An increasing volume of prostate biopsies and a worldwide shortage of urological pathologists puts a strain on pathology departments. Additionally, the high intra-observer and inter-observer ...variability in grading can result in overtreatment and undertreatment of prostate cancer. To alleviate these problems, we aimed to develop an artificial intelligence (AI) system with clinically acceptable accuracy for prostate cancer detection, localisation, and Gleason grading.
We digitised 6682 slides from needle core biopsies from 976 randomly selected participants aged 50–69 in the Swedish prospective and population-based STHLM3 diagnostic study done between May 28, 2012, and Dec 30, 2014 (ISRCTN84445406), and another 271 from 93 men from outside the study. The resulting images were used to train deep neural networks for assessment of prostate biopsies. The networks were evaluated by predicting the presence, extent, and Gleason grade of malignant tissue for an independent test dataset comprising 1631 biopsies from 246 men from STHLM3 and an external validation dataset of 330 biopsies from 73 men. We also evaluated grading performance on 87 biopsies individually graded by 23 experienced urological pathologists from the International Society of Urological Pathology. We assessed discriminatory performance by receiver operating characteristics and tumour extent predictions by correlating predicted cancer length against measurements by the reporting pathologist. We quantified the concordance between grades assigned by the AI system and the expert urological pathologists using Cohen's kappa.
The AI achieved an area under the receiver operating characteristics curve of 0·997 (95% CI 0·994–0·999) for distinguishing between benign (n=910) and malignant (n=721) biopsy cores on the independent test dataset and 0·986 (0·972–0·996) on the external validation dataset (benign n=108, malignant n=222). The correlation between cancer length predicted by the AI and assigned by the reporting pathologist was 0·96 (95% CI 0·95–0·97) for the independent test dataset and 0·87 (0·84–0·90) for the external validation dataset. For assigning Gleason grades, the AI achieved a mean pairwise kappa of 0·62, which was within the range of the corresponding values for the expert pathologists (0·60–0·73).
An AI system can be trained to detect and grade cancer in prostate needle biopsy samples at a ranking comparable to that of international experts in prostate pathology. Clinical application could reduce pathology workload by reducing the assessment of benign biopsies and by automating the task of measuring cancer length in positive biopsy cores. An AI system with expert-level grading performance might contribute a second opinion, aid in standardising grading, and provide pathology expertise in parts of the world where it does not exist.
Swedish Research Council, Swedish Cancer Society, Swedish eScience Research Center, EIT Health.
Particulate matter 2.5 (PM2.5) is a risk factor for lung cancer. In this study, we investigated the molecular mechanisms of PM2.5 exposure on lung cancer progression. We found that short‐term ...exposure to PM2.5 for 24 h activated the EGFR pathway in lung cancer cells (EGFR wild‐type and mutant), while long‐term exposure of lung cancer cells to PM2.5 for 90 days persistently promoted EGFR activation, cell proliferation, anchorage‐independent growth, and tumor growth in a xenograft mouse model in EGFR‐driven H1975 cancer cells. We showed that PM2.5 activated AhR to translocate into the nucleus and promoted EGFR activation. AhR further interacted with the promoter of TMPRSS2, thereby upregulating TMPRSS2 and IL18 expression to promote cancer progression. Depletion of TMPRSS2 in lung cancer cells suppressed anchorage‐independent growth and xenograft tumor growth in mice. The expression levels of TMPRSS2 were found to correlate with nuclear AhR expression and with cancer stage in lung cancer patient tissue. Long‐term exposure to PM2.5 could promote tumor progression in lung cancer through activation of EGFR and AhR to enhance the TMPRSS2‐IL18 pathway.
Synopsis
PM2.5 promotes lung cancer progression through activation of the AhR‐TMPRSS2‐IL18.
Exposure to PM2.5 activates EGFR pathway and promotes lung cancer progression.
Long‐term exposure to PM2.5 increases lung cancer cell proliferation, anchorage‐independent growth, and xenograft tumor growth in mice.
PM2.5 activates AhR to translocate into the nucleus and upregulates the expression of TMPRSS2.
