Abstract

Background: CRISPR-Cas systems have moved from a genome-engineering tool to a clinically deployed technology, with the first regulatory approval of a CRISPR-based therapeutic in December 2023 and a rapidly expanding body of CRISPR-based nucleic acid diagnostics. The primary evidence is now sufficiently mature that multiple systematic reviews with meta-analysis have been published in the diagnostic domain, creating both an opportunity for higher-order synthesis and a risk of redundant, overlapping reviews reporting near-identical estimates. Methods: Systematic reviews with meta-analysis evaluating the diagnostic accuracy of CRISPR-Cas-based assays against a reference standard were eligible. Reported pooled sensitivity and specificity were extracted and re-pooled in a DerSimonian-Laird random-effects model on the logit scale, following JBI guidance for synthesis at the review level and reported per PRISMA 2020. Results: Six systematic reviews with meta-analysis, collectively representing 150 primary diagnostic accuracy studies across SARSCoV- 2, Mycobacterium tuberculosis, hepatitis viruses and methicillin-resistant Staphylococcus aureus, met inclusion criteria. 95% prediction intervals were markedly wider than the confidence intervals (sensitivity 0.703-0.999), indicating that accuracy in a new pathogen setting is far less certain than the confidence interval alone implies. Leave-one-out analysis identified the tuberculosis review as the dominant heterogeneity source: its omission reduced I² from 71.2% to 0.0%. Subgroup analysis by target class showed no meaningful difference between viral (0.975, 95% CI 0.942-0.989) and bacterial (0.972, 95% CI 0.831- 0.996) targets. In the therapeutic domain, no randomised comparative trial data suitable for pooling were identified; the evidence comprises single-arm early-phase studies and one approved product. Conclusions: Therapeutic CRISPR evidence remains at a developmental stage incompatible with pooling. Registered, non-redundant reviews of prospectively conducted, clinically embedded accuracy studies are the priority need.

Keywords: CRISPR-Cas systems, Umbrella review, Diagnostic test accuracy, Gene editing, Meta-analysis, AMSTAR-2, Point-of-care testing