 483.The Center for Drug Evaluation and Research （COER） may have issued deficiency letters listing problems that the sponsor must

12、correct prior to the approval of NDA/ANDA and supplements. The iiispccrion team is expected to review such letters on file ac the district office, and they arc expected to ask the plane for access co such letters. The team should e-aluatc chc replies co these lectors co assure chac the data arc accu

13、rate and auchenric Complete chc inspection even choui there has been no response to these letcers or when chc response is judged inadequate.4. INSPECTION APPROACHA. GencrmlIn addicion to chc genera） approach utilized in a drug CGMP iiispccrion, rhe inspection of a laborat（）nT requires rhe use of obs

14、cnrari（）ns of the laboratory ill operation and of the raw laboratory data co craluatc compliance with CGMP*s and to specifically carry* out the commicmciits in ail applicarion or DMF. When conducting a comprehensive inspcccioii of a laboratory, all aspects of the laboratory operations will be evalua

15、tedLaborat（）nT records and logs represent a vical source of information chac allows a complete overview of the technical abilicy of the staff aild of overall quality coilcrol procedures. S（ ）Ps should be complete and adequate and the operations of the laboratories should conform to chc wrirren proce

16、dures. Specifications and analytical procedures should be suitable and, as applicable, in conformance with application commicmciits and compendialrequirements Evaluate raw laboratory data, laboratory- procedures and methods, laboratory cquipmcnc,including maintenance and calibration, and methods val

17、idation data co determine the overall quality of the laboracorr operation and the ability co comply with CCtMP regulations.Examine chromacograms and spectra for evidence of impuricics, poor technique, or lack of instrumenc calibration.s use systems that provider for the invesrigacion oflaboratory- t

18、est failures. These arc generally recorded in some cypu of log. Ask co see results of analyses for lots of produce chac have failed to meet specifications and review chc analysis of lots chac have been rcccsccd, rejected, or reworkud. Evaluate the decision to release lots of product when the laborat

19、ory results indicate chac the lot failed to meet specifications and determine who released themE Pre-Approva)Documents rclaring co the formulation of the product, synthesis of the bulk drug substance, product spucificncions, analysis of the pnxlucc, and others arc examined during the review process

20、in headquarters Howcrvur, these reviews and evaluations depend on accurate and auchenric data that truly represents the product.Prc-approval inspccrions arc designed to determine if the data submirced in an application arc auchenric and accurate and if rhe procedures listed in the applicauoii were a

21、cmally used to produce rhe data contained in the application. Addirionally, they arc designed to confirm chat plants (including chc quality control laboratory-) arc in compliance with CGMP rcgularions.The analycical sections of drug applications usually contain only cese results and the methods used

22、 to obtain them Sponsors arc not required to file all the test data because such acrioil would require voluminous submissions and would often result in filing redundant information. Sponsors may deliberately or unintentionally select and report data showing chat a drug is safe and effective and dcsc

23、nrcs co be approved. The inspecrion team muse decide if there is valid and scientific justification for the failure co reporc data which demonstrates the pnxluct failed to meet its predetermined specifications.Coordination bcra*ccii headquarters and the field is essential for a complete review of rh

24、e application and the plane. Experienced invesrigacors and analysts may conracc the rcncw chemise (with appropriate supervisory concurrence) when questions concerning specifications and standards ariseInspections should compare the results of analyses submirced with results of analysis of ocher batc

25、hes that may have been produced. Evaluate rhe methods and note any exceptions to the procedures or equipment accually used from those listed in rhe application and confirm that it is the same method listed in the applicarion. The analyst is cxpccrcd to evaluate raw laboratory data for tests performe

26、d on the test batches (biobacches and clinical batches) and to compare this raw data to the data filed in the application.5. FAILURE (OUT-OF-SPEC1F1CAT1ON) LABORATORY RESULTSEvaluate chc company system to investigate laboratory rest failures. Tlicsc invesrigacions represent a key issue in deciding w

27、hether a produce may be released or rejected and form the basis for rcccsriiig, and resampling.In a recent court decision the judge used the term nout-of-spccificnckm (OOS) laboratory- resulc rather ch ail the term product FaiIurcM which is more common to FDA investigators and analyses. He ruled tha

