Introduction to Pharmacokinetic Studies in Research and Development of Cell Therapy Products

Part 3: Pitfalls in Quantitative Analysis of Cell Therapy Products (2): Highly Sensitive Quantification of Human Cells! Research Pharmacokinetics Laboratory, Takeda Pharmaceutical Company Limited Miyu Nakayama, Shunsuke Yamamoto, Akihiko Goto, Yu Moriya

　Hello! In the last issue, we reported on “Pitfalls and countermeasures for quantitative analysis and interpretation of quantitative results in gene-modified immuno-cell therapy”. In this third issue, I would like to discuss “Points to keep in mind when quantifying human cells in animal biological samples using PCR method”, focusing on our own research.

　As you know, evaluation of biodistribution of cell therapy products provides important suggestions for evaluating their efficacy and safety. In particular, cell therapy products using induced pluripotent stem cells (iPS cells) or embryonic stem cells (ES cells) have tumorigenic potential because of the residual undifferentiated cells. Therefore, their biodistribution in vivo must be clarified and the risk of tumorigenicity must be evaluated. It has been reported that tumor formation was observed in 1 out of 6 NOG (NOD/Shi-scid ^IL2Rγnull) mice when 107 mesenchymal stem ^cells with only 10 Hela cells were transplanted subcutaneously with Matrigel [1], thus, when evaluating biodistribution, it is highly sensitive to detect very few Therefore, it is necessary to use a highly sensitive human cell quantification method to detect a very small number of cells when evaluating biodistribution. Known methods for human cell quantification include labeled fluorescence, radioactivity, magnetic measurement, flow cytometry (FCM), quantitative polymerase chain reaction (qPCR), etc. (Table 1). Table 1). Fluorescence, radioactivity, and magnetic measurements are useful biodistribution evaluation methods that allow easy measurement and imaging of these signals. However, it is necessary to genetically or chemically label the cell therapy product prior to administration, and it is important to note the changes in biodistribution characteristics and cellular functions caused by such labeling. Cell quantification using FCM is a powerful tool for quantitative counting of target cells in blood samples without the need for labeling prior to cell administration [2]. However, the stability of the sample must be taken into account due to its characteristic of detecting living cells. In addition, the detection of cells in tissue samples by FCM poses challenges for quantitative evaluation because it is difficult to ensure the recovery rate and viability of cells during tissue dispersion. On the other hand, qPCR is a specific and sensitive method for detecting human cells in animal biological samples by targeting human-specific sequences, without labeling the cells prior to administration. Since biological samples can be cryopreserved[3] and human genomic DNA can be easily extracted from blood and other body fluids and tissues, the qPCR method is considered to be a versatile assay. (Table 1) Table 1.

Table 1 Comparison of human cell assays

	PCR	Fluorescence labeling	Radioactive labeling	Magnetic labeling (MRI)	Flow cytometry
Detection Target	Human gDNA	Fluorescence	Radioactivity	Magnetic	Cell
Sensitivity	High	Low-Medium	High	Low	High
Quantitative	High	Low-Medium	High	Low	High
Units	Cell count, copy number, DNA weight	Cell count, fluorescence intensity, dose ratio	Cell count, radioactivity intensity, dose-to-dose ratio	Magnetic intensity	Number of cells, presence ratio
Matrix	All organs	All organs	All organs	All organs	Body fluid Difficult to apply to organs
Live cell measurement	Not available	Not available	Not possible	Not possible	Possible
Non-clinical study	Possible	Possible	Yes	Yes	Yes
Clinical trials	Not available	Not Possible	Not Possible	Possible	Possible
Sample Stability	Long-term storage possible	Long-term storage possible	Long-term storage possible	Mainly for biological measurements	Long-term storage not possible

　Many cell quantification methods using qPCR have been reported [4] [5] [6]. Conventional qPCR-based cell quantification methods use artificially synthesized target sequences or human genomic DNA extracted from biological samples as the calibration curve, and are generally expressed in units such as “copies/µg gDNA” or “µg human gDNA/µg host gDNA” [7]. However, it is difficult to calculate the number of cells in a biological sample using these units. In addition, it is not easy to compare the distribution of human cells among tissues because the amount of genomic DNA contained in animal tissues differs from tissue to tissue. Therefore, we have established a new method to express results in terms of “cells/mL blood or mg organ,” which has been commonly used in conventional pharmacokinetic evaluations[4]. This method is based on the same principle as the LC/MS measurement used for the quantification of small molecules and other pharmaceuticals, and is based on the preparation of a standard sample to which cell lysate is added in each control matrix. The number of cells per unit unknown sample can be calculated from the Ct value.

