Name: machine learning algorithms for early detection of bone metastases in an experimental rat model
Uploaded: 2020-08-16T13:00:00
Description: Watch this Scientific Journal Video about Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model at JoVE.com

This work was supported by the German Research Foundation (DFG, Collaborative Research Centre CRC 1181, subproject Z02; Priority Programme &#956;Bone, projects BA 4027/10-1 and BO 3811), including additional support for the scanning devices (INST&#160;410/77-1&#160;FUGG and INST&#160;410/93-1&#160;FUGG), and by the Emerging Fields Initiative (EFI) &#8220;Big Thera&#8221; of the Friedrich-Alexander-University Erlangen-N&#252;rnberg.

All care and experimental procedures were performed in accordance with national and regional legislation on animal protection, and all animal procedures were approved by the State Government of Franconia, Germany (reference number 55.2 DMS-2532-2-228).
1. Induction of breast cancer bone metastases in the right hind leg of nude rats
NOTE: A detailed description of the induction of breast cancer bone metastases in nude rats has been published elsewhere...

<ol>
	<li>D'Oronzo, S., Brown, J., Coleman, R. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+role+of+biomarkers+in+the+management+of+bone-homing+malignancies.">The role of biomarkers in the management of bone-homing malignancies.</a> Journal of Bone Oncology. 9, 1-9 (2017).</li><li>Ellmann, S., Beck, M., Kuwert, T., Uder, M., B&#228;uerle, T. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Multimodal+imaging+of+bone+metastases:+From+preclinical+to+clinical+applications.">Multimodal imaging of bone metastases: From preclinical to clinical applications.</a> Journal of Orthopaedic Translation. 3 (4), 166-177 (2015).</li><li>Braun, S., Pantel, K. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Clinical+significance+of+occult+metastatic+cells+in+bone+marrow+of+breast+cancer+patients.">Clinical significance of occult metastatic cells in bone marrow of breast cancer patients.</a> The Oncologist. 6 (2), 125-132 (2001).</li><li>Braun, S., Rosenberg, R., Thorban, S., Harbeck, N. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Implications+of+occult+metastatic+cells+for+systemic+cancer+treatment+in+patients+with+breast+or+gastrointestinal+cancer.">Implications of occult metastatic cells for systemic cancer treatment in patients with breast or gastrointestinal cancer.</a> Seminars in surgical oncology. 20 (4), 334-346 (2001).</li><li>Ellmann, S., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Prediction+of+early+metastatic+disease+in+experimental+breast+cancer+bone+metastasis+by+combining+PET/CT+and+MRI+parameters+to+a+Model-Averaged+Neural+Network.">Prediction of early metastatic disease in experimental breast cancer bone metastasis by combining PET/CT and MRI parameters to a Model-Averaged Neural Network.</a> Bone. 120, 254-261 (2018).</li><li>B&#228;uerle, T., Komljenovic, D., Berger, M. R., Semmler, W. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Multi-modal+imaging+of+angiogenesis+in+a+nude+rat+model+of+breast+cancer+bone+metastasis+using+magnetic+resonance+imaging,+volumetric+computed+tomography+and+ultrasound.">Multi-modal imaging of angiogenesis in a nude rat model of breast cancer bone metastasis using magnetic resonance imaging, volumetric computed tomography and ultrasound.</a> Journal of Visualized Experiments. (66), e4178 (2012).</li><li>Merz, M., Komljenovic, D., Semmler, W., B&#228;uerle, T. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Quantitative+contrast-enhanced+ultrasound+for+imaging+antiangiogenic+treatment+response+in+experimental+osteolytic+breast+cancer+bone+metastases.">Quantitative contrast-enhanced ultrasound for imaging antiangiogenic treatment response in experimental osteolytic breast cancer bone metastases.</a> Investigative Radiology. 47 (7), 422-429 (2012).</li><li>B&#228;uerle, T., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Characterization+of+a+rat+model+with+site-specific+bone+metastasis+induced+by+MDA-MB-231+breast+cancer+cells+and+its+application+to+the+effects+of+an+antibody+against+bone+sialoprotein.">Characterization of a rat model with site-specific bone metastasis induced by MDA-MB-231 breast cancer cells and its application to the effects of an antibody against bone sialoprotein.</a> International Journal of Cancer. 115 (2), 177-186 (2005).</li><li>Patel, J., Goyal, R. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Applications+of+Artificial+Neural+Networks+in+Medical+Science.">Applications of Artificial Neural Networks in Medical Science.</a> Current Clinical Pharmacology. 2 (3), 217-226 (2008).</li><li>Naftaly, U., Intrator, N., Horn, D. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Optimal+ensemble+averaging+of+neural+networks.">Optimal ensemble averaging of neural networks.</a> Network: Computation in Neural Systems. 8 (3), 283-296 (1997).</li><li>B&#228;uerle, T., Merz, M., Komljenovic, D., Zwick, S., Semmler, W. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Drug-induced+vessel+remodeling+in+bone+metastases+as+assessed+by+dynamic+contrast+enhanced+magnetic+resonance+imaging+and+vessel+size+imaging:+A+longitudinal+in+vivo+study.">Drug-induced vessel remodeling in bone metastases as assessed by dynamic contrast enhanced magnetic resonance imaging and vessel size imaging: A longitudinal in vivo study.</a> Clinical Cancer Research. 16 (12), 3215-3225 (2010).