A Fully Automatic Artificial Intelligence System Able to Detect and Characterize Prostate Cancer Using Multiparametric MRI: Multicenter and Multi-Scanner Validation.

In the last years, the widespread use of the prostate-specific antigen (PSA) blood examination to triage patients who will enter the diagnostic/therapeutic path for prostate cancer (PCa) has almost halved PCa-specific mortality. As a counterpart, millions of men with clinically insignificant cancer not destined to cause death are treated, with no beneficial impact on overall survival. Therefore, there is a compelling need to develop tools that can help in stratifying patients according to their risk, to support physicians in the selection of the most appropriate treatment option for each individual patient. The aim of this study was to develop and validate on multivendor data a fully automated computer-aided diagnosis (CAD) system to detect and characterize PCas according to their aggressiveness. We propose a CAD system based on artificial intelligence algorithms that a) registers all images coming from different MRI sequences, b) provides candidates suspicious to be tumor, and c) provides an aggressiveness score of each candidate based on the results of a support vector machine classifier fed with radiomics features. The dataset was composed of 131 patients (149 tumors) from two different institutions that were divided in a training set, a narrow validation set, and an external validation set. The algorithm reached an area under the receiver operating characteristic (ROC) curve in distinguishing between low and high aggressive tumors of 0.96 and 0.81 on the training and validation sets, respectively. Moreover, when the output of the classifier was divided into three classes of risk, i.e., indolent, indeterminate, and aggressive, our method did not classify any aggressive tumor as indolent, meaning that, according to our score, all aggressive tumors would undergo treatment or further investigations. Our CAD performance is superior to that of previous studies and overcomes some of their limitations, such as the need to perform manual segmentation of the tumor or the fact that analysis is limited to single-center datasets. The results of this study are promising and could pave the way to a prediction tool for personalized decision making in patients harboring PCa.

Augmented-reality robot-assisted radical prostatectomy using hyper-accuracy three-dimensional reconstruction (HA3D™) technology: a radiological and pathological study.

OBJECTIVES: To assess the use of hyper-accuracy three-dimensional (HA3D™; MEDICS, Moncalieri, Turin, Italy) reconstruction based on multiparametric magnetic resonance imaging (mpMRI) and superimposed imaging during augmented-reality robot-assisted radical prostatectomy (AR-RARP). PATIENTS AND METHODS: Patients with prostate cancer (clinical stages cT1-3, cN0, cM0) undergoing RARP at our Centre, from June 2017 to April 2018, were enrolled. In all cases, cancer was diagnosed with targeted biopsy at the level of index lesion based on high-resolution (1-mm slices) mpMRI. HA3D reconstruction was created by dedicated software to obtain the 3D virtual model of the prostate and surrounding structures. A specific system was used to overlay virtual data on the endoscopic video displayed by the remote da Vinci® surgical console (Intuitive Surgical Inc., Sunnyvale, CA, USA), and the virtual images were superimposed by the surgeon by the means of the TilePro™ multi-input display technology (Intuitive Surgical Inc.). The AR technology was used in four standardised key steps during RARP. The procedures were modulated differently in cases of prostate cancer without extracapsular extension (ECE) at mpMRI (Group A) or in cases of prostate cancer with ECE (Group B) at mpMRI. In Group A, the virtual image of the prostate was overlaid on the endoscopic view and the intraprostatic lesion was marked on the prostate surface by a metallic clip at the level of the suspicious lesion as identified by the 3D virtual AR image. In Group B, the same step was performed; moreover, a metallic clip was placed at the level of the suspicious ECE on the neurovascular bundles (NVBs) according to the virtual images. Finally, selective biopsies were taken from the NVBs at this level, and then, the entire NVBs were removed for final pathological examination, according to standard clinical indications. For Group A, the pathologist performed a targeted needle biopsy at the level of the metallic clip on the surface of prostate before the sample reduction. For Group B, the presence of tumour was evaluated during the reduction phase, at the level of metallic clip on the prostate surface and at the level of NVBs, sent separately. Finally, an image 3D scanner (Kinect, Microsoft) was used to perform a dimensional comparison between the mpMRI-based 3D virtual reconstruction and the whole-mount specimen. RESULTS: In all, 30 patients were enrolled in the present study, 11 (36.6%) included in Group A and 19 (63.4%) in Group B. In all cases (30/30), final pathology confirmed the location of the index lesion, as cancer was found at the level of the metallic clip. The suspected ECE was confirmed on final pathology in 15/19 cases (79%). The AR-guided selective biopsies at the level of the NVBs confirmed the ECE location, with 11/15 (73.3%) biopsies at the level of NVBs positive for cancer. The mismatch between the 3D virtual reconstruction and the prostate 3D scanning based on the whole-mount specimen was <3 mm in >85% of the gland. CONCLUSION: Our results suggest that a HA3D virtual reconstruction of the prostate based on mpMRI data and real-time superimposed imaging allow performance of an effective AR-RARP. Potentially, this approach translates into better outcomes, as the surgeon can tailor the procedure for each patient.

