Basic concept: Analytical sensitivity (limit of detection)

 Basic concept: Analytical sensitivity (limit of detection)

 

 

1. Analytical sensitivity (detection limit)

 

1. The lowest analyte concentration detected is the analytical sensitivity or detection limit of the detection system. This concentration limit is particularly important for drug testing in court, and it is critical to know whether there is any drug in the sample through testing. In addition, tumor markers and many specific proteins should have a detectable minimum concentration or a certain amount; such as: prostate specific protein (PSA), which is important information for monitoring recurrence of patients after treatment; A meaningful minimum amount of specific protein is required to be specified. Negative and positive nucleic acid test reports also require an explanation of how much virus the minimum copy of nucleic acid that can be detected is equivalent to. Therefore, determining the lower reportable limit of a detection system is an important analytical performance.

 

2. Currently, the term detection limit is confusing. Manufacturers use various terms such as: sensitivity, analytical sensitivity, minimum detection limit, functional sensitivity, limit of detection, limit of quantitation, etc. . There is as yet no standard definition, so it is necessary to know what each term actually means and the experimental manner in which it is determined, how the data are processed, how an estimate is made from the data, and whether this estimate is useful for the medical application of the test. The analytical sensitivities presented below are divided into the lower limit of detection, which has a qualitative meaning, and the biological detection limit and functional sensitivity, which has a quantitative meaning.

2. Lower Limit of Detection (LLD)

 

For each test, always make a blank sample. The detection method is usually calibrated to the zero point with the blank response, and then detect the reaction response of each test sample. The responses of these samples are the corresponding responses of the analytes after subtracting the response of the blank sample. However, the amount of blank responses also fluctuated. If the blank detection is repeated many times, the discrete range index of the blank (response) mean and standard deviation are expressed. When determining method performance or drawing a standard curve, the blank mean value is often used to indicate that the blank response is greater than or less than the blank mean value, each with a 50% probability. When the blank response is less than the blank mean value, the detection response of the same sample (without deducting the blank response) seems to reflect that there are more analytes, and the detection method seems to be more sensitive. When the blank response is greater than the blank mean, it appears that the previously detectable analyte is now undetectable. Therefore, the detection method must be clear: what is the amount of analyte that can be detected? The standard curve starts from zero, so can the reported analyte amounts start from zero? This is the question that the lower limit of detection has to answer. Statistics show that if the fluctuation of the blank response volume obeys the normal distribution law: the blank response volume x blank of each single detection has a 95% possibility of:

 

—2.s blank ≤ x blank ≤ —blank 2.s blank

 

That is: ∣x blank - blank ∣≤2.s blank

 

Among them, half of the average value of the blank will make the analyte easier to detect. This is not undetectable, so there is no need to consider it. If there is a blank that has a detection response that is 2s larger than the average value of the blank response, the possibility of still considering it as a blank response is only 5 %; There is a 95% probability that it belongs to the detection response of the analyte formed in the sample; it is more than 2s blank than the blank mean. In the same way, if the response amount is more than 3s different from the blank mean value, the possibility of the blank response amount is only 0.3; and there is a 99.7% possibility that the response amount is the response amount formed by the analyte in the sample. Therefore, if the response of the sample is detected If the response amount is larger than the blank mean value, but the difference from the blank mean value is less than 2s blank or 3s blank, it can only be said that these response amounts are the response amount of a single detection of the blank sample, and there is no analyte in the sample, or it means that the amount of analyte is zero. The response of more than 2s blank or 3s blank is considered to contain the analyte in the sample.

 

The lower limit of detection is defined as the amount of analyte corresponding to the amount of detection response that can be achieved by a single detection of a sample. The detection system or method can only report "no analyte detected" for analyte amounts less than or equal to the lower limit of detection. Two possibilities are usually estimated at 95% or 99.7%:

 

95% probability: LLD = blank 2.s blank

 

99.7% probability: LLD = blank 3.s blank

 

It should be noted that detection systems that directly read out concentration units will report zero for detections below zero, and their distribution is not normal, so the calculated mean and standard deviation cannot faithfully express the true situation of the lower limit of detection. If the detection response It can be represented by the initial value, such as: absorbance, fluorescence, etc., and the s blank is valid at this time. Therefore, the initial values should be used to calculate the mean and standard deviation, and then converted to concentration units.