</li><li>Cheng, C., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Evaluation+of+treatment+response+of+cilengitide+in+an+experimental+model+of+breast+cancer+bone+metastasis+using+dynamic+PET+with+18F-FDG.">Evaluation of treatment response of cilengitide in an experimental model of breast cancer bone metastasis using dynamic PET with 18F-FDG.</a> Hellenic Journal of Nuclear Medicine. 14 (1), 15-20 (2011).</li><li>Marturano-Kruik, A., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Human+bone+perivascular+niche-on-a-chip+for+studying+metastatic+colonization.">Human bone perivascular niche-on-a-chip for studying metastatic colonization.</a> Proceedings of the National Academy of Sciences of the United States of America. 115 (6), 1256-1261 (2018).</li><li>Sonntag, E., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=In+vivo+proof-of-concept+for+two+experimental+antiviral+drugs,+both+directed+to+cellular+targets,+using+a+murine+cytomegalovirus+model.">In vivo proof-of-concept for two experimental antiviral drugs, both directed to cellular targets, using a murine cytomegalovirus model.</a> Antiviral Research. 161, 63-69 (2019).</li><li>. Horos - Free DICOM Medical Image Viewer | Open-Source Available from: <a target="_blank" href="https://www.horosproject.org/">https://www.horosproject.org/</a> (2015)</li><li>. RStudio Team RStudio: Inteegrated Development for R Available from: <a target="_blank" href="https://rstudio.com">https://rstudio.com</a> (2015)</li><li>Dem&#353;ar, J., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Orange:+Data+Mining+Toolbox+in+Python.">Orange: Data Mining Toolbox in Python.</a> Journal of Machine Learning Research. 14, 2349-2353 (2013).</li><li>Saeys, Y., Inza, I., Larra&#241;aga, P. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+review+of+feature+selection+techniques+in+bioinformatics.">A review of feature selection techniques in bioinformatics.</a> Bioinformatics. 23 (19), 2507-2517 (2007).</li><li>. CRAN - Package caret Available from: <a target="_blank" href="https://cran.r-project.org/web/packages/caret/index.html">https://cran.r-project.org/web/packages/caret/index.html</a> (2016)</li><li>. CRAN: Package xgboost - Extreme Gradient Boosting Available from: <a target="_blank" href="https://cran.r-project.org/web/packages/xgboost/">https://cran.r-project.org/web/packages/xgboost/</a> (2019)</li><li>Fern&#225;ndez-Delgado, M., Cernadas, E., Barro, S., Amorim, D., Fern&#225;ndez-Delgado, A. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Do+we+Need+Hundreds+of+Classifiers+to+Solve+Real+World+Classification+Problems.">Do we Need Hundreds of Classifiers to Solve Real World Classification Problems.</a> Journal of Machine Learning Research. 15, 3133-3181 (2014).</li><li>Hira, Z. M., Gillies, D. F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+Review+of+Feature+Selection+and+Feature+Extraction+Methods+Applied+on+Microarray+Data.">A Review of Feature Selection and Feature Extraction Methods Applied on Microarray Data.</a> Advances in Bioinformatics. 2015, 198363 (2015).</li><li>S&#225;nchez-Maro&#241;o, N., Alonso-Betanzos, A., Tombilla-Sanrom&#225;n, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Filter+methods+for+feature+selection+-+A+comparative+study.">Filter methods for feature selection - A comparative study.</a> Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). 4881, 178-187 (2007).</li><li>Cawley, G. C., Talbot, N. L. C. C. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Fast+exact+leave-one-out+cross-validation+of+sparse+least-squares+support+vector+machines.">Fast exact leave-one-out cross-validation of sparse least-squares support vector machines.</a> Neural Network. 17 (10), 1467-1475 (2004).</li><li>Forghani, R., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Radiomics+and+Artificial+Intelligence+for+Biomarker+and+Prediction+Model+Development+in+Oncology.">Radiomics and Artificial Intelligence for Biomarker and Prediction Model Development in Oncology.</a> Computational and Structural Biotechnology Journal. 17, 995-1008 (2019).</li><li>Jaffe, C. C. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Measures+of+response:+RECIST,+WHO,+and+new+alternatives.">Measures of response: RECIST, WHO, and new alternatives.</a> Journal of Clinical Oncology Official Journal of the American Society of Clinical Oncology. 24 (20), 3245-3251 (2006).</li><li>Lambin, P., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Radiomics:+Extracting+more+information+from+medical+images+using+advanced+feature+analysis.">Radiomics: Extracting more information from medical images using advanced feature analysis.</a> European Journal of Cancer. 48 (4), 441-446 (2012).</li><li>Gillies, R. J., Kinahan, P. E., Hricak, H. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Radiomics:+Images+are+more+than+pictures,+they+are+data.">Radiomics: Images are more than pictures, they are data.</a> Radiology. 278 (2), 563-577 (2016).</li><li>Nioche, C., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Lifex:+A+freeware+for+radiomic+feature+calculation+in+multimodality+imaging+to+accelerate+advances+in+the+characterization+of+tumor+heterogeneity.">Lifex: A freeware for radiomic feature calculation in multimodality imaging to accelerate advances in the characterization of tumor heterogeneity.</a> Cancer Research. 78 (16), 4786-4789 (2018).</li><li>Ellmann, S., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Application+of+machine+learning+algorithms+for+multiparametric+MRI-based+evaluation+of+murine+colitis.">Application of machine learning algorithms for multiparametric MRI-based evaluation of murine colitis.</a> PLOS ONE. 13 (10), 0206576 (2018).</li></ol>

