Artificial intelligence and radiologists in prostate cancer detection on MRI (PI-CAI): an international, paired, non-inferiority, confirmatory study.
Artificial intelligence (AI) systems can potentially aid the diagnostic pathway of prostate cancer by alleviating the increasing workload, preventing overdiagnosis, and reducing the dependence on experienced radiologists. We aimed to investigate the performance of AI systems at detecting clinically significant prostate cancer on MRI in comparison with radiologists using the Prostate Imaging-Reporting and Data System version 2.1 (PI-RADS 2.1) and the standard of care in multidisciplinary routine practice at scale.
In this international, paired, non-inferiority, confirmatory study, we trained and externally validated an AI system (developed within an international consortium) for detecting Gleason grade group 2 or greater cancers using a retrospective cohort of 10 207 MRI examinations from 9129 patients. Of these examinations, 9207 cases from three centres (11 sites) based in the Netherlands were used for training and tuning, and 1000 cases from four centres (12 sites) based in the Netherlands and Norway were used for testing. In parallel, we facilitated a multireader, multicase observer study with 62 radiologists (45 centres in 20 countries; median 7 [IQR 5-10] years of experience in reading prostate MRI) using PI-RADS (2.1) on 400 paired MRI examinations from the testing cohort. Primary endpoints were the sensitivity, specificity, and the area under the receiver operating characteristic curve (AUROC) of the AI system in comparison with that of all readers using PI-RADS (2.1) and in comparison with that of the historical radiology readings made during multidisciplinary routine practice (ie, the standard of care with the aid of patient history and peer consultation). Histopathology and at least 3 years (median 5 [IQR 4-6] years) of follow-up were used to establish the reference standard. The statistical analysis plan was prespecified with a primary hypothesis of non-inferiority (considering a margin of 0·05) and a secondary hypothesis of superiority towards the AI system, if non-inferiority was confirmed. This study was registered at ClinicalTrials.gov, NCT05489341.
Of the 10 207 examinations included from Jan 1, 2012, through Dec 31, 2021, 2440 cases had histologically confirmed Gleason grade group 2 or greater prostate cancer. In the subset of 400 testing cases in which the AI system was compared with the radiologists participating in the reader study, the AI system showed a statistically superior and non-inferior AUROC of 0·91 (95% CI 0·87-0·94; p<0·0001), in comparison to the pool of 62 radiologists with an AUROC of 0·86 (0·83-0·89), with a lower boundary of the two-sided 95% Wald CI for the difference in AUROC of 0·02. At the mean PI-RADS 3 or greater operating point of all readers, the AI system detected 6·8% more cases with Gleason grade group 2 or greater cancers at the same specificity (57·7%, 95% CI 51·6-63·3), or 50·4% fewer false-positive results and 20·0% fewer cases with Gleason grade group 1 cancers at the same sensitivity (89·4%, 95% CI 85·3-92·9). In all 1000 testing cases where the AI system was compared with the radiology readings made during multidisciplinary practice, non-inferiority was not confirmed, as the AI system showed lower specificity (68·9% [95% CI 65·3-72·4] vs 69·0% [65·5-72·5]) at the same sensitivity (96·1%, 94·0-98·2) as the PI-RADS 3 or greater operating point. The lower boundary of the two-sided 95% Wald CI for the difference in specificity (-0·04) was greater than the non-inferiority margin (-0·05) and a p value below the significance threshold was reached (p<0·001).
An AI system was superior to radiologists using PI-RADS (2.1), on average, at detecting clinically significant prostate cancer and comparable to the standard of care. Such a system shows the potential to be a supportive tool within a primary diagnostic setting, with several associated benefits for patients and radiologists. Prospective validation is needed to test clinical applicability of this system.
Health~Holland and EU Horizon 2020.
Saha A
，Bosma JS
，Twilt JJ
，van Ginneken B
，Bjartell A
，Padhani AR
，Bonekamp D
，Villeirs G
，Salomon G
，Giannarini G
，Kalpathy-Cramer J
，Barentsz J
，Maier-Hein KH
，Rusu M
，Rouvière O
，van den Bergh R
，Panebianco V
，Kasivisvanathan V
，Obuchowski NA
，Yakar D
，Elschot M
，Veltman J
，Fütterer JJ
，de Rooij M
，Huisman H
，PI-CAI consortium
... -
《-》
Effect of testing for cancer on cancer- or venous thromboembolism (VTE)-related mortality and morbidity in people with unprovoked VTE.
Venous thromboembolism (VTE) is a collective term for two conditions: deep vein thrombosis (DVT) and pulmonary embolism (PE). A proportion of people with VTE have no underlying or immediately predisposing risk factors and the VTE is referred to as unprovoked. Unprovoked VTE can often be the first clinical manifestation of an underlying malignancy. This has raised the question of whether people with an unprovoked VTE should be investigated for an underlying cancer. Treatment for VTE is different in cancer and non-cancer patients and a correct diagnosis would ensure that people received the optimal treatment for VTE to prevent recurrence and further morbidity. Furthermore, an appropriate cancer diagnosis at an earlier stage could avoid the risk of cancer progression and lead to improvements in cancer-related mortality and morbidity. This is the third update of the review first published in 2015.