　What target sequences are used in the qPCR assay? For sensitive detection of the human genome in animal samples, it is important to select human-specific and multi-copy sequences. Alu-qPCR, which targets the primate-specific Arthrobacter luteus (Alu) element of approximately 300 nucleotides in length, accounting for more than 10% of the human genome and approximately 1.1 million copies, is known as a particularly sensitive qPCR method. In fact, we have reported that human cell counts in mouse samples can be quantified with a sensitivity of 10 cells/50 μL blood or 15 mg tissue (7.5 mg for lung and spleen) using the aforementioned additive calibration curve [4]. On the other hand, the following issues have been identified with Alu qPCR. (1) Alu sequences are highly homologous among primates, making it difficult to apply Alu qPCR to non-human primate (NHP) samples, (2) non-specific amplification signals are observed even in non-template control (NTC) samples that do not contain human genomic DNA, and (3) Alu qPCR is not suitable for the analysis of human cells. (3) The qPCR method is easily affected by matrix effects during PCR measurement, and the PCR amplification curve may deviate for each matrix, and (4) The DNA recovery rate during DNA extraction varies from sample to sample, resulting in loss of precision and accuracy of measurement values.

　As mentioned above, biodistribution studies of cell therapy products using rodents provide useful information for understanding distribution characteristics and considering efficacy and safety. However, species differences in biodistribution and extrapolation to humans are limited by physiological differences, and the use of large animals, including NHPs, is increasing to bridge rodent-human studies and to accelerate the transition from non-clinical to clinical settings. In particular, NHPs can provide useful information due to their high genetic and physiological similarity to humans [8]. Therefore, we searched for a new target sequence to replace Alu that is also applicable to NHP samples [9]. to detect human cells in NHP samples with high sensitivity and specificity, a sequence must be human-specific, not present in NHP, and must be a multicopy gene sequence like Alu sequence. Long Interspersed Element-1 (LINE1, also known as L1 or LINE-1) is a retrotransposon that occupies about 17% of the human genome and has a copy number of about 500,000. LINE1 can be divided into several subfamilies (non-Ta, Ta-0, and Ta-1) like other mammalian retrotransposons. Among them, the Ta-1 family contains human-specific sequences that are not present even in chimpanzees, which have the highest homology to humans[10]. Therefore, we designed primers and probes for human-specific sequences of the Ta-1 family and evaluated their specificity. The results showed that the primers and probes did not cross the genomic DNA of NHPs, but amplified only human genomic DNA. In addition, when a cell-added standard was prepared using a control sample of crab-eating macaque, cell concentration-dependent amplification was confirmed, and it was confirmed that the sensitivity was equivalent to that of Alu-qPCR. NTC samples extracted from crab-eating macaque control tissues showed no nonspecific amplification observed with Alu-qPCR. Furthermore, LINE1-qPCR is applicable to rodent, rabbit, marmoset and porcine samples, demonstrating its high versatility.

　Biodistribution evaluation of iPS cell- and ES cell-derived cell therapy products with tumorigenicity concerns requires measurement of major tissues throughout the body. On the other hand, human cell quantification using qPCR requires the preparation of a standard for each matrix due to matrix effects in the PCR reaction. Therefore, the preparation of standards for each matrix requires a large number of concentrations of standards, which leads to a significant increase in the number of samples and requires a large amount of cost and man-hours. In addition, since control samples are used for the preparation of standards, a large number of mice for control tissue collection are required, especially in cases where trace tissue samples such as genital organs and bone marrow are to be measured. Therefore, from the viewpoint of animal ethics, we attempted to construct a method that enables measurement of human cells in various tissue samples with a single calibration curve in order to reduce the number of experimental animals. In constructing the method, we needed to solve two bottlenecks: the matrix effect during PCR reaction and the variation in DNA recovery rate mentioned above. First, to avoid matrix effects, we focused on Droplet digital PCR (ddPCR). ddPCR disperses template DNA into approximately 20,000 generated droplets per well according to Poisson distribution theory, and PCR amplification is performed independently in each droplet [1]. PCR amplification is performed independently in each droplet[11]. The key point here is that ddPCR is an endpoint assay. qPCR measures the fluorescence intensity of each PCR well at each PCR cycle and draws an amplification curve, whereas ddPCR determines whether the fluorescence intensity of each droplet is positive or negative at the endpoint. ddPCR determines if the endpoint is positive or negative according to the fluorescence intensity of each droplet. Matrix effects may affect PCR efficiency, but they are unlikely to affect the digital determination of endpoints. Furthermore, simultaneous measurement of different target sequences by qPCR is generally difficult because the PCR reaction of each sequence tends to affect the reaction of other sequences, but ddPCR has the advantage that simultaneous measurement is relatively easy. Next, to compensate for variations in DNA recovery, we focused on the addition of external standards, as described in the previous newsletter, which, like internal standards in LC/MS assays, can be added to each sample to correct for DNA recovery.