ML algorithms are powerful tools used to integrate several predictive features into a combined model and obtain an accuracy that exceeds that of its separate constituents when used alone. Nonetheless, the actual result depends on several critical steps. First, the ML algorithm used is a crucial factor, because different ML algorithms yield different results. The algorithm used in this protocol is an avNNet, but other promising algorithms include Extreme Gradient Boosting21 or Random Forests. The c...

The purpose of this protocol is to integrate several functional imaging parameters from MRI and PET/CT into a model-averaged neural network (avNNet) ML algorithm. This algorithm predicts the growth of macrometastases in a rat model of breast cancer bone metastases at an early timepoint, when macroscopic changes within the bone are not yet visible.
Prior to the growth of macrometastases, a bone marrow invasion of disseminated tumor cells occurs, commonly referred to as micrometastatic disease1,2. This initial invasion can be considered an early step in metastatic disease, but is typically missed during conventional staging examinations3,4. Although the currently available imaging modalities cannot detect bone marrow microinvasion when used alone, a combination of imaging parameters yielding information on vascularization and metabolic activity has been shown to perform better5. This complementary benefit is achieved by combining different imaging parameters into an avNNet, which is an ML algorithm. Such an avNNet allows for the reliable prediction of bone macrometastases formation before any visible metastases are present. Therefore, integrating imaging biomarkers into an avNNet could serve as a surrogate parameter for bone marrow microinvasion and early metastatic disease.
To develop the protocol, a previously described model of breast cancer bone metastases in nude rats was used6,7,8. The advantage of this model is its site-specificity, meaning that the animals develop bony metastases exclusively in their right hind leg. However, the tumor-take rate of this approach is 60%&#8211;80%, so a considerable number of the animals do not develop any metastases during the study. Using imaging modalities such as MRI and PET/CT, the presence of metastases is detectable from day 30 postinjection (PI). At earlier time points (e.g., 10 PI) imaging does not distinguish between animals that will develop metastatic disease and those will not (Figure 1).
An avNNet trained on functional imaging parameters acquired on day 10 PI, as described in the following protocol, reliably predicts or excludes the growth of macrometastases within the following ~3 weeks. Neural Networks combine artificial nodes within different layers. In the study protocol, the functional imaging parameters for bone marrow blood supply and metabolic activity represent the bottom layer, while the prediction of malignancy represents the top layer. An additional intermediate layer contains hidden nodes that are connected to both the top and the bottom layer. The strength of the connections between the different nodes is updated during the training of the network to perform the respective classification task with high accuracy9. The accuracy of such a neural network can be further increased by averaging the outputs of several models, resulting in an avNNet10.

<table border="0" cellpadding="0" cellspacing="0" style="width:801px;" width="800">
	<colgroup>
		<col />
		<col />
		<col />
		<col />
	</colgroup>
	<tbody>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">Binocular Operating Microscope</td>
			<td style="width:239px;">Leica</td>
			<td style="width:154px;">NA</td>
			<td style="width:167px;"></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">ClinScan MR System</td>
			<td style="width:239px;">Bruker</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">DICOM Viewer</td>
			<td style="width:239px;">Horos</td>
			<td style="width:154px;">NA</td>
			<td style="width:167px;">www.horosproject.org</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">Excel: Spreadsheet</td>
			<td style="width:239px;">Microsoft</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">FCS</td>
			<td style="width:239px;">Sigma</td>
			<td style="width:154px;">F2442-500ML</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">Gadovist</td>
			<td style="width:239px;">Bayer-Schering</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">Inveon PET/CT</td>
			<td style="width:239px;">Siemens</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:241px;">Inveon Research Workplace Software</td>
			<td style="width:239px;">Siemens Healthcare GmbH</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">IVIS Spectrum</td>
			<td style="width:239px;">PerkinElmer</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
		<tr height="42">
			<td height="42" style="height:42px;width:241px;">MDA-MB-231 human breast cancer cells</td>
			<td style="width:239px;">American Type Culture Collection</td>
			<td style="width:154px;">N/A</td>
			<td></td>
		</tr>
		<tr height="63">
			<td height="63" style="height:63px;width:241px;">Open-source data visualization, machine learning and data mining toolkit.</td>
			<td style="width:239px;">Orange3, University of Ljubljana</td>
			<td style="width:154px;">NA</td>
			<td>https://orange.biolab.si/</td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">RPMI-1640</td>
			<td>Invitrogen/ThermoFisher</td>
			<td align="right" style="width:154px;">11875093</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">Trypsin</td>
			<td style="width:239px;">Sigma</td>
			<td style="width:154px;">9002-07-7</td>
			<td></td>
		</tr>
		<tr height="21">
			<td height="21" style="height:21px;width:241px;">Vevo 3100</td>
			<td style="width:239px;">VisualSonics</td>
			<td style="width:154px;">NA</td>
			<td></td>
		</tr>
	</tbody>
</table>

