**2. Tumor or target antigen-expressing cells related assay**

#### **2.1 Target screening and specificity evaluation**

Although the diversity of tumor cells leads to a high tumor escape rate in traditional single-target CAR-T, and new technologies are beginning to use bi-specific targets and even more complex designs, screening of specific target antigens is still the most important part of CAR-T design. MFC is the most important tool to achieve this purpose at present [3, 4, 17, 18]. The ideal CAR-T target should meet the following requirements: a high rate of occurrence in certain diseases (high availability of CAR-T), a high percentage of expression on tumor cells (covering all tumor cells to minimize relapse), a high intensity of expression (the expression intensity of target antigen on tumor cells is related to the efficacy, although there are contradictory results) [36**–**39], and good specificity (no or little expression in normal cells will not cause serious impact on patients) [1–4]. Target screening is performed on the basis of immunophenotyping or MRD, but with more stringent precautions than routine clinical diagnosis. It is necessary to accurately gate and define tumor cells, especially to cover all heterogeneous tumor cell populations. Otherwise, target-negative tumor cells will become the source of relapse. The detection of potential targets on tumor cells needs to be reported as a percentage. The mean fluorescence intensity (MFI) or median fluorescence intensity (MdFI) of target antigen expression in tumor cells and the ratio of MdFI to control cells may be used to describe the relative expression intensity. Some studies even use fluorescence microbeads for accurate quantitative

detection [36, 37]. Standardization of the operation and calibration of the instrument are required for the testing process, as well as a selection of appropriate internal controls.

### *2.1.1 Overview of the CAR-T targets in various tumors*

The target selection of ALL is relatively consistent. Generally, lineage markers with high coverage are preferred, such as CD19 [1–3, 36, 40, 41], CD22 [42], or CD19/ CD22 dual targets [37, 43] for B-ALL and CD7-CAR-T by genetic engineering technology [4, 44, 45] for T-ALL. Although some clinical studies have selected markers expressed in subgroups of ALL, such as CD20 [46] and CRLF2 [31] for B-ALL, and TRBC1 and CD1a for T-ALL [47, 48]. As to lymphoproliferative disease (LPD), other studies have tried to select lineage markers expressed by mature lymphocytes beside CD19 and CD22 [6, 37], for example, CD20, CD37, and BAFFR for B-LPD, CD5, CD4, TRBC1 for T-LPD, and CD38, BCMA or other markers for multiple myeloma (MM) [9, 10, 49**–**51]. Special subtypes of lymphoma have selected corresponding specific antigens as targets, such as CD30 for anaplastic large cell lymphoma (ALCL) and Hodgkin lymphoma (HL) [52**–**54]. Acute myeloid leukemia (AML) is a highly heterogeneous tumor, and there are a lot of studies on its targets, including CD33, CD123, CLL1 (CD371), CD25, CD117, Tim3, NKG2D, CD44, CD96, and CD38, or the combination of the above targets [5, 12, 17, 18, 55**–**60]. Although studies on solid tumors have made some achievements, searching for solid tumor-specific or associated antigens is still an interesting field for researchers. Therefore, current researches focus on immunosuppressive TME and modified CAR-T design [9, 10, 61, 62].

### *2.1.2 Standardized evaluation of target antigen expression*

Although some studies have shown that the efficacy of CAR-T is highly dependent on the density of target antigen expression [36, 37], and clinical trials has found that high tumor burden is a high-risk factor to relapse [40, 41], more detailed data remains unclear. For example, the percentage and numbers of antigen expressed on tumor cells and the expression intensity that can activate CAR-T to obtain the best response rate and longest survival rate; the suitable target antigen expression in tumor cells that allows the patient to be enrolled in the CAR-T study; the corresponding relations between absolute counts of target positive tumor cells or antigens on total tumor cells and dose of CAR-T needed for treatment [63, 64], etc. On the other side, the efficacy, stability, and difference of CAR detection antibody, qualitative and quantitative heterogeneity of antigen expression on tumor cells in the same disease, differences in antigen expression intensity caused by different fluorescence, and the influence factors in the process of antibody staining, can lead to significant intra-lab and inter-lab differences. Thus, the accurate relationship between the expression of CAR-T target antigens and the efficacy/side effects/survival rate is not comparable in different studies, which is more obvious in studies of weakly expressed target antigens. Given the diversity of CAR-T products and the multiple factors affecting MFC testing, uniform standard operating procedures (SOPs) may not be available in a short time. However, we hope to make the technique relatively stable and objective by standardizing the process of MFC detection, which will be helpful in exploring the most ideal target, accurately evaluating the efficacy and side effects of CAR-T, studying the complex relapse mechanism and promoting the update of CAR-T products to obtain the best effect for individual study [21–25]. To do this, the same protocol should be used

in a study, especially in a multicenter study, including antibody clone and fluorescein combination; titration and inter-batch comparison of the antibodies are required. Use the same instrument as far as possible, accurate comparisons are required if different instruments are used, and daily calibration and regular maintenance is also mandatory; residual normal counterparts in the specimen are good controls to evaluate the expression of target antigen with high or moderate intensity, such as CD19, CD22, and CD7, and we can describe target antigen as dim (dimmer than normal) or bright (brighter than normal) besides percentage; MFI/MdFI or quantitative fluorescence microbeads need to use to determine the expression of target antigen with very low intensity [37–39].

#### *2.1.3 Evaluation of the specificity of target antigen*

The ideal target antigen has been described above, where the specificity is evaluated by the expression of the antigen in normal cells, which is very important for CAR-T target selection. Because CAR-T is a very powerful and specific targeted therapy, most cells expressing the target will be killed, whether normal or malignant. Killing tumor cells is effective, while killing normal cells is toxic, not to say this effect lasts 1–3 months [3, 4, 34–37].

An ideal target is only specifically expressed in tumor cells but not in normal cells, or in normal cells the expression rate is low or the functions of these cells can be replaced by other cells or drugs. Unfortunately, almost no antigen is absolutely specific or low expressed in normal cells [9, 10] except those associated with B cells and plasma cells. Fortunately, with the development of modern life science, more and more gene modification methods and other technologies are overcoming this problem, such as the emergence of gene knockout or selecting CD7-CAR-T [4, 44, 45]. MFC plays an important role during the process. Accurate analysis of the target antigen expression on different cells can predict the toxicity and side effects, and help researchers to modify CAR-Ts.

#### **2.2 MRD**

MRD is an important indicator for evaluating efficacy and is closely related to prognosis [1–4, 19, 65]. MRD monitoring after CAR-T therapy is difficult due to tumor adaptation and off-target effects. The issues that need to be paid attention to are gating with multiple markers and recognizing malignant or normal cell loss target antigens [3, 19, 20].
