Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-2075
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-21T06:53:55.366515Z
status: published
lens: trace
sector: health
headline: MRI-Based Machine Learning to Differentiate True Progression from Pseudoprogression in Glioblastoma: An Updated Systematic Review and Meta-Analysis
dek: Background Differentiating true progression (TP) from pseudoprogression (PsP) after chemoradiation for glioblastoma remains clinically difficult. We evaluated externally validated diagnostic performance of magnetic resonance imaging (MRI)-based machine learning models and appraised study quality and transparency. Methods PubMed, Embase, MEDLINE, and Cochrane Library were searched through October 1, 2025 (PROSPERO registration CRD420251144894). Eligible adult glioblastoma studies used MRI-based machine learning t…
gain_title: In external test cohorts, MRI-based machine learning models distinguished true glioblastoma progression from pseudoprogression with pooled sensitivity of 0.78 and specificity of 0.75.
problem_title: The supporting evidence is weak: external cohorts were small, risk of bias was often high or unclear, and reporting was incomplete, so the models are not ready for routine clinical use.
trace_subject: MRI-based machine learning for glioblastoma true progression versus pseudoprogression
gain_reading: In external test cohorts, MRI-based machine learning models distinguished true glioblastoma progression from pseudoprogression with pooled sensitivity of 0.78 and specificity of 0.75.
gain_evidence: Pooled sensitivity was 0.78 (95% confidence interval [CI], 0.63-0.88), specificity was 0.75 (95% CI, 0.63-0.84) | MRI-based machine learning showed moderate-to-good external validation performance
problem_reading: The supporting evidence is weak: external cohorts were small, risk of bias was often high or unclear, and reporting was incomplete, so the models are not ready for routine clinical use.
problem_evidence: Risk of bias was frequently high or unclear, and reporting transparency was heterogeneous. | evidence remains limited by small external test cohorts, bias, and incomplete reporting
quick_read: A systematic review and meta-analysis searched four databases through October 1, 2025, and pooled studies in which machine learning applied to MRI was used to tell true tumor progression from pseudoprogression after chemoradiation in adults with glioblastoma. Thirty-one studies were included, but only seven external test cohorts had data suitable for the primary analysis.

This matters because telling these two conditions apart is clinically difficult after treatment, and the models showed useful but imperfect external accuracy. The authors say routine clinical use should wait for larger multicenter validations with transparent reporting, so how well the tools work in everyday practice is still unknown.
limitation: The external evidence base is small and often biased or incompletely reported, so the pooled accuracy may not hold up in routine care; larger transparent multicenter validations are still needed.
tag: Dual reading
key_points: The review included 31 studies covering 2,344 patients undergoing imaging assessment. | The primary analysis used seven external test cohorts with 219 patient observations. | The pooled summary AUC in external testing was 0.81 (bootstrap 95% CI, 0.69-0.91). | Internal test cohorts had a higher AUC (0.91) than external cohorts (0.81), but the difference was not statistically significant (P = 0.12). | Studies were checked against histopathology and/or clinicoradiologic follow-up.
rundown: The authors searched PubMed, Embase, MEDLINE, and Cochrane Library and registered the protocol in PROSPERO (CRD420251144894). They included adult glioblastoma studies in which MRI-based machine learning was tested against histopathology and/or clinicoradiologic follow-up.

Only external test cohorts with extractable 2 x 2 data entered the primary meta-analysis, which used bivariate random-effects models. Internal and combined internal and external analyses were secondary.

Internal performance looked better than external performance, though the gap was not statistically significant. Study quality was also a concern, as risk of bias was frequently high or unclear and reporting transparency varied.
sources:
- peer_reviewed | Neuro-Oncology | https://doi.org/10.1093/neuonc/noag259 | 2026-10-19
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

397ce5b5386e037d66183b9dc502615998df16671f7b55cd3db234fbebd6ae00

previous

0000000000000000000000000000000000000000000000000000000000000000