machine learning algorithms for early detection of bone metastases in an experimental rat model

Machine learning (ML) algorithms permit the integration of different features into a model to perform classification or regression tasks with an accuracy exceeding its constituents. This protocol describes the development of an ML algorithm to predict the growth of breast cancer bone macrometastases in a rat model before any abnormalities are observable with standard imaging methods. Such an algorithm can facilitate the detection of early metastatic disease (i.e., micrometastasis) that is regularly missed during staging examinations.
The applied metastasis model is site-specific, meaning that the rats develop metastases exclusively in their right hind leg. The model&#8217;s tumor-take rate is 60%&#8211;80%, with macrometastases becoming visible in magnetic resonance imaging (MRI) and positron emission tomography/computed tomography (PET/CT) in a subset of animals 30 days after induction, whereas a second subset of animals exhibit no tumor growth.
Starting from image examinations acquired at an earlier time point, this protocol describes the extraction of features that indicate tissue vascularization detected by MRI, glucose metabolism by PET/CT, and the subsequent determination of the most relevant features for the prediction of macrometastatic disease. These features are then fed into a model-averaged neural network (avNNet) to classify the animals into one of two groups: one that will develop metastases and the other that will not develop any tumors. The protocol also describes the calculation of standard diagnostic parameters, such as overall accuracy, sensitivity, specificity, negative/positive predictive values, likelihood ratios, and the development of a receiver operating characteristic. An advantage of the proposed protocol is its flexibility, as it can be easily adapted to train a plethora of different ML algorithms with adjustable combinations of an unlimited number of features. Moreover, it can be used to analyze different problems in oncology, infection, and inflammation.

The rats recovered quickly from the surgery and injection of the MDA-MB-231 breast cancer cells and were then subjected to MR- and PET/CT imaging on days 10 and 30 PI (Figure 1). A representative DCE analysis of a rat&#8217;s right proximal tibia is presented in Figure 2A. The DCE raw measurements were saved by selecting the &#8220;Export&#8221; button and choosing &#8220;DCEraw.txt&#8221; as the file name.
Subsequent...

Watch this Scientific Journal Video about Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model at JoVE.com

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model

This protocol was designed to train a machine learning algorithm to use a combination of imaging parameters derived from magnetic resonance imaging (MRI) and positron emission tomography/computed tomography (PET/CT) in a rat model of breast cancer bone metastases to detect early metastatic disease and predict subsequent progression to macrometastases.

Machine learning (ML) algorithms permit the integration of different features into a model to perform classification or regression tasks with an accuracy ...