Indication to pelvic lymph nodes dissection for prostate cancer: the role of multiparametric magnetic resonance imaging when the risk of lymph nodes invasion according to Briganti updated nomogram is <5.

BACKGROUND: The Briganti updated nomogram (BN) is the most popular predictive model aiming to predict the presence of lymph node invasion (LNI) in patients with prostate cancer (PCa), but it lacks information obtained by preoperative imaging. The primary aim of the study was to evaluate the role of multiparametric prostate magnetic resonance imaging (mp-MRI) in the indication to perform pelvic lymph nodes dissection (PLND) or not in patients with risk of LNI according to BN below 5%. METHODS: Since March 2012 and September 2016, 310 patients who underwent a preoperative mp-MRI for staging purpose and subsequent robot-assisted extended PLND (RAEPLND) were retrospectively evaluated. Mp-MRIs were prospectively analyzed by two experienced radiologists. The imaging parameters analyzed were the presence of extracapsular extension (ECE), seminal vesicles invasion (SVI) and predominant Gleason pattern 4 (pG4). All patients underwent RAEPLND by two experienced surgeons with a standardized technique. A dedicated uropathologist performed all pathological analysis. Univariate analysis and multivariate logistic regression analysis were used in order to identify the predictors of LNI in patients with PCa. RESULTS: In the overall population, 57 (18.4%) patients had histologically proven pN1 disease. 48/250 patients (19.2%) with a risk of LNI ≥5% as calculated by the BN were staged pN1 at final histopathological analysis. 9/60 patients (15.0%) with a risk of LNI <5% as calculated by BN, who underwent RAEPLND anyway according to the findings at mp-MRI, were staged pN1 at final histopathological analysis. At multivariate logistic regression analysis, all the three mp-MRI parameters were significant independent predictors of LNI after RAEPLND. CONCLUSIONS: The role of mp-MRI seemed to be crucial in patients with a risk of LNI <5% as calculated by the BN. The presence of ECE, SVI, or pG4 at mp-MRI was found to be an independent predictor of LNI by itself.

Multiparametric Magnetic Resonance/Ultrasound Fusion Prostate Biopsy: Number and Spatial Distribution of Cores for Better Index Tumor Detection and Characterization.

