Table 1.
Quality assessment of included studies
| Risk of bias | Indirectness | Imprecision | ||||||
|---|---|---|---|---|---|---|---|---|
| Selection bias | Attrition bias | Reporting/Detection bias | ||||||
| First Author | IRC involved in patient selection (Yes; No) |
Loss to follow-up (< 5%; 5–20%; >20%) |
Objective outcomes assessed (Yes; No) | IRC involved in assessment of response (Yes; No) | Safety outcomes reported (Yes; No) |
Heterogeneity (Single sub-type; 2 sub-types; >2 sub-types in the study) |
Sample size (< 30; 30–50; >50 patients treated) |
Duration of follow-up (< 6 months; 6–12 months; >12 months) |
| Studies with single dose and fractionated-dose patients | ||||||||
| Frey NV(37) | No* | > 20% | Yes | No | Yes | Single sub-type | 30–50 | > 12 months |
| Frey NV(35) | No* | 5–20% | Yes | No | Yes | Single sub-type | 30–50 | > 12 months |
| Xu J(36) | No* | Consort Diagram Not Reported | Yes | No | Yes | Single sub-type | < 30 | > 12 months |
| Anti-CD19 CAR-T cells | ||||||||
| Li M(48) | No* | > 20% | Yes | No | Yes | Single sub-type | > 50 | NR |
| Jiang H(22) | No* | Consort Diagram Not Reported | Yes | No | Yes | Single sub-type | > 50 | NR |
| Ying Z(28) | No* | Consort Diagram Not Reported | Yes | No | Yes | > 2 sub-types | < 30 | NR |
| Roddie C(40) | No* | Consort Diagram Not Reported | Yes | No | Yes | Single sub-type | < 30 | > 12 months |
| Wang CM(45) | No | Consort Diagram Not Reported | Yes | No | Yes | > 2 sub-types | < 30 | NR |
| Geyer MB(38) | No | > 20% | Yes | No | Yes | > 2 sub-type | < 30 | > 12 months |
| Davila ML(41) | No | < 5% | Yes | No | Yes | Single sub-type | < 30 | > 12 months |
| Sauter CS(39) | No | Consort Diagram Not Reported | Yes | No | Yes | > 2 sub-types | < 30 | > 12 months |
| Hu Y(46) | No | 5–20% | Yes | No | Yes | Single sub-type | < 30 | < 6 months |
| Porter DL(47) | No* | > 20% | Yes | No | Yes | Single sub-type | < 30 | > 12 months |
| Geyer MB(27) | No | Consort Diagram Not Reported | Yes | No | Yes | Single sub-type | < 30 | > 12 months |
| Zhang Q(29) | No* | Consort Diagram Not Reported | Yes | No | No | Single sub-type | < 30 | NR |
| Kalos M(44) | No* | Consort Diagram Not Reported | Yes | No | Yes | Single sub-type | < 30 | NR |
| Anti-BCMA CAR-T cells | ||||||||
| Zhao WH(43) | No* | Consort Diagram Not Reported | Yes | No | Yes | Single sub-type | > 50 | 6–12 months |
| Cohen AD(42) | No | 5–20% | Yes | Yes | Yes | Single sub-type | < 30 | > 12 months |
* Independent review committee/board approved the study’s protocol and had patients sign consent forms
IRC, independent review committee
All observational and single arm unblinded studies are given low grade and the grade is moved upwards based on quality assessment [48–50]
Risk of Bias mainly involves selection bias and reporting or detection bias. Selection bias is low, and quality is high for studies that included an IRC for patient selection and that had < 5% loss of patients to follow-up. Studies with 5–20% loss to follow-up are considered to have medium selection bias and studies with over 20% loss to follow-up are considered to have high selection bias
Reporting or detection bias is considered low for studies that evaluated objective outcomes, included an IRC for response assessment, and reported treatment-related adverse events (safety). Studies that reported subjective outcomes (e.g. patient reported outcomes) or studies that did not include IRC for response assessment or studies that did not report safety outcomes are rated as high for reporting or detection bias
Indirectness (comparability) of the cohort between studies is considered low and quality is also high for studies that have a homogenous cohort (single type of cancer). Studies with up to 2 cancer-subtypes are rated as medium for indirectness and with > 2 cancer-subtypes are rated as low for comparability
Imprecision of the cohort is considered high and quality is low for studies that have low sample size (< 30 patients) and small follow-up (< 6 months). Studies that have a sample size of 30–50 patients or with 6–12 months follow-up are rated medium for imprecision. Studies with sample size of > 50 patients and with follow-up over 12 months are rated low for imprecision and high for quality