Depletion of TMPRSS2 in lung cancer cells suppresses anchorage‐independent growth and xenograft tumor growth in mice.
TMPRSS2 upregulates IL I8 expression and promotes lung cancer progression.
PM2.5 promotes lung cancer progression through activation of the AhR‐TMPRSS2‐IL18.
Sclerosing pneumocytoma is a unique benign neoplasm of the lungs. The molecular alterations in sclerosing pneumocytoma are not well understood. In a previous whole-exome sequencing study, recurrent ...AKT1 point mutation was observed in about half of the cases of sclerosing pneumocytoma. However, in the remaining half, cancer-related mutations have still not been identified. In this study, we first analyzed the raw sequence data from the previous whole-exome sequencing study (PRJNA297066 cohort). Using Genomon-ITDetector, a special software for detection of internal tandem duplications, we identified recurrent internal tandem duplications in the AKT1 gene in 22 of the 44 tumor samples (50%). All the cases positive for AKT1 internal tandem duplications lacked AKT1 point mutations. Next, we performed targeted next-generation sequencing in an independent cohort of sclerosing pneumocytoma from our hospital (VGH-TPE cohort), and again identified recurrent AKT1 internal tandem duplications in 20 of the 40 (50%) tumor samples analyzed. The internal tandem duplications resulted in duplications of 7 to 16 amino acids in a narrow region of the Pleckstrin homology domain of the AKT1 protein. This region contains the interaction interface between the Pleckstrin homology and kinase domains, which is known to play a critical role in the activation of the AKT1 protein. Moreover, we found that AKT1 internal tandem duplications were mutually exclusive of other forms of AKT1 mutations, including point mutations and short indels. Taking all forms of AKT1 mutations together, we detected AKT1 mutations in almost all the sclerosing pneumocytomas in our study (PRJNA297066 cohort: 41 out of 44 cases, 93%; VGH-TPE cohort: 40 out of 40 cases, 100%). Our results suggest that AKT1 mutation is the genetic hallmark of sclerosing pneumocytoma. These results would help in better understanding of the pathogenesis of sclerosing pneumocytoma.
The Gleason score is the most important prognostic marker for prostate cancer patients, but it suffers from significant observer variability. Artificial intelligence (AI) systems based on deep ...learning can achieve pathologist-level performance at Gleason grading. However, the performance of such systems can degrade in the presence of artifacts, foreign tissue, or other anomalies. Pathologists integrating their expertise with feedback from an AI system could result in a synergy that outperforms both the individual pathologist and the system. Despite the hype around AI assistance, existing literature on this topic within the pathology domain is limited. We investigated the value of AI assistance for grading prostate biopsies. A panel of 14 observers graded 160 biopsies with and without AI assistance. Using AI, the agreement of the panel with an expert reference standard increased significantly (quadratically weighted Cohen's kappa, 0.799 vs. 0.872; p = 0.019). On an external validation set of 87 cases, the panel showed a significant increase in agreement with a panel of international experts in prostate pathology (quadratically weighted Cohen's kappa, 0.733 vs. 0.786; p = 0.003). In both experiments, on a group-level, AI-assisted pathologists outperformed the unassisted pathologists and the standalone AI system. Our results show the potential of AI systems for Gleason grading, but more importantly, show the benefits of pathologist-AI synergy.
Aims
Papillary renal neoplasm with reverse polarity (PRNRP) is a newly defined entity with distinct histomorphology and recurrent KRAS mutation. It has been estimated to constitute 4% of previously ...diagnosed papillary renal cell carcinoma (PRCC). Renal papillary adenoma (PA) is suggested to be the precursor of PRCC. This study aimed to investigate the association between PRNRP and PA, particularly the morphologically similar type D PA.