This protocol facilitates the development of a machine learning algorithm for predicting the growth of bone metastases in an experimental model at the stage of early organ colonization. The main advantage of this technique is that it combines several imaging parameters into a machine learning algorithm that significantly outperforms the productive ability of each individual parameter. Though, this protocol aims for the early diagnosis of metastases in bone, it can be adapted to different organs or areas of multimodal and multiparametric imaging research.10 days post-surgery and injection in a DICOM viewer with a DCE plugin. Pick import to load the DCE sequence in 4D mode and select the DICOM folder containing the MRI images. Place a circular 1.5 square millimeter to dimensional region of interest in the proximal tibial shaft bone marrow of the right hind leg within the fourth or fifth MRI image and select relative enhancement in the plot type.Enter the baseline range from time points, one to five into the respective fields and export the analysis as a dot TXT file named DCE raw dot TXT. Open RStudio and load the provided DCE script dot R file. Select code, run region and run all to run the entire script and copy the output to the provided imaging features template file.In the DICOM viewer, place a second region of interest within the back muscle of the animal and repeat the DCE measurements as demonstrated. In the imaging feature spreadsheet the respective bone measurements will be automatically divided by the respective muscle measurements for data normalization. To analyze the PET/CT images open the PET/CT analysis software and import the PET/CT images.Click general analysis and select region of interest quantification, create and create a region of interest from a template. Place a four by six millimeter, to dimensional region of interest within the bone marrow of the proximal tibial shaft of the right hind leg and select regions of interest target one overlay. Note the mean, minimum and maximum values in becquerels per milliliter.Then, divide the maximum value by the injected activity and multiply the result by the animal weight in grams to calculate the standardized uptake value. To diagnose the tumor growth rate in the injected hind leg. After obtaining MRI and PET/CT images on day 30 post-injection, analyze the images as demonstrated and add a tumor column to the imaging features spreadsheet.Then enter a one or zero for animals with metastasis or without a visible tumor burden respectively. To determine the most relevant features for the prediction of future tumor growth, import the spreadsheet into an open source data visualization machine learning and data mining toolkit. Drag the file sub-routine from the data menu into the workspace and double-click the file.Click the folder icon to load the spreadsheet and select the imaging features spreadsheet. Select the export worksheet and assign the target attribute to the variable tumor. Assign the skip function to the animal number.Drag the rank sub-routine into the workspace and draw a line to connect to the file and rank sub-routines. Then, double click to open the rank sub-routine and select the information gain algorithm. For machine learning analysis, open RStudio 3.4.1 and load the provided train model R script.Select lines three through six within the script to load the required libraries. Click code to run the selected code and click run selected lines to run the selected lines. To train a model average to neural network algorithm, select lines eight through 39 from the train model R script and click code and run selected lines.Then to assess the standard parameters of diagnostic accuracy, select lines 41 through 50 from the train model R script and click code and run selected lines. On day 10 after surgery and cancer cell injection, MRI and PET/CT images can be acquired. DCE analysis allows the measurement of muscle and bone tissue areas of interest.These values can be normalized by dividing the bone measurements by the muscle measurements. On day 30 post-injection, all of animals are evaluated to determine whether they have developed metasticies with a one indicating a positive tumor burden and a zero indicating healthy animals without visible tumors. Running the train model R script allows the optimal hyper parameter combination to be determined and the final model to be calculated using the optimal hyper parameter combination.These data allow a set of standard diagnostic parameters to be calculated and the receiver operating characteristic curve of the model to be plotted. For example, in this analysis of 28 samples the model performs significantly better than all of its three constituents. Several machine learning algorithms tend to perform better when the input data are normalized.In this protocol normalization is achieved using the Box Cox function. This protocol uses a model average during network as a machine learning algorithm. However, the provided framework can easily be adapted to other algorithms such as random forests or support vector machines.Extracting numerical information from image material has become essential. Algorithms such as these may facilitate the integration of large amounts of data to allow patient stratification.

machine-learning-algorithms-for-early-detection-bone-metastases-an

Research

JoVE Journal

Cancer Research

Intra-iliac Artery Injection for Efficient and Selective Modeling of Microscopic Bone Metastasis

Intra-iliac artery (IIA) injection is an efficient approach to introduce metastatic lesions of various cancer cells in animals. Compared to the widely used intra-cardiac and intra-tibial injections, IIA injection brings several advantages. First, it can deliver a large quantity of cancer cells specifically to hind limb bones, thereby providing spatiotemporally synchronized early-stage colonization events and allowing robust quantification and swift detection of disseminated tumor cells. Second, it injects cancer cells into the circulation without damaging the local tissues, thereby avoiding inflammatory and wound-healing processes that confound the bone colonization process. Third, IIA injection causes very little metastatic growth in non-bone organs, thereby preventing animals from succumbing to other vital metastases, and allowing continuous monitoring of indolent bone lesions. These advantages are especially useful for the inspection of progression from single cancer cells to multi-cell micrometastases, which has largely been elusive in the past. When combined with cutting-edge approaches of biological imaging and bone histology, IIA injection can be applied to various research purposes related to bone metastases.

This manuscript provides the detailed procedure of intra-iliac artery (IIA) injection, a technique to deliver cancer cells specifically to hind limb tissues including bones to establish experimental bone metastases. Although initially established with breast tumor models, this protocol can be easily extended to other cancer types.

Intra-iliac artery (IIA) injection is an efficient approach to introduce metastatic lesions of various cancer cells in animals. Compared to the widely ...

A 3D Organotypic Melanoma Spheroid Skin Model

Malignant transformation of melanocytes, the pigment cells of human skin, causes formation of melanoma, a highly aggressive cancer with increased metastatic potential. Recently, mono-chemotherapies continue to improve by melanoma specific combination therapies with targeted kinase inhibitors. Still, metastatic melanoma remains a life-threatening disease because tumors exhibit primary resistance or develop resistance to novel therapies, thereby regaining tumorigenic capacity. In order to improve the therapeutic success of malignant melanoma, the determination of molecular mechanisms conferring resistance against conventional treatment approaches is necessary; however, it requires innovative cellular in vitro models. Here, we introduce an in vitro three-dimensional (3D) organotypic melanoma spheroid model that can portray the in vivo architecture of malignant melanoma and may warrant new insights into intra-tumoral as well as tumor-host interactions. The model incorporates defined numbers of mature and differentiated melanoma spheroids in a 3D human full skin reconstruction model consisting of primary skin cells. The cellular composition and differentiation status of the embedded melanoma spheroids is similar to the one of cutaneous melanoma metastasis in vivo. Using this organotypic melanoma spheroid model as a drug screening platform may support the identification of responders to selected combination therapies, while sparing the unnecessary treatment burden for non-responders, thereby increasing the benefit of therapeutic interventions.