To determine whether testing for undiagnosed cancer in people with a first episode of unprovoked VTE (DVT of the lower limb or PE) is effective in reducing cancer- or VTE-related mortality and morbidity and to determine which tests for cancer are best at identifying treatable cancers early.
The Cochrane Vascular Information Specialist searched the Cochrane Vascular Specialised Register, CENTRAL, MEDLINE, Embase and CINAHL databases and World Health Organization International Clinical Trials Registry Platform and ClinicalTrials.gov trials registers to 5 May 2021. We also undertook reference checking to identify additional studies.
Randomised and quasi-randomised trials in which people with an unprovoked VTE were allocated to receive specific tests for identifying cancer or clinically indicated tests only were eligible for inclusion.
Two review authors independently selected studies, assessed risk of bias and extracted data. We assessed the certainty of the evidence using GRADE criteria. We resolved any disagreements by discussion. The main outcomes of interest were all-cause mortality, cancer-related mortality and VTE-related mortality.
No new studies were identified for this 2021 update. In total, four studies with 1644 participants are included. Two studies assessed the effect of extensive tests including computed tomography (CT) scanning versus tests at the physician's discretion, while the other two studies assessed the effect of standard testing plus positron emission tomography (PET)/CT scanning versus standard testing alone. For extensive tests including CT versus tests at the physician's discretion, the certainty of the evidence, as assessed according to GRADE, was low due to risk of bias (early termination of the studies). When comparing standard testing plus PET/CT scanning versus standard testing alone, the certainty of evidence was moderate due to a risk of detection bias. The certainty of the evidence was downgraded further as detection bias was present in one study with a low number of events. When comparing extensive tests including CT versus tests at the physician's discretion, pooled analysis on two studies showed that testing for cancer was consistent with either benefit or no benefit on cancer-related mortality (odds ratio (OR) 0.49, 95% confidence interval (CI) 0.15 to 1.67; 396 participants; 2 studies; low-certainty evidence). One study (201 participants) showed that, overall, malignancies were less advanced at diagnosis in extensively tested participants than in participants in the control group. In total, 9/13 participants diagnosed with cancer in the extensively tested group had a T1 or T2 stage malignancy compared to 2/10 participants diagnosed with cancer in the control group (OR 5.00, 95% CI 1.05 to 23.76; low-certainty evidence). There was no clear difference in detection of advanced stages between extensive tests versus tests at the physician's discretion: one participant in the extensively tested group had stage T3 compared with four participants in the control group (OR 0.25, 95% CI 0.03 to 2.28; low-certainty evidence). In addition, extensively tested participants were diagnosed earlier than control group (mean: 1 month with extensive tests versus 11.6 months with tests at physician's discretion to cancer diagnosis from the time of diagnosis of VTE). Extensive testing did not increase the frequency of an underlying cancer diagnosis (OR 1.32, 95% CI 0.59 to 2.93; 396 participants; 2 studies; low-certainty evidence). Neither study measured all-cause mortality, VTE-related morbidity and mortality, complications of anticoagulation, adverse effects of cancer tests, participant satisfaction or quality of life. When comparing standard testing plus PET/CT screening versus standard testing alone, standard testing plus PET/CT screening was consistent with either benefit or no benefit on all-cause mortality (OR 1.22, 95% CI 0.49 to 3.04; 1248 participants; 2 studies; moderate-certainty evidence), cancer-related mortality (OR 0.55, 95% CI 0.20 to 1.52; 1248 participants; 2 studies; moderate-certainty evidence) or VTE-related morbidity (OR 1.02, 95% CI 0.48 to 2.17; 854 participants; 1 study; moderate-certainty evidence). Regarding stage of cancer, there was no clear difference for detection of early (OR 1.78, 95% 0.51 to 6.17; 394 participants; 1 study; low-certainty evidence) or advanced (OR 1.00, 95% CI 0.14 to 7.17; 394 participants; 1 study; low-certainty evidence) stages of cancer. There was also no clear difference in the frequency of an underlying cancer diagnosis (OR 1.71, 95% CI 0.91 to 3.20; 1248 participants; 2 studies; moderate-certainty evidence). Time to cancer diagnosis was 4.2 months in the standard testing group and 4.0 months in the standard testing plus PET/CT group (P = 0.88). Neither study measured VTE-related mortality, complications of anticoagulation, adverse effects of cancer tests, participant satisfaction or quality of life.
Specific testing for cancer in people with unprovoked VTE may lead to earlier diagnosis of cancer at an earlier stage of the disease. However, there is currently insufficient evidence to draw definitive conclusions concerning the effectiveness of testing for undiagnosed cancer in people with a first episode of unprovoked VTE (DVT or PE) in reducing cancer- or VTE-related morbidity and mortality. The results could be consistent with either benefit or no benefit. Further good-quality large-scale randomised controlled trials are required before firm conclusions can be made.
Robertson L
，Broderick C
，Yeoh SE
，Stansby G
... -
《Cochrane Database of Systematic Reviews》
ResearchGate
(全网免费下载)
钛学术
(全网免费下载)