PURPOSE: We evaluated the minimum core number for better index tumor detection to determine the best core site as well as biopsy Gleason score heterogeneity in the same index lesion. The aim was to optimize the highest Gleason score detection. MATERIALS AND METHODS: A total of 327 patients with negative digital rectal examination underwent magnetic resonance imaging/transrectal ultrasound fusion targeted biopsy for elevated/rising prostate specific antigen and/or 1 or more detectable lesions on multiparametric magnetic resonance imaging after a previous negative standard biopsy. Depending on the diameter of each index lesion (8 or less, or greater than 8 mm) 4 or 6 cores, respectively, were taken according to a well determined sequence. RESULTS: Of the patients 166 (50.7%) had prostate cancer, including 79 (47.6%) with an 8 mm or less index lesion and 87 (52.4%) with a greater than 8 mm index lesion. Of patients with an index tumor 8 mm or less 7 (8.9%) had 1, 31 (39.2%) had 2, 27 (34.2%) had 3 and 14 (17.7%) had 4 positive cores. Similarly, of patients with a lesion greater than 8 mm 8 (9.2%) had 1, 30 (34.5%) had 2, 13 (14.9%) had 3, 14 (16.1%) had 4, 12 (13.8%) had 5 and 10 (11.5%) had 6 positive cores. The major prevalence of positive cores was observed in the center of the target. Gleason score heterogeneity was found in 12.6% of those with an 8 mm or less target vs 26.4% with a target greater than 8 mm. In the center of the target there was a slight prevalence of Gleason pattern 4 or greater, or a lesser pattern. CONCLUSIONS: Approaching magnetic resonance imaging/transrectal ultrasound fusion targeted biopsy with a single core might be inadequate. Rather, taking 2 cores in the center of the index lesion may provide more accurate cancer detection and optimize the chances of finding the highest Gleason pattern.

Multiparametric-Magnetic Resonance/Ultrasound Fusion Targeted Prostate Biopsy Improves Agreement Between Biopsy and Radical Prostatectomy Gleason Score.

AIM: To investigate if targeted prostate biopsy (TBx) has superior performance to standard untargeted biopsy (SBx) in determining the optimal agreement between biopsy and surgical Gleason Score (GS). PATIENTS AND METHODS: An analysis of our institutional longitudinal database identified 683 consecutive patients who underwent either SBx (18-20 standardized transrectal ultrasound peripheral/transitional zone cores) or TBx alone (4-6 cores for each multiparametric magnetic resonance suspicious lesion, Prostate Imaging Reporting and Data System [(PI-RADS)≥3] after a previous negative first SBx. A total of 246 consecutive patients with diagnosis of prostate cancer (117 SBx and 129 TBx diagnoses) who underwent robot-assisted radical prostatectomy between January 2014 and December 2015, were enrolled. The concordance of biopsy GS to pathological GS, as well as the association between categorical variables [age, digital rectal exam (DRE), TNM, PI-RADS], were analyzed by Fisher's exact test. RESULTS: Prostate cancer was diagnosed in 32.0% of the SBx group and in 49.3% of TBx. The rate of correctly classified, up-graded and down-graded GS was 53.8% vs. 91.5%, 39.3% vs. 7.8% and 6.8% vs. 0.8% for SBx and TBx, respectively (p<0.001). The GS concordance rates for SBx and TBx cohorts were: 14.3% vs. 41.7% for GS 6, 61.0% vs. 83.8% for GS 3+4, 56.3% vs. 75.0% for GS 4+3, 27.3% vs. 100% for GS 8 and 80% vs. 100% for GS 9, respectively. CONCLUSION: TBx ensured a higher of accuracy of prostate cancer detection and a better performance in discriminating significant from insignificant prostate cancer, when compared to SBx. TBx significantly reduced the risk of GS up-/down-grading at radical prostatectomy for all histopathological categories. This is a notable advance in the selection of candidates for active surveillance.

Multiparametric magnetic resonance imaging and active surveillance: How to better select insignificant prostate cancer?