Methods and results
Nephrectomy specimens of PRCC and PA from our 10‐year pathology archives were retrieved and reviewed. GATA3 immunohistochemistry and RAS/BRAF testing were performed in all cases reclassified as PRNRP and all PAs with sufficient materials. Overall, PRNRP accounted for 9.1% (10 of 110) of PRCC and there was no recurrence/metastasis with a mean follow‐up period of 39 months. Three novel morphological features were described, including clear cell change, mast cell infiltration and metaplastic ossification. Nine of the 10 PRNRPs showed diffuse and strong GATA3 expression and KRAS missense mutations at codon 12. One case exhibited moderate GATA3 staining on 80% of the tumour cells and RAS/BRAF wild‐type. In a total of 73 PAs, 11 were classified as type D. GATA3 expression was significantly more frequent in type D versus non‐type D PAs (100 versus 35%, P < 0.01). KRAS missense mutations were identified in six of eight (75%) of the type D PAs but none of the 18 non‐type D PAs.
Conclusions
Type D PA was different from other types of PA and represented an analogue or a small‐sized PRNRP for their identical morphology, immunophenotype and molecular signature.
ALK
-rearranged renal cell carcinoma is a provisional entity in the 2016 WHO Classification of Tumors of the Urinary System and Male Genital Organs. The reported fusion partners included
VCL
,
TPM3
,
...EML4
,
STRN
, and
HOOK1
. Herein, we present a peculiar renal cell carcinoma morphologically resembling metanephric adenoma and harboring a novel
PLEKHA7
-
ALK
fusion. Microscopically, the tumor is composed of bland epithelial cells with scant to moderate amount of amphophilic cytoplasm, round and uniform nuclei, delicate chromatin, and inconspicuous nucleoli, arranged in tightly packed small acini and angulated tubules. Papillary formation, intraluminal glomeruloid tufts, microcysts, and solid nests were focally observed. Psammomatous calcifications were evident. The tumor cells were diffusely reactive for CK7, AMACR, PAX8, and ALK, while non-reactive for WT1, BRAF V600E, CD57, carbonic anhydrase IX, TFE3, and cathepsin K. Fluorescence in situ hybridization showed breaking apart of
ALK
. A novel
PLEKHA7
exon18-
ALK
exon20 fusion was detected using ArcherDX FusionPlex next-generation sequencing panel and was further confirmed with reverse-transcriptase PCR. Our case demonstrates that in contrast to prior cases showing high-grade tumor cells,
ALK
-rearranged renal cell carcinoma may also present as a low-grade renal tumor mimicking metanephric adenoma. Immunohistochemistry and molecular testing are helpful to identify this tumor, which may be eligible for ALK inhibitor-targeted therapy.
•We surveyed the use of machine learning to inform predictive models in mood disorders.•We include studies that use machine learning algorithms to identify predictors of therapeutic outcomes in ...uni/bipolar depression.•Classification algorithms informed by neuroimaging, phenomenological, and genetic data were able to predict therapeutic outcomes with an overall accuracy of 0.82.•Predictive models integrating multiple data types performed better when compared to models with single lower-dimension data types (p <0.01).•Machine learning provides opportunity to parse clinical heterogeneity and characterize moderators of disease risk and trajectory.
No previous study has comprehensively reviewed the application of machine learning algorithms in mood disorders populations. Herein, we qualitatively and quantitatively evaluate previous studies of machine learning-devised models that predict therapeutic outcomes in mood disorders populations.
We searched Ovid MEDLINE/PubMed from inception to February 8, 2018 for relevant studies that included adults with bipolar or unipolar depression; assessed therapeutic outcomes with a pharmacological, neuromodulatory, or manual-based psychotherapeutic intervention for depression; applied a machine learning algorithm; and reported predictors of therapeutic response. A random-effects meta-analysis of proportions and meta-regression analyses were conducted.
We identified 639 records: 75 full-text publications were assessed for eligibility; 26 studies (n=17,499) and 20 studies (n=6325) were included in qualitative and quantitative review, respectively. Classification algorithms were able to predict therapeutic outcomes with an overall accuracy of 0.82 (95% confidence interval CI of 0.77, 0.87). Pooled estimates of classification accuracy were significantly greater (p < 0.01) in models informed by multiple data types (e.g., composite of phenomenological patient features and neuroimaging or peripheral gene expression data; pooled proportion 95% CI = 0.930.86, 0.97) when compared to models with lower-dimension data types (pooledproportion=0.680.62,0.74to0.850.81,0.88).
