ObjectiveMonitoring biomarkers using machine learning (ML) may determine glioblastoma treatment response. We systematically reviewed quality and performance accuracy of recently published studies.MethodsFollowing Preferred Reporting Items for Systematic Reviews and Meta-Analysis: Diagnostic Test Accuracy, we extracted articles from MEDLINE, EMBASE and Cochrane Register between 09/2018–01/2021. Included study participants were adults with glioblastoma having undergone standard treatment (maximal resection, radiotherapy with concomitant and adjuvant temozolomide), and follow-up imaging to determine treatment response status (specifically, distinguishing progression/recurrence from progression/recurrence mimics, the target condition). Using Quality Assessment of Diagnostic Accuracy Studies Two/Checklist for Artificial Intelligence in Medical Imaging, we assessed bias risk and applicability concerns. We determined test set performance accuracy (sensitivity, specificity, precision, F1-score, balanced accuracy). We used a bivariate random-effect model to determine pooled sensitivity, specificity, area-under the receiver operator characteristic curve (ROC-AUC). Pooled measures of balanced accuracy, positive/negative likelihood ratios (PLR/NLR) and diagnostic odds ratio (DOR) were calculated. PROSPERO registered (CRD42021261965).ResultsEighteen studies were included (1335/384 patients for training/testing respectively). Small patient numbers, high bias risk, applicability concerns (particularly confounding in reference standard and patient selection) and low level of evidence, allow limited conclusions from studies. Ten studies (10/18, 56%) included in meta-analysis gave 0.769 (0.649-0.858) sensitivity [pooled (95% CI)]; 0.648 (0.749-0.532) specificity; 0.706 (0.623-0.779) balanced accuracy; 2.220 (1.560-3.140) PLR; 0.366 (0.213-0.572) NLR; 6.670 (2.800-13.500) DOR; 0.765 ROC-AUC.ConclusionML models using MRI features to distinguish between progression and mimics appear to demonstrate good diagnostic performance. However, study quality and design require improvement.

OBJECTIVEThis prospective case-control study was conducted to examine whether spherical deconvolution (SD) can unveil microstructural abnormalities in the corticospinal tract (CST) caused by IDH-mutant gliomas. To determine the significance of abnormal microstructure, the authors investigated the correlation between diffusion parameters and neurophysiological data collected with navigated transcranial magnetic stimulation (nTMS).METHODSTwenty participants (10 patients and 10 healthy controls) were recruited. Diffusion-weighted images were acquired on a 3-T MRI scanner using a cardiac-gated single-shot spin echo echo-planar imaging multiband sequence (TE 80 msec, TR 4000 msec) along 90 diffusion directions with a b-value of 2500 sec/mm2 (FOV 256 × 256 mm). Diffusion tensor imaging tractography and SD tractography were performed with deterministic tracking. The anterior portion of the ipsilateral superior peduncle and the precentral gyrus were used as regions of interest to delineate the CST. Diffusion indices were extracted and analyzed for significant differences between hemispheres in patients and between patient and control groups. A navigated brain stimulation system was used to deliver TMS pulses at hotspots at which motor evoked potentials (MEPs) for the abductor pollicis brevis, first digital interosseous, and abductor digiti minimi muscles are best elicited in patients and healthy controls. Functional measurements such as resting motor threshold (rMT), amplitude of MEPs, and latency of MEPs were noted. Significant differences between hemispheres in patients and between patients and controls were statistically analyzed. The Spearman rank correlation was used to investigate correlations between diffusion indices and functional measurements.RESULTSThe hindrance modulated orientational anisotropy (HMOA), measured with SD tractography, is lower in the hemisphere ipsilateral to glioma (p = 0.028). The rMT in the hemisphere ipsilateral to a glioma is significantly greater than that in the contralateral hemisphere (p = 0.038). All measurements contralateral to the glioma, except for the mean amplitude of MEPs (p = 0.001), are similar to those of healthy controls. Mean diffusivity and axial diffusivity from SD tractography are positively correlated with rMT in the hemisphere ipsilateral to glioma (p = 0.02 and 0.006, respectively). The interhemispheric difference in HMOA and rMT is correlated in glioma patients (p = 0.007).CONCLUSIONSSD tractography can demonstrate microstructural abnormality within the CST of patients with IDH1-mutant gliomas that correlates to the functional abnormality measured with nTMS.

Novel approaches for classification, including molecular features, are needed to direct therapy for men with low-grade prostate cancer (PCa), especially men on active surveillance. Risk alleles identified from genome-wide association studies (GWAS) could improve prognostication. Those risk alleles that coincided with genes and somatic copy number aberrations associated with progression of PCa were selected as the most relevant for prognostication.In a systematic literature review, a total of 698 studies were collated. Fifty-three unique SNPs residing in 29 genomic regions, including 8q24, 10q11 and 19q13, were associated with PCa progression. Functional studies implicated 21 of these single nucleotide polymorphisms (SNPs) as modulating the expression of genes in the androgen receptor pathway and several other oncogenes. In particular, 8q24, encompassing MYC, harbours a high density of SNPs conferring unfavourable pathological characteristics in low-grade PCa, while a copy number gain of MYC in low-grade PCa was associated with prostate-specific antigen recurrence after radical prostatectomy.By combining GWAS data with gene expression and structural rearrangements, risk alleles were identified that could provide a new basis for developing a prognostication tool to guide therapy for men with early prostate cancer.

The aim of the systematic review was to assess recently published studies on diagnostic test accuracy of glioblastoma treatment response monitoring biomarkers in adults, developed through machine learning (ML). Articles published 09/2018-09/2020 were searched for using MEDLINE, EMBASE, and the Cochrane Register. Included study participants were adult patients with high grade glioma who had undergone standard treatment (maximal resection, radiotherapy with concomitant and adjuvant temozolomide) and subsequently underwent follow-up imaging to determine treatment response status (specifically, distinguishing progression/recurrence from progression/recurrence mimics -the target condition). Risk of bias and applicability was assessed with QADAS 2 methodology. Contingency tables were created for hold-out test sets and recall, specificity, precision, F1-score, balanced accuracy calculated. Fifteen studies were included with 1038 patients in training sets and 233 in test sets. To determine whether there was progression or a mimic, the reference standard combination of follow-up imaging and histopathology at re-operation was applied in 67% (10/15) of studies. External hold-out test sets were used in 27% (4/15) to give ranges of diagnostic accuracy measures: recall = 0.70-1.00; specificity = 0.67-0.90; precision = 0.78-0.88; F1 score = 0.74-0.94; balanced accuracy = 0.74-0.83; AUC = 0.80-0.85. The small numbers of patient included in studies, the high risk of bias and concerns of applicability in the study designs (particularly in relation to the reference standard and patient selection due to confounding), and the low level of evidence, suggest that limited conclusions can be drawn from the data. There is likely good diagnostic performance of machine learning models that use MRI features to distinguish between progression and mimics. The diagnostic performance of ML using implicit features did not appear to be superior to ML using explicit features. There are a range of ML-based solutions poised to become treatment response monitoring biomarkers for glioblastoma. To achieve this, the development and validation of ML models require large, well-annotated datasets where the potential for confounding in the study design has been carefully considered. Therefore, multidisciplinary efforts and multicentre collaborations are necessary.