OBJECTIVES: To evaluate the role of multiparametric magnetic resonance imaging in improving the predictive accuracy of the Prostate Cancer Research International: Active Surveillance and Epstein criteria for active surveillance in prostate cancer. METHODS: A retrospective study was carried out with 126 prostate cancer patients treated with robot-assisted radical prostatectomy, but eligible for active surveillance according to the Prostate Cancer Research International: Active Surveillance criteria; 63 patients were also eligible according to the Epstein criteria. All patients underwent preoperative multiparametric magnetic resonance imaging, after at least 6 weeks from biopsy. The images from the multiparametric magnetic resonance imaging were assessed, and diagrams showing prostate sextants were used to designate regions of abnormalities within the prostate. Findings in the prostate were assigned to one of five categories according the Prostate Imaging-Reporting and Data System guidelines (v1.0), and considered positive for prostate cancer if the final Prostate Imaging-Reporting and Data System guidelines were >3 and negative if ≤3. Multivariate logistic regression analysis was carried out to evaluate the gain in accuracy of the Prostate Cancer Research International: Active Surveillance and Epstein criteria when added to multiparametric magnetic resonance imaging. Decision curve analysis was carried out to identify the net benefit of each model. RESULTS: The inclusion of multiparametric magnetic resonance imaging to the Epstein criteria and the Prostate Cancer Research International: Active Surveillance multivariate model significantly increased their accuracy in predicting pathologically-confirmed insignificant prostate cancer by 7% and 5%, respectively. At the decision curve analysis evaluation, the model including the Prostate Cancer Research International: Active Surveillance criteria and multiparametric magnetic resonance imaging improved the clinical risk prediction over the other models. CONCLUSIONS: The present findings suggest that multiparametric magnetic resonance imaging is able to increase the predictive accuracy of Prostate Cancer Research International: Active Surveillance and Epstein criteria to identify prostate cancer patients eligible for active surveillance.

Revisiting nephrocalcinosis: A single-centre perspective. A northern Italian experience.

AIM: Nephrocalcinosis is a clinical-pathological entity characterized by the deposition of calcium salts within the kidney parenchyma. Both the protean presentation and multiple causes may explain the lack of data regarding its prevalence. The aim of this study is to report the prevalence and main clinical features of nephrocalcinosis diagnosed in a newly opened nephrology outpatient unit. METHODS: Analysis on the data we prospectively gathered from the start of activity (2007-2013) was carried out. Clinical and laboratory data were collected from the medical records and from the general laboratory; diagnosis was based upon imaging data reviewed by the same radiologists. RESULTS: Sixty-five of 2695 patients referred to our unit were diagnosed with nephrocalcinosis (2.4%). The affected patients were younger than the overall out-patient population (median: 37.7 (min-max: 8-82) vs 63 years (2-102) P < 0.001), with higher female prevalence (68% vs 51.4%: P < 0.05) and better preserved kidney function (CKD-EPI 103 (23-165) vs 60 mL/min (3.2-169) P < 0.001). Kidney stones were the main reason for referral (35.4%), followed by electrolyte disturbances (22.7%), acute pyelonephritis (4.6%), AKI or CKD (4.6%). Nephrocalcinosis was associated with autoimmune diseases in 29% and with microcythaemia in 23%, while positive family history was present in 23% of patients. Various electrolyte disturbances were observed, with hypercalciuria being the hallmark of beta thalassaemic patients. CONCLUSIONS: Nephrocalcinosis is a rare, but not exceptional disease in nephrology. In Mediterranean countries, microcythaemia would appear to be a major cause of this disease. Greater awareness of nephrocalcinosis is needed for an integrated approach involving various branches of internal medicine and radiology.

Diabetic Kidney Disease: A Syndrome Rather Than a Single Disease.

The term "diabetic kidney" has recently been proposed to encompass the various lesions, involving all kidney structures that characterize protean kidney damage in patients with diabetes. While glomerular diseases may follow the stepwise progression that was described several decades ago, the tenet that proteinuria identifies diabetic nephropathy is disputed today and should be limited to glomerular lesions. Improvements in glycemic control may have contributed to a decrease in the prevalence of glomerular lesions, initially described as hallmarks of diabetic nephropathy, and revealed other types of renal damage, mainly related to vasculature and interstitium, and these types usually present with little or no proteinuria. Whilst glomerular damage is the hallmark of microvascular lesions, ischemic nephropathies, renal infarction, and cholesterol emboli syndrome are the result of macrovascular involvement, and the presence of underlying renal damage sets the stage for acute infections and drug-induced kidney injuries. Impairment of the phagocytic response can cause severe and unusual forms of acute and chronic pyelonephritis. It is thus concluded that screening for albuminuria, which is useful for detecting "glomerular diabetic nephropathy", does not identify all potential nephropathies in diabetes patients. As diabetes is a risk factor for all forms of kidney disease, diagnosis in diabetic patients should include the same combination of biochemical, clinical, and imaging tests as employed in non-diabetic subjects, but with the specific consideration that chronic kidney disease (CKD) may develop more rapidly and severely in diabetic patients.