Most studies were retrospective; differences in machine learning algorithms and their implementation (e.g., cross-validation, hyperparameter tuning); cannot infer importance of individual variables fed into learning algorithm.
Machine learning algorithms provide a powerful conceptual and analytic framework capable of integrating multiple data types and sources. An integrative approach may more effectively model neurobiological components as functional modules of pathophysiology embedded within the complex, social dynamics that influence the phenomenology of mental disorders.
ALK rearranged renal cell carcinoma (ALK-RCC) has recently been included in 2016 WHO classification as a provisional entity. In this study, we describe 12 ALK-RCCs from 8 institutions, with detailed ...clinical, pathological, immunohistochemical (IHC), fluorescence in situ hybridization (FISH), and next generation sequencing (NGS) analyses. Patients' age ranged from 25 to 68 years (mean, 46.3 years). Seven patients were females and five were males (M:F = 1:1.4). Tumor size ranged from 17 to 70 mm (mean 31.5, median 25 mm). The pTNM stage included: pT1a (n = 7), pT1b (n = 1), and pT3a (n = 4). Follow-up was available for 9/12 patients (range: 2 to 153 months; mean 37.6 months); 8 patients were alive without disease and one was alive with distant metastases. The tumors demonstrated heterogeneous, ‘difficult to classify' morphology in 10/12 cases, typically showing diverse architectural and cellular morphologies, including papillary, tubular, tubulocystic, solid, sarcomatoid (spindle cell), rhabdoid, signet-ring cell, and intracytoplasmic vacuoles, often set in a mucinous background. Of the remaining two tumors, one showed morphology resembling mucinous tubular and spindle cell renal cell carcinoma (MTSC RCC-like) and one was indistinguishable from metanephric adenoma. One additional case also showed a focal metanephric adenoma-like area, in an otherwise heterogeneous tumor. By IHC, all tumors were diffusely positive for ALK and PAX8. In both cases with metanephric adenoma-like features, WT1 and ALK were coexpressed. ALK rearrangement was identified in 9/11 tumors by FISH. ALK fusion partners were identified by NGS in all 12 cases, including the previously reported: STRN (n = 3), TPM3 (n = 3), EML4 (n = 2), and PLEKHA7 (n = 1), and also three novel fusion partners: CLIP1 (n = 1), KIF5B (n = 1), and KIAA1217 (n = 1). ALK-RCC represents a genetically distinct entity showing a heterogeneous histomorphology, expanded herein to include unreported metanephric adenoma-like and MTSC RCC-like variants. We advocate a routine ALK IHC screening for “unclassifiable RCCs” with heterogeneous features.
Metastasis is a predominant cause of death for prostate cancer (PCa) patients; however, the underlying mechanisms are poorly understood. We report that monoamine oxidase A (MAOA) is a clinically and ...functionally important mediator of PCa bone and visceral metastases, activating paracrine Shh signaling in tumor-stromal interactions. MAOA provides tumor cell growth advantages in the bone microenvironment by stimulating interleukin-6 (IL6) release from osteoblasts, and triggers skeletal colonization by activating osteoclastogenesis through osteoblast production of RANKL and IL6. MAOA inhibitor treatment effectively reduces metastasis and prolongs mouse survival by disengaging the Shh-IL6-RANKL signaling network in stromal cells in the tumor microenvironment. These findings provide a rationale for targeting MAOA and its associated molecules to treat PCa metastasis.
Display omitted
•MAOA is associated with prostate cancer metastasis in clinical specimens•MAOA promotes metastasis by activating the paracrine Shh-IL6-RANKL signaling•MAOA drives tumor-stromal cell interactions in a vicious-cycle manner•MAOA inhibitor treatment reduces metastasis and prolongs survival in mice
Wu et al. show that monoamine oxidase A (MAOA) is an important mediator of prostate cancer bone and visceral metastases by activating paracrine Shh-IL6-RANKL signaling in tumor-stromal interactions. Pharmacological inhibition of MAOA restricts metastasis and extends survival in a mouse prostate cancer model.