　We have constructed a LINE1-ddPCR system in which the external standard gene is added to each sample containing human cells, and the LINE1 sequence and the external standard gene sequence in the extracted solution are simultaneously measured by ddPCR[12]. The liver of SCID mice was used as the standard sample for the calibration curve, and each major tissue and blood were used as the quality control (QC) sample. The relative error and coefficient of variation were -18.9%-31.5% and 0.7%-33.1%, respectively, indicating excellent quantitative performance. The accuracy and precision of the system were significantly improved compared to those obtained by qPCR and those obtained without DNA recovery correction using an external standard gene, suggesting that the system was constructed as originally envisioned. The LINE1-ddPCR assay is also applicable to NOG and NSG mice, which are severely immunodeficient animals used in the development of cell therapy products, using a surrogate calibration curve prepared from SCID mouse liver. Since LINE1 is a human-specific sequence, it can be applied not only to mice but also to other animal species. In particular, the advantage of LINE1-ddPCR that a single surrogate matrix calibration curve can be used to quantify human cells in various biological samples will be maximized in biodistribution studies using species for which control animals are difficult to obtain, such as NHPs. On the other hand, the use of ddPCR has some disadvantages, such as dynamic range. In ddPCR, the copy number of the target sequence is calculated from the number of negative droplets in the total number of droplets, assuming a Poisson distribution. Therefore, the dynamic range is limited by the total number of droplets, resulting in a much narrower dynamic range compared to qPCR. In biodistribution studies of cell therapy products, there are cases where significant amplification in the body is confirmed [13] [14], and dynamic range is an extremely important factor in cell quantification. It is necessary to assume the number of cells in a biological sample before using this method.

　So far, we have described methods for quantification of human cells in biological samples by PCR. Are the results obtained by these methods comparable? Are the results comparable with those obtained by other methods, such as fluorescence, radioactivity, or FCM? To verify the equivalence between the quantitative methods, we measured the number of human cells in each sample within 30 hours after intravenous administration of ¹⁰⁶ cells/animal of human peripheral blood mononuclear cells (PBMCs) to SCID mice and compared the results. Alu-qPCR and LINE1-ddPCR results showed that the cell counts were higher in the lungs and liver, and disappeared more rapidly in the lungs than in other tissues, a profile that captures the kinetic characteristics of cell therapy products. This characteristic profile was also observed in a report that measured the radioactivity of ^51Cr-labeled murine splenocytes after intravenous administration to mice[15]. In addition, comparable results for human PBMC concentrations in blood have been obtained by FCM measurement. However, it is possible to obtain comparable results by understanding the advantages and disadvantages (points of concern) of each assay method in advance and designing a more appropriate study. In addition, the results of the PCR assay are not yet available.

　Finally, as a future issue, we would like to briefly discuss the standardization of the quantification of human cells in animal biological samples using the PCR method. Currently, although guidelines and white papers on PCR assays for gene therapy products have been reported [16] [17] [18] [19] [20], there are no guidelines or guidance on human cell quantification using the PCR method. As mentioned above, the existence of various methodologies within the PCR method is one of the reasons for the lack of guidance and guidelines. Even for a single reference material, there are various options, such as artificial sequences, extracted human genomic DNA, or cells themselves. In addition, there are many other issues to be discussed, such as calibration curves, tissue homogenate preparation methods, DNA extraction methods, confirmation methods for matrix effects, number of PCR reactions, qPCR or ddPCR, and criteria for required sensitivity, trueness, and accuracy [21]. It is necessary to reach a consensus on these points one by one in the future (Table 2). At that time, we hope that the PCR method will be standardized into an evaluation method that is suitable for the purpose of the biodistribution test and that can obtain results that enable kinetic analysis, without being overly constrained by the conventional wisdom of PCR measurement.

Table 2: Points to be considered when performing PCR

Item	PCR
Reference materials	Artificial sequences, human genomic DNA, cells
Calibration curve	Standard solution calibration curve, calibration curve with cell addition
Homogenate preparation	Partial tissue, whole tissue
DNA extraction	Silica membrane, magnetic beads, organic solvents
Confirmation of matrix effect	Add a certain amount of DNA to N1 among the samples to be measured, and confirm with QC samples
Number of PCR measurements	singlicate, duplicate, triplicate
Measuring instruments	qPCR, ddPCR
Unit	ng human gDNA/ng host gDNA, copies/ug gDNA, cells/g tissue or mL blood
Acceptance criteria	Accuracy, precision, sensitivity, recovery, selectivity, reproducibility

　How did we do? As mentioned above, various methods are used to measure human cells in animal samples for the development of cell therapy products, including fluorescence, radioactivity, magnetometry, FCM, and PCR. In this article, we focus on the PCR method, which does not require labeling of cells prior to administration and is highly sensitive and quantifiable. When you start quantification of human cells in animal samples in the future, please remember this article and make use of it. In the next issue, we will report on “Examples of cellular kinetics (CK)/biodistribution (BD) in non-clinical and clinical settings”. Please stay tuned!