Diffusion-weighted magnetic resonance imaging: new perspectives in the diagnostic pathway of non-complicated acute pyelonephritis.

OBJECTIVES: The diagnosis of acute pyelonephritis (APN) requires demonstration of parenchymal involvement. When no predisposing conditions are found, non-complicated APN is suspected and CT or MRI should be performed. Diffusion-weighted (DW) MRI might be useful, quicker and cheaper than the standard gadolinium-enhanced (GE) MRI. The aim of this study is to compare DW-MRI with GE-MRI to test its diagnostic accuracy in APN. METHODS: Of 318 consecutive patients hospitalised for APN, 279 underwent MRI. Four hundred and fourteen MR studies (first test and follow-up examinations) were gathered and data were processed using Diffusion Analysis software. DW-MRI has been compared with GE-MRI for evaluating diagnostic agreement. RESULTS: Two hundred and forty-four patients were diagnosed as having APN; 35 were negative. One hundred and sixty-three APN cases were considered non-complicated and selected for the study. Among the 414 MR examinations, comparing DW-MRI with GE-MRI, positive correlation was found in 258 cases, negative in 133. There were 14 false-negatives and 9 false-positives. DW-MRI achieved sensitivity 95.2 %, specificity 94.9 %, positive predictive value 96.9 %, negative predictive value 92.3 % and accuracy 94.6 %. CONCLUSIONS: DW-MRI is reliable for diagnosing non-complicated APN. The high diagnostic agreement between DW-MRI and GE-MRI offers new perspectives in diagnostic management, enabling diagnosis of non-complicated APN without using ionising radiation or contrast media. KEY POINTS: • The diagnosis of acute pyelonephritis (APN) requires demonstration of renal involvement. • Hitherto magnetic resonance imaging required gadolinium enhancement (GE-MRI) to establish this diagnosis. • But diagnostic agreement between diffusion-weighted and GE-MRI offers new diagnostic opportunities. • Quantification of ADC values can help diagnose and monitor APN. • DW-MRI avoids ionising radiation and paramagnetic contrast medium administration.

Excessive urinary tract dilatation and proteinuria in pregnancy: a common and overlooked association?

BACKGROUND: Proteinuria and dilatation of the urinary tract are both relatively common in pregnancy, the latter with a spectrum of symptoms, from none to severe pain and infection. Proteinuria is a rare occurrence in acute obstructive nephropathy; it has been reported in pregnancy, where it may pose a challenging differential diagnosis with pre-eclampsia.The aim of the present study is to report on the incidence of proteinuria (≥ 0.3; ≥ 0.5 g/day) in association with symptomatic-severe urinary tract dilatation in pregnancy. METHODS: Case series. SETTING: Nephrological-Obstetric Unit dedicated to pregnancy and kidney diseases (January 2000-April 2011). SOURCE: database prospectively updated since the start of the Unit. Retrospective review of clinical charts identified as relevant on the database, by a nephrologist and an obstetrician. RESULTS: From January 2000 to April 2011, 262 pregnancies were referred. Urinary tract dilatation with or without infection was the main cause of referral in 26 cases (predominantly monolateral in 19 cases): 23 singletons, 1 lost to follow-up, 1 twin and 1 triplet. Patients were referred for urinary tract infection (15 cases) and/or renal pain (10 cases); 6 patients were treated by urologic interventions ("JJ" stenting). Among them, 11 singletons and 1 triple pregnancy developed proteinuria ≥ 0.3 g/day (46.1%). Proteinuria was ≥ 0.5 g/day in 6 singletons (23.1%). Proteinuria resolved after delivery in all cases. No patient developed hypertension; in none was an alternative cause of proteinuria evident. No significant demographic difference was observed in patients with renal dilatation who developed proteinuria versus those who did not. An association with the presence of "JJ" stenting was present (5/6 cases with proteinuria ≥ 0.5 g/day), which may reflect both severer obstruction and a role for vescico-ureteral reflux, induced by the stent. CONCLUSIONS: Symptomatic urinary tract dilatation may be associated with proteinuria in pregnancy. This association should be kept in mind in the differential diagnosis with other causes of proteinuria in pregnancy, including pre-eclampsia.

