QC/T 502-2025Metallographic examination for steel automobile parts by induction hardened (English PDF)

4.1 General

The laboratory shall validate its established methods, standard methods used beyond the intended scope, and expanded and modified standard methods. Validation shall be as comprehensive as possible to meet the needs of the intended use or application area.

For validated methods, laboratories shall develop work instructions.

The laboratory shall first conduct in-laboratory method validation research tests according to the technical plan. Then, interlaboratory collaborative testing is performed by the collaborating laboratory using the same samples. The characteristic parameters of the method to be validated can be validated by comparing the method to be validated with the reference method.

sufficient number of naturally contaminated samples cannot be obtained, the use of artificially contaminated samples is allowed. The matrix of the test sample shall include samples with different contents of background microorganisms (high content, low content), samples with different processing methods, and original (unprocessed) samples.

Example 1: If the method to be validated is only applicable to a certain type of matrix, it only needs to be validated for that type of matrix. If the method to be validated is applicable to multiple matrices, such as food, select at least 5 food categories. For each type of food, no less than 3 different types of samples (such as unprocessed food, primary processed food, and deeply processed food, etc.) are selected for testing and research. There are 20 copies of each type of sample.

Example 2: For qualitative methods, the ideal situation is to have 10 positive samples and 10 negative samples of each type of sample. At least 50% of positive samples shall be set for each type of matrix. Sample levels that are all negative or positive are meaningless.

4.3.2 Inter-laboratory method validation

4.3.2.1 General principle

Inter-laboratory method validation is when different laboratories use the same sample for testing, compare the characteristic parameters of the method to be validated and the reference method, and compare these results with the limiting standards for acceptable differences between the method to be validated and the reference method. Inter- laboratory method validation is usually completed by 8 collaborating laboratories.

Under special circumstances, it may be completed by at least 5 collaborative laboratories.

4.3.2.2 Test sample requirements

Uniform test samples are prepared by the organizing laboratory and a combination of artificial and naturally contaminated samples is used wherever possible. If there are no naturally contaminated samples, artificially contaminated samples can be used. See

4.3.1.2 for substrate selection.

For qualitative methods, each matrix includes at least three test levels of samples, including negative controls and two test levels of positive samples. Samples with at least one level are set to obtain partially positive results. Based on the probability distribution, theoretically, a test level of 1 colony forming unit (CFU)/sample shall yield

30% negative results. Each collaborative laboratory needs to conduct 8 replicate tests for each level, using two methods. That is, each collaborative laboratory provides at least 48 result data (3 levels × 2 methods × 8 repetitions).

For quantitative methods, each matrix includes at least 3 test levels and covers the test range of the method. Two methods are used, with 8 repeated tests for each level. Each

LODalt - the level of detection of the method to be validated;

LODref - the level of detection of the reference method.

The acceptability limit (AL) of a paired study is 1.5, that is, the LODx of the method to be validated must not be higher than 1.5 times the LODx of the reference method. The

AL for unpaired studies is set to 2.5, that is, the LODx of the method to be validated must not be higher than 2.5 times the LODx of the reference method. It is acceptable for the LODx value of the method to be validated to be smaller than the LODx value of the reference method. This indicates that the method to be validated can detect lower levels of contamination than the reference method.

When the calculated value of the result is higher than AL, it is validated that the result does not meet the requirements of AL. When AL is not met, root cause analysis shall be performed to explain the observed results. Based on the AL and attachment information, validate whether the method to be validated is suitable or not suitable for the matrix type or category being analyzed.

4.4.1.3 Inclusivity study and exclusivity study

4.4.1.3.1 General

Each strain used for inclusivity study and exclusivity study shall be biochemically and/or serologically and/or genetically characterized in sufficient detail. For quantitative methods, inclusivity study and exclusivity study tests are suitable for counting specific microorganisms (for example, Listeria). They are not suitable for methods that count the total number of all microorganisms, such as total bacterial colonies, mold and yeast counts, etc.

4.4.1.3.2 Inclusivity study

Only inclusivity study testing is required using the method to be validated. No sample matrix needs to be added during the test. For inclusivity study testing, at least 50 pure cultures of the (target) microorganism shall be tested. Pure cultures of test strains shall be cell populations in stationary phase cultured on non-selective medium under optimal growth conditions. The inoculum level shall be 10 to 100 times higher than the minimum level of detection of the validated method. When the inclusivity study result is negative or suspicious, the reference method shall be used for repeated testing at the same time to analyze whether the relevant strains can be detected using the reference method.

4.4.1.3.3 Exclusivity study

Only exclusivity study testing using the method to be validated is required. No sample matrix needs to be added during the test. For exclusivity study testing, pure cultures of at least 30 (non-target) microorganisms shall be tested. The pure culture of the test strain shall be a cell population cultured on a non-selective medium under optimal growth conditions and in the stationary phase. If there is a selective enrichment step in the method to be validated, for the purpose of exclusivity study testing, a non-selective medium shall be used instead of the selective medium used in this step. When an exclusivity study result is positive or doubtful, the method to be validated shall be repeated. At the same time, the reference method is tested to analyze whether the reference method can detect relevant strains.

4.4.1.4 Relative trueness

A relative trueness study is a comparative study between the results of a reference method and the results of the method to be validated. It is performed using natural and/or artificial contamination of samples. Select at least 3 different types of samples for each matrix type, at least 5 samples of each type. The contamination levels of all samples shall cover the concentration range of microbial species.