28、t an OOS result identified as a laboratorT error by a failure investigation or an outlier test. The court provided explicit limitacions on the use of oudicr tests and these arc discussed in a later segment of this document., or overcome by retesting. The court ruled on rhe use of retesting which is

29、covered in a later segment of this document, is not a product failure. OOS results fall into three categories:laboratorT errornon-proccss related or operator errorprocess related or maiiufaccuring process errorA. LABORATORY ERRORSLaborat()nT errors occur when analyses make mistakes ill following the

30、 method of analysis, use incorrect standards, and/or simply miscalculate the data. Laborat()r)r errors muse be determined chroui a failure investigation to identify the cause of the OOS. Once the narurc of the OOS result has been identified ic call be classified into one of the three categories abov

31、e The inquiry may vary with the object under invescigarion.E. LABORATORY INVESTIGATIONSThe exact cause of analyst error or mis cake call be difficult co determine specifically and it is unrealistic to expect chat analyse error will always be determined and documeared. Nevertheless, a laborac()r)r in

32、vcsrigatioii consists of more than a retest. The inability to identify* an errors cause with confidence affects retesting procedures, not rhe invesrigarion inquiry- required for chc initial OOS result.The firms analyse should follow a written procedure, checking off each step as it is completed duri

33、ng the analytical procedure. We expect laboratoryT test data to be recorded di reedy ill notebooks; use of scrap paper and loose paper must be avoided. These common sense measures enhance the accuracy and intcgrinr of data.Rcncw and evaluate rhe laboracory- SOP for product failure iiivcstigauons. Sp

34、ecific procedures must be Followed when single and multiple OOS results arc investigated. For the single OOS result rhe invescigarion should include the following steps and these inquiries muse be conducted before there is a reccsc of the sample:o the analyse conducting the test should report the 0(

35、 )S result to the supervisoro the analyse and chc supervisor should conduct an inf()rmal laboracorr invesrigacion which addresses rhe following areas:1. discuss the testing procedure2. discuss the calculation3. examine the inscrumencs4. review the notcb(x)ks containing the OOS resultAn altcrnacn-c m

36、calls to invalidate ail initial OOS result, provided rhe failure invesrigauon proves inconclusive, is the ,outlicrM cesc. Howler, specific res trie cions muse be placed on the use of this wst1. Firms cannot frequently reject results on this basis.2. The LSP standards govern its use in specific eases

37、 only.3. The test cannot be used for chemical tesring results. An initial concent uiiif)rmirr test was OOS followed by a passing retest. The ini ci a) OOS result was claimed the result of aiialvsr error based on a scaristical evaluarion of the data. The court ruled char the use of an outlier test is

38、 inappropriatein this ease.When che lab()rat()nT investigacion is inconclusive (reason for the error is not identified) the firm:1. Cannot conduct 2 retescs and base release on average of three tests2. Cannot use outlier test in chemical rests3. Caimot use a re-sample co assume a sampling or prepara

39、tion error4. Cail conduct a retest oFdiffcrent tablets from the same sample when a retest is considered appropriate (see criteria clswhcrc)C. FORMAL INVESTIGATIONSFormal invesrigauons ex rending beyond the laboracorr must follow ail outline with particular attcDuon to corrective action. Thu company

40、must:1. Scare chc reason for the invesugarion2. PrmHdc summation of the process sequences that may have caused the problem3. Outline corrective actions ncccssarT to save che batch and prevent similar recurrence4. Lise ocher batches and pnxlucts possibly affected, rhe results of invesrigacion of thes

41、e batches and products, and ally corrective action Specifically:o examine other bacches of produce made by che crraiir employee or machine o examine other products pnxluccd by che errant process or operation5. Present rhe comments and signatures of all production and qualicy control personnel who co

42、nducted the investigation and approved ally reprocessed material after addirional res ringP. INVESTIGATION DOCUMENTATIONAnalyst*s mistakes, such as undetected calculation errors, should be specified with particularity and supported by cviduncu. Invcsrigatioiis along with conclusions reached must be

43、presetved with wrirccii documentacion that enumerates each seep of the invescigarion. The valuation, conclusion and cort ccuvc action, if ally, should be presented in an invesrigarion or failure report aiid placed into a central file.E. INVESTIGATION TIME FRAMESAll failure invcstigarioiis should be