Protean presentation and multiple challenges of nephrocalcinosis in pregnancy (six pregnancies in four patients).

BACKGROUND: Nephrocalcinosis is an umbrella term covering increased content of calcium salts in the renal parenchyma, interstitial damage and potential evolution towards renal failure. Pregnancy is often the first occasion for biochemical or imaging tests in young women and may allow early diagnosis. Conversely, even mild kidney disease may represent a challenge in pregnancy. AIM: The aim of this study was to report on four patients in whom nephrocalcinosis was first diagnosed during pregnancy, exemplifying the protean presentation and multiple challenges of nephrocalcinosis in pregnancy. METHODS: This is a case series study including data on all pregnancies prospectively gathered in the Nephrological-Obstetric Unit dedicated to pregnancy and kidney diseases (2000-11). RESULTS: Six pregnancies were observed in four patients (31-35 years; one twin pregnancy, one ongoing, one patient with three pregnancies). Symptoms were oedema in two (later developed in a further patient), renal functional impairment and electrolyte imbalance in two each. Two patients developed hypertension late in pregnancy. Electrolyte imbalance was life-threatening in one patient (severe acidosis, severe hyperkalaemia: 7.5 mEq/L). Delivery was by Caesarean section in three patients, preterm in one. Multiple or long hospitalizations for metabolic reasons were needed in three patients, the fourth was hospitalized for obstetric reasons. In all patients, diagnosis of nephrocalcinosis was made at ultrasounds during basic nephrological evaluation, confirmed at computerized tomography scan in three. The pathogenesis was linked to diuretic abuse in one case and to collagen disease, inborn errors and prematurity, possibly associated with diuretic misuse, in the others. CONCLUSION: Nephrocalcinosis may have protean presentations in pregnancy. The risk of severe electrolyte derangements, oedema and hypertension warrants strict clinical surveillance.

The clinical and imaging presentation of acute "non complicated" pyelonephritis: a new profile for an ancient disease.

BACKGROUND: Acute pyelonephritis (APN) is differently defined according to imaging or clinical criteria. In adults information on the relationship between imaging and clinical data is lacking.Our study was aimed at analysing the relationship between the clinical and imaging presentation of APN, defined according to imaging criteria (parenchymal involvement at MR or CT scan). METHODS: All consecutive patients hospitalized for "non-complicated" APN were considered (June 2005-December 2009). Clinical, biochemical and imaging data at hospitalization were analyzed by univariate and logistic regression analysis. RESULTS: There were 119 patients, all females, median age 32 years (15-72). At hospitalization, inflammatory markers were elevated (CRP median: 12.1 mg/dL, normal < 0.8). Incomplete presentations were frequent: fever was absent in 6.7%, pain in 17.8%, lower urinary tract symptoms in 52.9%. At CT or MR scan the lesions were bilateral in 12.6%, multiple in 79.8%; abscesses were present in 39.5%. Renal scars were found in 15.1%. Positive cultures were correlated with multiple foci (multivariate OR 4.2; CI 1.139-15.515). No other sign/symptom discriminated between small lesions, abscesses or multifocal involvement. CONCLUSIONS: APN is a protean disease. In the absence of strict correlation with clinical or biochemical markers, imaging studies are required to assess the severity of kidney involvement.