3. Biological Limit of Detection (BLD)

 

A response signal greater than the lower limit of detection indicates that the analyte is present in the sample, but the method is not yet able to correctly report quantitative results. Because in such a low concentration or other magnitude range, the response response of a single test sample has poor repeatability. So at what detection response size is it good to report quantitative results? Two approaches are described: biological detection limit and functional sensitivity. In principle, multiple samples with concentrations close to the detection limit (certainly not blank samples) are tested repeatedly, and the response of the samples after deducting the blank response is summarized as the mean and standard deviation. According to the law of normal distribution, there is a 95% or 99.7% probability that the response quantity of a single detection sample is 2 or 3 times the standard deviation of the response quantity from the mean value of the non-detection response quantity. A single response that is larger than the mean is definitely fine; however, if the single response that is smaller than the mean crosses the lower limit of detection (the upper limit of the blank response), it means that the detection method cannot distinguish it by a single test. The blank still contains the analyte. Therefore, the minimum value of 95% or 99.7% of the detection response of these samples is also greater than the lower limit of detection (LLD), so that it can be guaranteed that the single detection response of the sample must not be the response of the blank under any circumstances. ; The sample has an analyte concentration that can be reported quantitatively. Among multiple samples that are close to the detection limit, the lowest analyte concentration (or other quantity) that meets this condition is the lower biological detection limit of the detection system or method.

 

The lower limit of biological detection is defined as: the lower limit of detection plus 2 or 3 times the standard deviation of the detection limit sample, to determine the limit of the detection system or method that can quantitatively report the low concentration of the analyte extracted or other values.

 

The specific measure of biological detection limit (BLD) is:

 

95% probability: BLD=LLD 2s detection limit sample

 

99.7% probability: BLD=LLD 3.s detection limit sample

 

This term perfectly expresses the detection limit of the actual sample, such as what concentration is the zero value or there is no difference in analyte. When confirming the manufacturer's BLD statement, the manufacturer's statement of the detection limit sample and concentration should be the same.

4. Functional Sensitivity (FS)

 

Functional sensitivity is defined as: the average concentration of the sample corresponding to the detection limit when the CV between days is 20%, and the lowest concentration or other quantitative value of the analyte that can be quantitatively reported by the detection system or method is determined. To estimate FS, multiple detection limit concentrations were used to determine precision performance at low concentrations, from which the concentration with 20% CV was selected. When confirming the specification of the manufacturer's FS, the detection limit sample concentration used should be the same as the manufacturer's specification.

5. Factors to be considered in the experiment

 

Typically two different types of samples are prepared. One is a "blank" sample, ie contains no analyte and has zero analyte concentration. The other is a "limit of detection" sample that contains a low concentration of the analyte. In some cases, several "limit of detection" samples need to be prepared. Blank and detection limit samples were tested in duplicate by the test method. Calculate the respective mean and standard deviation. Different detection limits were calculated from blank and detection limit sample data.

 

1. Blank solution: One part is used as a blank and the other is used to prepare detection limit samples. Ideally, the blank should have the same matrix as the patient sample being tested. However, it is common to use the "zero standard" from the series standards of the detection system as a blank. For some projects, samples from patients without certain diseases after surgery (such as: PSA-free serum from patients with prostate tumor surgery) can be used as blank samples.

 

2. Detection limit sample: When confirming the sensitivity performance of a certain method, add the analyte to the blank solution to prepare the detection sample. The amount of analyte added should be the detection concentration specified by the manufacturer. In establishing the limit of detection, it is necessary to prepare several limit-of-detection samples at concentrations within the upper and lower ranges of the expected limit of detection. Number of repeated tests: There is no specific regulation, but it is often recommended to do 20 times. It meets the requirements of clinical testing for repeated testing experiments. Manufacturers often recommend 10 times, and laboratories often adopt 10 times to reduce costs

 

3. Experiments take time: If the lower limit of detection is mainly understood from the repeatability of the blank solution, intra-batch or short-term experiments are often done. If the quantitative detection limit is mainly understood from the repeatability of the "detection limit" sample, it is recommended to conduct a longer-term experiment, which represents the performance of the day-to-day determination. Actually do 10 tests (10d).