There is no suitable repository for systematic review of in vitro studies in which our protocol could be registered. We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.
Eligibility criteria
We included studies published after 1993 that evaluated the effect of temozolomide monotherapy on cell viability of at least one malignant glioma cell line. The evaluation of temozolomide could be a control experiment within a study. There was no limit on the type of cell viability measure. We excluded studies that utilised a drug-resistant cell line, a modified preparation of temozolomide, or an animal model. Studies using cell line-based xenotransplant animal models or reporting outcome measures other than cell viability were excluded.
Information sources
We searched Ovid Medline and Embase between January 1994 and January 2021 using a combination of search terms relating to gliomas, cell lines, temozolomide, and cell death. The full search strategies are available in Supplementary Material. Our search date was on 28th January 2021. There was no hand search or search in the grey literature.
Study selection & data extraction
The online tool Covidence was our platform for conducting the primary screening and data extraction in this review. We applied the built-in deduplication function on the platform to records retrieved from our database search. One reviewer (MB) screened titles and abstracts of all records, of which 10% excluded records were reviewed by a second reviewer (MTCP). We used the same review approach for full-text eligibility assessment. One reviewer (MB) extracted data from all eligible articles and a second reviewer (MTCP) independently extract data from 40% of these studies. We resolved disagreements by discussion between the two reviewers or by seeking resolution from a third reviewer (PMB).
Data items
We collected data on study characteristics, experimental setup and cell viability measures. Study characteristics included year of publication, country of primary affiliation, the primary aim of study, and types of cell lines used. ‘Primary aim of study’ had three categories: “therapeutic evaluation” referred to studies comparing the effects on cell lines of another therapeutic agent alone against temozolomide; “pathway modification” referred to studies that altered cellular pathways with a molecularly targeted compound and assessed the effect of temozolomide on the cell lines; “gene or protein measurement” referred to studies that measured levels of molecular markers in response to temozolomide. We categorised cell lines into human cell lines, murine cell lines, and cancer stem cell-like patient-derived cell lines. Patient-derived cell lines were developed from patients with malignant glioma at the investigators’ institute. Experimental setup data included temozolomide concentration, temozolomide exposure duration, base medium, addition of serum, cell density, cell passage number, use of hypoxic environment, and temperature and percentage of carbon dioxide in the ambient incubator environment.
Cell viability was our outcome of interest. Data on cell viability included the assay, quantification technique used and cell viability measurement. Cell viability assays included 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyl-2H-tetrazolium bromide (MTT), sulphondamine B (SRB), annexin V, propium iodide (PI), trypan blue, bromodeoxyuridine (BrdU) and luciferase-based techniques. Quantification techniques included flow cytometry, colorimetric or fluorometric plate reader, microscopy with manual counting, and microscopy with automated counting. Where reported, we collected drug sensitivity measures on half-maximal inhibitory concentration (IC50), half-maximal effective concentration (EC50), dose reduces initial population to 10% (LD10), dose reduces survival number of cells to 37% (D37), approximate concentration of drug tolerated without lethality (DT), and median-effect dose (Dm). To assess the internal validity, we assessed whether a study reported the number of replicate plates and the number of replicates per plate.
Risk of bias
There was no suitable risk of bias tool for the type of studies included in this review. The completeness of reporting of the data items described above was instead used to indicate the transparency of the studies to permit summarisation of comparable studies.
Data synthesis
We used descriptive statistics to summarise the data items. Our reporting of temozolomide sensitivity measures was stratified for the two most common cell lines (U87, U251) and patient-derived cell lines. We divided studies into groups that had the same cell lines exposed to temozolomide for the same duration, in a non-hypoxic environment, and reported the same drug sensitivity measures. We tested the homogeneity of variances in temozolomide sensitivity across cell lines using Levene’s and Fligner-Killeen tests. There was no plan for meta-analysis because studies were anticipated to be heterogeneous and drug sensitivity measures without a measure of variance are not amenable to meta-analysis. These reasons also precluded the assessment of reporting bias. We performed all analyses in R version 4.1.0 using ‘tidyverse’ (v1.3.1), ‘gtsummary’ (v1.4.1), and ‘car’ (v.3.0–10) packages.