Crowdsourcing image annotation for nucleus detection and segmentation in computational pathology: evaluating experts, automated methods, and the crowd.

The development of tools in computational pathology to assist physicians and biomedical scientists in the diagnosis of disease requires access to high-quality annotated images for algorithm learning and evaluation. Generating high-quality expert-derived annotations is time-consuming and expensive. We explore the use of crowdsourcing for rapidly obtaining annotations for two core tasks in com- putational pathology: nucleus detection and nucleus segmentation. We designed and implemented crowdsourcing experiments using the CrowdFlower platform, which provides access to a large set of labor channel partners that accesses and manages millions of contributors worldwide. We obtained annotations from four types of annotators and compared concordance across these groups. We obtained: crowdsourced annotations for nucleus detection and segmentation on a total of 810 images; annotations using automated methods on 810 images; annotations from research fellows for detection and segmentation on 477 and 455 images, respectively; and expert pathologist-derived annotations for detection and segmentation on 80 and 63 images, respectively. For the crowdsourced annotations, we evaluated performance across a range of contributor skill levels (1, 2, or 3). The crowdsourced annotations (4,860 images in total) were completed in only a fraction of the time and cost required for obtaining annotations using traditional methods. For the nucleus detection task, the research fellow-derived annotations showed the strongest concordance with the expert pathologist- derived annotations (F-M =93.68%), followed by the crowd-sourced contributor levels 1,2, and 3 and the automated method, which showed relatively similar performance (F-M = 87.84%, 88.49%, 87.26%, and 86.99%, respectively). For the nucleus segmentation task, the crowdsourced contributor level 3-derived annotations, research fellow-derived annotations, and automated method showed the strongest concordance with the expert pathologist-derived annotations (F-M = 66.41%, 65.93%, and 65.36%, respectively), followed by the contributor levels 2 and 1 (60.89% and 60.87%, respectively). When the research fellows were used as a gold-standard for the segmentation task, all three con- tributor levels of the crowdsourced annotations significantly outperformed the automated method (F-M = 62.21%, 62.47%, and 65.15% vs. 51.92%). Aggregating multiple annotations from the crowd to obtain a consensus annotation resulted in the strongest performance for the crowd-sourced segmentation. For both detection and segmentation, crowd-sourced performance is strongest with small images (400 × 400 pixels) and degrades significantly with the use of larger images (600 × 600 and 800 × 800 pixels). We conclude that crowdsourcing to non-experts can be used for large-scale labeling microtasks in computational pathology and offers a new approach for the rapid generation of labeled images for algorithm development and evaluation.

Blocks to thyroid cancer cell apoptosis can be overcome by inhibition of the MAPK and PI3K/AKT pathways.

Current treatment for recurrent and aggressive/anaplastic thyroid cancers is ineffective. Novel targeted therapies aimed at the inhibition of the mutated oncoprotein BRAF(V600E) have shown promise in vivo and in vitro but do not result in cellular apoptosis. TNF-related apoptosis-inducing ligand (TRAIL) induces apoptosis in a tumor-selective manner by activating the extrinsic apoptotic pathway. Here, we show that a TRAIL-R2 agonist antibody, lexatumumab, induces apoptosis effectively in some thyroid cancer cell lines (HTh-7, TPC-1 and BCPAP), while more aggressive anaplastic cell lines (8505c and SW1736) show resistance. Treatment of the most resistant cell line, 8505c, using lexatumumab in combination with the BRAF(V600E) inhibitor, PLX4720, and the PI3K inhibitor, LY294002, (triple-drug combination) sensitizes the cells by triggering both the extrinsic and intrinsic apoptotic pathways in vitro as well as 8505c orthotopic thyroid tumors in vivo. A decrease in anti-apoptotic proteins, pAkt, Bcl-xL, Mcl-1 and c-FLIP, coupled with an increase in the activator proteins, Bax and Bim, results in an increase in the Bax to Bcl-xL ratio that appears to be critical for sensitization and subsequent apoptosis of these resistant cells. Our results suggest that targeting the death receptor pathway in thyroid cancer can be a promising strategy for inducing apoptosis in thyroid cancer cells, although combination with other kinase inhibitors may be needed in some of the more aggressive tumors initially resistant to apoptosis.

killerFLIP: a novel lytic peptide specifically inducing cancer cell death.

One of the objectives in the development of effective cancer therapy is induction of tumor-selective cell death. Toward this end, we have identified a small peptide that, when introduced into cells via a TAT cell-delivery system, shows a remarkably potent cytoxicity in a variety of cancer cell lines and inhibits tumor growth in vivo, whereas sparing normal cells and tissues. This fusion peptide was named killerFLIP as its sequence was derived from the C-terminal domain of c-FLIP, an anti-apoptotic protein. Using structure activity analysis, we determined the minimal bioactive core of killerFLIP, namely killerFLIP-E. Structural analysis of cells using electron microscopy demonstrated that killerFLIP-E triggers cell death accompanied by rapid (within minutes) plasma membrane permeabilization. Studies of the structure of the active core of killerFLIP (-E) indicated that it possesses amphiphilic properties and self-assembles into micellar structures in aqueous solution. The biochemical properties of killerFLIP are comparable to those of cationic lytic peptides, which participate in defense against pathogens and have also demonstrated anticancer properties. We show that the pro-cell death effects of killerFLIP are independent of its sequence similarity with c-FLIPL as killerFLIP-induced cell death was largely apoptosis and necroptosis independent. A killerFLIP-E variant containing a scrambled c-FLIPL motif indeed induced similar cell death, suggesting the importance of the c-FLIPL residues but not of their sequence. Thus, we report the discovery of a promising synthetic peptide with novel anticancer activity in vitro and in vivo.

Sensitization of prostate carcinoma cells to Apo2L/TRAIL by a Bcl-2 family protein inhibitor.

Overexpression of anti-apoptotic Bcl-2 family proteins may play an important role in the aggressive behavior of prostate cancer cells and their resistance to therapy. The Bcl-2 homology 3 domain (BH3) is a uniquely important functional element within the pro-apoptotic class of the Bcl-2-related proteins, mediating their ability to dimerize with other Bcl-2-related proteins and promote apoptosis. The BH3 inhibitors (BH3Is) function by disrupting the interactions mediated by the BH3 domain between pro- and anti-apoptotic members of the Bcl-2 family and liberating more Bax/Bak to induce mitochondrial membrane permeabilization. LNCaP-derived C4-2 human prostate cancer cells are quite resistant to non-tagged, human recombinant soluble Apo2 ligand [Apo2L, also Tumor necrosis factor (TNF)-related apoptosis-inducing ligand, TRAIL], a tumor specific drug that is now in clinical trials. However, when Apo2L/TRAIL was combined with the Bcl-xL inhibitor, BH3I-2', it induced apoptosis synergistically through activation of Caspase-8 and the proapoptotic Bcl-2 family member Bid, resulting in the activation of effector Caspase-3 and proteolytic cleavage of Poly(ADP-ribose) polymerase, events that were blocked by the pan-caspase inhibitor zVAD-fmk. Our data indicate that, in combination with the BH3 mimetic, BH3I-2', Apo2L/TRAIL synergistically induces apoptosis in C4-2 human prostate cancer cells through both the extrinsic and intrinsic apoptotic pathways.