The results obtained are analyzed using the Bland-Altman method. Calculate the average and difference between the method to be validated and the reference method.

Bias can be estimated by using the mean D of the difference between the two methods.

The variation of the mean value D is described by the standard deviation Sd of the difference. If the distribution of differences follows a normal distribution, the vast majority of differences shall lie within the 95% confidence interval of the bias.

Therefore, the difference between the two measurements is acceptable if it lies within the limits of agreement.

4.4.1.5 Accuracy

An accuracy study is a comparative study between the results of a reference method and the results of the method to be validated. It is performed using artificial contamination of samples. There is a total of 6 test samples for each type of sample.

Two of the samples show low concentration contamination levels, 2 are intermediate concentration pollution levels and 2 are high contamination levels. They cover the entire contamination range for the selected type of sample. Each test sample is tested in duplicate 5 times.

For each test sample, two methods are used to perform quantitative detection under repeated conditions. Convert count results to logarithms. If the logarithmic value (yi) of the result of the method to be validated is assumed to be normally distributed, calculate the beta-expectation tolerance interval (beta-ETI) for the value of yi.

The acceptability limit for accuracy was set to: AL = ±0.5 log units. Plot the accuracy distribution graph. The upper and lower limits of the tolerance interval are connected by straight lines to interpolate the limiting behavior between different levels of the validation sample. The horizontal line represents the reference value obtained with the reference method. The difference between the reference value and the mean value for each contamination level of the method to be validated is represented by a black dot.

When there is no bias, these values lie on the horizontal reference line. In addition, the

Where,

N- -Number of negative results at L0 level;

P0 - The number of false positive results obtained from blank control samples before result identification;

CP0 - The number of false positive results obtained for blank control samples.

For the results at L1 and L2 levels, calculate the sensitivity of the method to be validated (SEalt) and the sensitivity of the reference method (SEref), the relative trueness (RT) and the false positive rate (FPR) of the method to be validated.

In paired studies, partial positive results shall be obtained for at least one of levels L1 or L2. Calculate the difference between the negative deviation (ND) and the positive deviation (PD) of the level (ND - PD) and the sum of the negative deviation (ND) and the positive deviation (PD) (ND+PD). The resulting values shall not be higher than the acceptability limits (ALs). The number and acceptability limits (ALs) of different collaborating laboratories can be found in Table A.2.

In unpaired studies, the difference between the negative deviation (ND) and the positive deviation (PD) at the level (L1 or L2) where partial positive results are obtained (ND-

PD) is calculated. The value of (ND - PD) shall not be higher than AL. The AL value is defined as [(ND - PD)max]. Its calculation can refer to ISO 16140-2.

4.4.2.2 Quantitative method

Calculate the accuracy of results. Plot the accuracy distribution graph. The upper and lower limits of the tolerance interval limits are connected by straight lines to interpolate the limiting behavior between different levels of the verification sample. The horizontal line represents the reference value obtained with the reference method. The difference between the reference value and the mean value for each contamination level of the method to be validated is represented by a black dot. When there is no bias, these values lie on the horizontal reference line. In addition, the acceptability limits are represented by two horizontal dashed lines and the beta-ETI limits are represented by a broken line.

The acceptability limit is set to ±0.5 log units.

The method is considered equivalent to the reference method when the values of beta-ETI lie within the acceptable limits for all contamination levels.

Specificity of reference method

Specificity of method to be validated

5.3 Qualitative method

Each matrix needs to be tested under at least three different inoculation levels: high, medium, low and no inoculation. Each case is tested twice, and a total of 8 test results are obtained. It includes 2 high levels: the inoculum is no more than ten times the expected LOD50; 2 medium levels: the high-level inoculum is diluted 2 times to obtain the intermediate level inoculum; 2 low levels: the high-level inoculum is diluted 10 times to obtain low level inoculum; 2 non-inoculums are served as blank controls.

Only positive results can be obtained with high levels of vaccination (10×LOD50). If negative results are obtained, the experiment shall be repeated for all levels. Blank controls shall not yield positive results. If a positive result is obtained, the experiment shall be repeated for all levels.

The number of positive results obtained for each vaccination level is recorded. Use

Table B.1 in Annex B to determine the estimated LOD50. Compare to LOD50 from method validation study. If there is no validation data for the method, the estimated

LOD50 shall be <=3 CFU/test unit.

5.4 Quantitative method

5.4.1 Estimated bias

Select the type of matrix to be used for method validation. Validation tests for quantitative methods shall use naturally contaminated samples whenever possible.

Samples can be from different batches (different products, different manufacturers, etc.) to cover the measurement range of the method. Three different batches are required for each validated matrix type. Each batch represents one contamination level. Three contamination levels shall cover the range of use of this method. Depending on the level of contamination and the target microbial species, if the expected natural contamination level of the sample test unit is below a certain limit, artificial contamination is recommended.

The laboratory uses the reference method (preferred) or the method currently used by the laboratory as a control method to perform tests simultaneously with the method to be verified. If the laboratory does not have a control method that can be performed simultaneously, the laboratory shall use non-selective medium to count the concentration of the culture.

For each contamination level, the difference between the count results of the method to

......