Here, we present a protocol to generate a 3D organotypic melanoma spheroid skin model that recapitulates both the architecture and multicellular complexity of an organ/tumor in vivo but at the same time accommodates systematic experimental intervention.

Malignant transformation of melanocytes, the pigment cells of human skin, causes formation of melanoma, a highly aggressive cancer with increased ...

Evaluation of the Feasibility, Safety, and Accuracy of an Intraoperative High-intensity Focused Ultrasound Device for Treating Liver Metastases

Today, the only potentially curative option in patients with colorectal liver metastases is surgery. However, liver resection is feasible in less than 20% of patients. Surgery has been widely used in association with radiofrequency, cryotherapy, or microwaves to expand the number of treatments performed with a curative intent. Nevertheless, several limitations have been documented when using these techniques (i.e., a traumatic puncture of the parenchyma, a limited size of lesions, and an inability to monitor the treatment in real-time).High-intensity focused ultrasound (HIFU) technology can achieve precise ablations of biological tissues without incisions or radiation. Current HIFU devices are based on an extracorporeal approach with limited access to the liver. We have developed a HIFU device designed for intraoperative use. The use of a toroidal transducer enables an ablation rate (10 cm3&#183;min-1) higher than any other treatment and is independent of perfusion.
The feasibility, safety, and accuracy of intraoperative HIFU ablation were evaluated during an ablate-and-resect prospective study. This clinical phase I and IIa study was performed in patients undergoing hepatectomy for liver metastases. The HIFU treatment was performed on healthy tissue scheduled for resection.Liver metastases measuring less than 20 mm will be targeted during phase IIb (currently ongoing). This set-up allows the real-time evaluation of HIFU ablation while protecting participating patients from any adverse effects related to this new technique.
Fifteen patients were included in phase I&#8211;IIa and 30 HIFU ablations were safely created within 40 s and with a precision of 1&#8211;2 mm. The average dimensions of the HIFU ablations were 27.5 x 21.0 mm2, corresponding to a volume of approximately 7.5 cm3. The aim of the ongoing phase IIb is to ablate metastases of less than 20 mm in diameter with a 5 mm margin.

Here, we present an ablate-and-resect prospective study to evaluate the feasibility, safety, and accuracy of intraoperative high-intensity focused ultrasound ablation in patients undergoing hepatectomy for liver metastases.

Today, the only potentially curative option in patients with colorectal liver metastases is surgery. However, liver resection is feasible in less than 20% ...

Three-Dimensional Bone Extracellular Matrix Model for Osteosarcoma

Osteosarcoma (OS) is the most common and a highly aggressive primary bone tumor. It is characterized with anatomic and histologic variations along with diagnostic or prognostic difficulties. OS comprises genotypically and phenotypically heterogeneous cancer cells. Bone microenvironment elements are proved to account for tumor heterogeneity and disease progression. Bone extracellular matrix (BEM) retains the microstructural matrices and biochemical components of native extracellular matrix. This tissue-specific niche provides a favorable and long-term scaffold for OS cell seeding and proliferation. This article provides a protocol for the preparation of BEM model and its further experimental application. OS cells can grow and differentiate into multiple phenotypes consistent with the histopathological complexity of OS clinical specimens. The model also allows visualization of diverse morphologies and their association with genetic alterations and underlying regulatory mechanisms. As homologous to human OS, this BEM-OS model can be developed and applied to the pathology and clinical research of OS.

The bone extracellular matrix (BEM) model for osteosarcoma (OS) is well established and shown here. It can be used as a suitable scaffold for mimicking primary tumor growth in vitro and providing an ideal model for studying the histologic and cytogenic heterogeneity of OS.

Osteosarcoma (OS) is the most common and a highly aggressive primary bone tumor. It is characterized with anatomic and histologic variations along with ...

Detection of Lung Tumor Progression in Mice by Ultrasound Imaging

With ~1.6 million victims per year, lung cancer contributes tremendously to the worldwide burden of cancer. Lung cancer is partly driven by genetic alterations in oncogenes such as the KRAS oncogene, which constitutes ~25% of lung cancer cases. The difficulty in therapeutically targeting KRAS-driven lung cancer partly stems from having poor models that can mimic the progression of the disease in the lab. We describe a method that permits the relative quantification of primary KRAS lung tumors in a Cre-inducible LSL-KRAS G12D mouse model via ultrasound imaging. This method relies on brightness (B)-mode acquisition of the lung parenchyma. Tumors that are initially formed in this model are visualized as B-lines and can be quantified by counting the number of B-lines present in the acquired images. These would represent the relative tumor number formed on the surface of the mouse lung. As the formed tumors develop with time, they are perceived as deep clefts within the lung parenchyma. Since the circumference of the formed tumor is well-defined, calculating the relative tumor volume is achieved by measuring the length and width of the tumor and applying them in the formula used for tumor caliper measurements. Ultrasound imaging is a non-invasive, fast and user-friendly technique that is often used for tumor quantifications in mice. Although artifacts may appear when obtaining ultrasound images, it has been shown that this imaging technique is more advantageous for tumor quantifications in mice compared to other imaging techniques such as computed tomography (CT) imaging and bioluminescence imaging (BLI). Researchers can investigate novel therapeutic targets using this technique by comparing lung tumor initiation and progression between different groups of mice.