44、performed within 20 business days of the problem occurrence and recorded and wriucn into a failure orinvcsrigarion report.6. PRODUCT FAILURESAn ( )( )S laboracon* result can be overcome (invalidated) when laboratory- error has been documented. However, nonprocc：ss and process related errors rcsulrin

45、g from operators making mistakes, equipment (other ch ail laboratorx- equipment) malfunctions, or a maiiufaccuring process that is fundamentally deficient, such as an improper mixing time, represent produce failures.Examine the results of invescigations using the guidance in section 5 above and eval

46、uate the decision to release, retest, or rework products.7. RETESTINGEvaluate chc companys re resting SOP for compliance with scientifically sound and appropriate procedures. A very- important ruling in one recent court decision secs forth a procedure co govern rhe retesting program. This discricc c

47、ourt ruling provides an cxccllcnc guide to use in craluaring some aspects of a pharmaceutical laboracor buc should not be considered as law, regulation or binding legal precedent. The court ruled chac a firm should have a prcdcccrmiiicd testing procedure and ic should consider a point at which testi

48、ng ends and the ptcxlucr is evaluated. If results arc nor satisfactory, chc produce is rejected.Addicionally, chc company should consider all retest results in the context of chc overall record of chc produce. This includes the history of the produce. The court ordered a recall of one batch of produ

49、ce on chc basis of an ini rial content unif()rmicv failure and no basis co invalidate the test result and on a history of conccnc uniformin problems with the product., r pc of ccsc performed, and process test results. Failing assay results cannot be disregarded simply on chc basis of acceptable conc

50、ent uniformity results The number of recescs performed before a firm concludes chac an unexplained OOS result is invalid or chat a produce is unacccpcablc is a macter of scientific judgment. The goal of retesting is to isolate OOS results buc retesting caimoc continue ad infinirumIn the ease of mmpr

51、occss and process-related errors, rccescing is suspect. Because the initial tests arc genuine, in these circumstances, additional res ring alone caimot contriburc co product quality. The court acknowledged chat some retesting may precede a finding of nonproccss or process-based errors. ()ncc this dc

52、tcrmiiiauon is made, however, additional re testing fr purposes of resting a product into compliance is noc acceptableFor example, in the case of content uniformity testing designed co detect variabilic- in the blend or cablets, failing and noil-failing results arc not inherently incoiisisccnt and p

53、assing results on limited retesting do nor rule out the possibility chat the batch is not uniform As part of the invcsrigacion firms should consider the record of pluvious batches, since similar or related failures on differenc batches would be a cause of concern.Retesting following an 0（ ）S result

54、is ruled appropriate only after the failure invcsdgarioii is underway and the failure investigation determines in pare whether retesting is appropriate. It is appropriate when analyst error is documented or the review of analysts work is Hinc（）nclusivcM , but it is noc appropriate for known and undi

55、sputed nonprocuss or process related errorsThe court ruled chac rctcsriilg:o must be done on the same, not a different sampleo may be done on a second aliquot from chc same portion of chc sample chac was chc source of rhe first aliquoto may be done oil a porrion of the same larger sample prcnously c

56、ollected for laborat（）nT purposes8. RESAMPLINGFirms cannot rely on resampling. The court ordered the recall of one batch of product after having concluded chac a successful resample result alone cannoc invalidate an initial OOS result, co release a product that has failed ccsuiig and tecesring unles

57、s chc failure invesrigauon discloses evidence chat the original sample is not representative or was improperly prepared Evaluate each resampling activity for compliance with this guidance.9. AVE RAGING RESULTS OF ANALYSISAveraging call be a rational and valid approach when the object under considera

58、tion is cocal product assay, buc as a general rule this practice should be avoided. Thu court ruled that the firm must recall a batch that was released for content uniformity on chc basis of averaged test results because averages hide chc variabilit）f among individual test results. This phenomenon i

59、s parcicularly troubling if cescing generates both OOS and passing individual results which when averaged arc within specification Here, relying on chc average figure without examining and explaining the individual OOS results is highly mislcadin百 and unacccpcablc.Content uniformity and dissolution results never should be averaged co obta