This protocol describes the steps taken to induce KRAS lung tumors in mice as well as the quantification of formed tumors by ultrasound imaging. Small tumors are visualized in early timepoints as B-lines. At later timepoints, relative tumor volume measurements are achieved by the measurement tool in the ultrasound software.

With ~1.6 million victims per year, lung cancer contributes tremendously to the worldwide burden of cancer. Lung cancer is partly driven by genetic ...

Quantifying the Brain Metastatic Tumor Micro-Environment using an Organ-On-A Chip 3D Model, Machine Learning, and Confocal Tomography

Brain metastases are the most lethal cancer lesions; 10-30% of all cancers metastasize to the brain, with a median survival of only ~5-20 months, depending on the cancer type. To reduce the brain metastatic tumor burden, gaps in basic and translational knowledge need to be addressed. Major challenges include a paucity of reproducible preclinical models and associated tools. Three-dimensional models of brain metastasis can yield the relevant molecular and phenotypic data used to address these needs when combined with dedicated analysis tools. Moreover, compared to murine models, organ-on-a-chip models of patient tumor cells traversing the blood brain barrier into the brain microenvironment generate results rapidly and are more interpretable with quantitative methods, thus amenable to high throughput testing. Here we describe and demonstrate the use of a novel 3D microfluidic blood brain niche (&#181;mBBN) platform where multiple elements of the niche can be cultured for an extended period (several days), fluorescently imaged by confocal microscopy, and the images reconstructed using an innovative confocal tomography technique; all aimed to understand the development of micro-metastasis and changes to the tumor micro-environment (TME) in a repeatable and quantitative manner. We demonstrate how to fabricate, seed, image, and analyze the cancer cells and TME cellular and humoral components, using this platform. Moreover, we show how artificial intelligence (AI) is used to identify the intrinsic phenotypic differences of cancer cells that are capable of transit through a model &#181;mBBN and to assign them an objective index of brain metastatic potential. The data sets generated by this method can be used to answer basic and translational questions about metastasis, the efficacy of therapeutic strategies, and the role of the TME in both.

Here, we present a protocol for preparing and culturing a blood brain barrier metastatic tumor micro-environment and then quantifying its state using confocal imaging and artificial intelligence (machine learning).

Brain metastases are the most lethal cancer lesions; 10-30% of all cancers metastasize to the brain, with a median survival of only ~5-20 months, ...

An Orthotopic Resectional Mouse Model of Pancreatic Cancer

There is a lack of satisfactory animal models to study adjuvant and/or neoadjuvant therapy in patients being considered for surgery of pancreatic cancer (PC). To address this deficiency, we describe a mouse model involving orthotopic implantation of PC followed by distal pancreatectomy and splenectomy. The model has been demonstrated to be safe and suitably flexible for the study of various therapeutic approaches in adjuvant and neo adjuvant settings.
In this model, a pancreatic tumor is first generated by implanting a mixture of human pancreatic cancer cells (luciferase-tagged AsPC-1) and human cancer associated pancreatic stellate cells into the distal pancreas of Balb/c athymic nude mice. After three weeks, the cancer is resected by re-laparotomy, distal pancreatectomy and splenectomy. In this model, bioluminescence imaging can be used to follow the progress of cancer development and effects of resection/treatments. Following resection, adjuvant therapy can be given. Alternatively, neoadjuvant treatment can be given prior to resection.
Representative data from 45 mice are presented. All mice underwent successful distal pancreatectomy/splenectomy with no issues of hemostasis. A macroscopic proximal pancreatic margin greater than 5 mm was achieved in 43 (96%) mice. The technical success rate of pancreatic resection was 100%, with 0% early mortality and morbidity. None of the animals died during the week after resection.
In summary, we describe a robust and reproducible technique for a surgical resection model of pancreatic cancer in mice which mimics the clinical scenario. The model may be useful for the testing of both adjuvant and neoadjuvant treatments.

In the clinical context, patients with localized pancreatic cancer will undergo pancreatectomy followed by adjuvant treatment. This protocol reported here aims to establish a safe and effective method of modelling this clinical scenario in nude mice, through orthotopic implantation of pancreatic cancer followed by distal pancreatectomy and splenectomy.

There is a lack of satisfactory animal models to study adjuvant and/or neoadjuvant therapy in patients being considered for surgery of pancreatic cancer ...

Experimental Melanoma Immunotherapy Model Using Tumor Vaccination with a Hematopoietic Cytokine

Fms-like tyrosine kinase 3 ligand (Flt3L) is a hematopoietic cytokine that promotes the survival and differentiation of dendritic cells (DCs). It has been used in tumor vaccines to activate innate immunity and enhance antitumor responses. This protocol demonstrates a therapeutic model using cell-based tumor vaccine consisting of Flt3L-expressing B16-F10 melanoma cells along with phenotypic and functional analysis of immune cells in the tumor microenvironment (TME). Procedures for cultured tumor cell preparation, tumor implantation, cell irradiation, tumor size measurement, intratumoral immune cell isolation, and flow cytometry analysis are described. The overall goal of this protocol is to provide a preclinical solid tumor immunotherapy model, and a research platform to study the relationship between tumor cells and infiltrating immune cells. The immunotherapy protocol described here can be combined with other therapeutic modalities, such as immune checkpoint blockade (anti-CTLA-4, anti-PD-1, anti-PD-L1 antibodies) or chemotherapy in order to improve the cancer therapeutic effect of melanoma.

The protocol presents a cancer immunotherapy model using cell-based tumor vaccination with Flt3L-expressing B16-F10 melanoma. This protocol demonstrates the procedures, including preparation of cultured tumor cells, tumor implantation, cell irradiation, measurement of tumor growth, isolation of intratumoral immune cells, and flow cytometry analysis.

Fms-like tyrosine kinase 3 ligand (Flt3L) is a hematopoietic cytokine that promotes the survival and differentiation of dendritic cells (DCs). It has been ...

Mapping the Tumor Immune Landscape: An In-Depth Multiplex IHC Analysis Protocol

Multiplex Immunohistochemical Analysis of the Spatial Immune Cell Landscape of the Tumor Microenvironment

The immune cell landscape of the tumor microenvironment potentially contains information for the discovery of prognostic and predictive biomarkers. Multiplex immunohistochemistry is a valuable tool to visualize and identify different types of immune cells in tumor tissues while retaining its spatial information. Here we provide detailed protocols to analyze lymphocyte, myeloid, and dendritic cell populations in tissue sections. Starting from cutting formalin-fixed paraffin-embedded sections, automatic multiplex staining procedures on an automated platform, scanning of the slides on a multispectral imaging microscope, to the analysis of images using an in-house-developed machine learning algorithm ImmuNet. These protocols can be applied to a variety of tumor specimens by simply switching tumor markers to analyze immune cells in different compartments of the sample (tumor versus invasive margin) and apply nearest-neighbor analysis. This analysis is not limited to tumor samples but can also be applied to other (non-)pathogenic tissues. Improvements to the equipment and workflow over the past few years have significantly shortened throughput times, which facilitates the future application of this procedure in the diagnostic setting.

Author Spotlight: Unlocking Insights into the Immune Cell Landscape of Tumors

This protocol describes in detail how immune cell characterization of the tumor microenvironment using multiplex immunohistochemistry is carried out.

The immune cell landscape of the tumor microenvironment potentially contains information for the discovery of prognostic and predictive biomarkers. ...

Reprogramming PDAC and Pancreatic Cells into Induced Pluripotent Stem Cells

Reprogramming Pancreatic Ductal Adenocarcinoma to Pluripotency

The generation of induced pluripotent stem cells (iPSCs) using transcription factors has been achieved from almost any differentiated cell type and has proved highly valuable for research and clinical applications. Interestingly, iPSC reprogramming of cancer cells, such as pancreatic ductal adenocarcinoma (PDAC), has been shown to revert the invasive PDAC phenotype and override the cancer epigenome. The differentiation of PDAC-derived iPSCs can recapitulate PDAC progression from its early pancreatic intraepithelial neoplasia (PanIN) precursor, revealing the molecular and cellular changes that occur early during PDAC progression. Therefore, PDAC-derived iPSCs can be used to model the earliest stages of PDAC for the discovery of early-detection diagnostic markers. This is particularly important for PDAC patients, who are typically diagnosed at the late metastatic stages due to a lack of reliable biomarkers for the earlier PanIN stages. However, reprogramming cancer cell lines, including PDAC, into pluripotency remains challenging, labor-intensive, and highly variable between different lines. Here, we describe a more consistent protocol for generating iPSCs from various human PDAC cell lines using bicistronic lentiviral vectors. The resulting iPSC lines are stable, showing no dependence on the exogenous expression of reprogramming factors or inducible drugs. Overall, this protocol facilitates the generation of a wide range of PDAC-derived iPSCs, which is essential for discovering early biomarkers that are more specific and representative of PDAC cases.

Author Spotlight: Reprogramming Cancer Cells to iPSCs to Study Disease Progression and Treatment Targets

The present protocol describes the reprogramming of Pancreatic Ductal Adenocarcinoma (PDAC) and normal pancreatic ductal epithelial cells into induced pluripotent stem cells (iPSCs). We provide an optimized and detailed, step-by-step procedure, from preparing lentivirus to establishing stable iPSC lines.

The generation of induced pluripotent stem cells (iPSCs) using transcription factors has been achieved from almost any differentiated cell type and has ...