Skip to main content
Discovering the causes of cancer and the means of prevention
 

Molecular and Digital Pathology Laboratory Expands Capacity for DCEG Studies

, by Maura Kate Costello, M.A.

In 2016, the Molecular and Digital Pathology Laboratory was founded to offer investigators in the Division of Cancer Epidemiology and Genetics (DCEG) an array of specialized services for high-quality tissue pathology in epidemiological studies.

Capitalizing on Existing Resources

Group photo of MDPL team in two rows.

(Back, left to right) Yelena Golubeva, senior scientist; Scott Lawrence, associate scientist; Belynda Hicks, CGR Deputy Director

(Front, left to right) Lisa Garland, research associate; Petra Lenz, pathologist; Mary Olanich, scientific project manager

DCEG maintains a rich collection of over 12 million biospecimens from high-quality epidemiological and clinical studies on a diverse array of cancer types. These include 300,000 tissue specimens from 275 family and population-based studies. With repositories like this and advancing analytical technologies, tissue-based analyses in epidemiological research are rapidly expanding.

The Molecular and Digital Pathology Laboratory (MDPL) is an extension of the Cancer Genomics Research Laboratory (CGR) that integrates histological and molecular tissue profiling with analyses of environmental and genetic risk factors. Through molecular pathology analysis, researchers can examine the impact of genetic variation on the biology of gene expression and protein function at the tissue level and provide critical insights on cancer risk and progression.

Montserrat García-Closas, M.D., Dr.P.H., deputy director of DCEG and acting chief of the Integrative Tumor Epidemiology Branch (ITEB), and others recognized the need to capitalize on existing resources and expertise and called for the creation of MDPL in 2016.

Montserrat García-Closas

Montserrat García-Closas

“The use of innovative technologies and data analytics is providing unprecedented opportunities to study etiological and clinical questions in molecular epidemiological studies,” notes Dr. García-Closas.

In addition to conducting analyses, the chief tasks of the MDPL are to standardize and streamline methods, as well as increase capacity of processing tissue samples for histopathology, digital pathology, and molecular/genomic analyses. The often-varying quality of tissue specimens poses a challenge to maintaining consistency and reproducibility across findings. Samples may be processed as fresh-frozen tissue, formalin-fixed paraffin-embedded (FFPE) tissue (including blocks, tissue microarrays, and loose cores); or glass slides (stained or unstained) kept at ambient temperatures or in a cold room.

From the outset, the goal of MDPL has been to offer DCEG researchers specialized services for high-quality tissue pathology in epidemiological studies. Housing MDPL within the well-established CGR provides efficiencies across the board and ensures full access to the biospecimen repository and other resources, not the least of which is the expertise and support of CGR staff for nucleic acid extractions, genomic analyses, and bioinformatics.

“The MDPL allows CGR to provide a more comprehensive view into the whole life cycle of an analysis—germline and somatic genetics, gene expression and methylation, proteomics, as well as viral, bacterial, or fungal influences on cancer etiology,” explains Belynda Hicks, M.S., CGR Deputy Director. “Previously, we focused on processing germline samples and a small number of somatic analyses. Now, we can start directly from the tissue and either perform protein biomarker analyses or begin with nucleic acid extractions for somatic and germline genomic analyses.”

Pipeline for Processing a Tissue Sample

Processing and analyzing tissue samples for large etiologic studies can take several months and often well over a year, primarily due to the varying quality of tissue samples. Mary Olanich, Ph.D., project manager of MDPL, explains, “We’re working with all sorts of tissues, including archival material, or materials from hospitals that may follow different protocols than ours.” Hospital lab technicians collect tissues from patients for diagnostic purposes and typically do not follow study-specific protocols. Samples from tissue banks collected decades earlier may have been stored using different methods and may be of lower quality. Tissue preparation and loan policies can vary by laboratory. Sometimes it is possible to get a full FFPE block, usually as a short-term loan. More often, labs will share only a few slides or tissue scrolls.

Although tissue collection may not be standardized, MDPL is establishing protocols for processing and analyzing samples to reduce heterogeneity and increase the quality of data generated for epidemiological investigations. A team of highly experienced pathologists, molecular biologists, and imaging experts carry out the steps needed to accomplish the scientific aims of tissue-based studies.

What Happens in the Lab: Services Provided by MDPL

There are several analytical approaches that can be applied to any given sample. Each approach requires different treatment at the outset, making coordination and communication between the investigator and the MDPL staff essential.

Dr. Olanich serves as the initial point of contact. She reviews the study goals and types of samples available and drafts a project plan with the investigator. “Understanding the quality and usability of the specimen is key, as well as knowing the research question at the beginning," Dr. Olanich explains. The first step almost always includes a pathologist to review the specimens. Next, depending on the research question, the tissue may undergo various procedures, including immunohistochemistry (IHC) or immunofluorescent (IF) staining and digitization, macrodissection or laser-capture microdissection, followed by nucleic acid extraction.

Before and after slides of stained slide of breast tumor tissue showing the repair of a tear in the original sample.

Compromised archival FFPE breast tumor block (left) has been re-embedded to obtain a high-quality section (right).

After quality control and nucleic acid extraction, Dr. Olanich works with DCEG researchers to determine suitability for DNA and RNA genomic analyses. An investigator can select from multiple options which analyses are best suited to the analytical goal and the sample quality and quantity, including targeted or whole exome sequencing, expression analysis (either targeted, direct, digital detection or whole transcriptome sequencing) and both genotyping and methylation analysis. Streamlining all the procedures through one integrated facility is critical to increase efficiency and ensure high-quality, coordinated analyses of precious tissue samples.

Dr. Golubeva sections an FFPE block with machine.

Dr. Golubeva sections an FFPE block.

With a project plan in place, MDPL and DCEG staff work together to optimize experimental and analytical approaches for a wide variety of research questions. For example, an investigator who wants to focus on biomarker analysis will collaborate with pathologist Petra Lenz, M.D., to determine how to use effectively the existing specimens, slides, images, or data. Since analysis may be limited by the amount or quality of specimen, an iterative process of evaluation is often necessary to strike the right balance between optimal methods and the amount of sample required to conduct that analysis.

Various types of tissues come with potential challenges. In some instances, slides arrive in the lab with bubbles, or with damaged or folded tissue, any of which makes the sample unusable. If a slide is heavily marked from previous annotations, fortunately, MDPL staff can digitally erase those markings to refurbish the slide. Occasionally, FFPE block scrolls are rejected because they do not contain any tumor tissue. Fragile archival material has potential to break during sectioning. With re-embedding techniques, however, MDPL staff have been able to salvage many such samples, some over 30 years old.

A multi-tumor tissue array was stained and imaged with a fluorescent multiplex assay. Next, each core was analyzed to measure the density of the various target biomarkers expressed.

Yelena Golubeva, Ph.D., a senior scientist with over 40 years of experience in general and molecular histopathology, is responsible for preparing tissue samples for analysis. Here again, knowing the research goal at the outset is essential. If the lab receives an FFPE block, she can section it into scrolls for nucleic acid extraction or cut it into sections for automated staining with hematoxylin and eosin (H&E) or IHC. Dr. Golubeva cuts and stains a few sections throughout the block at set intervals in order to chart the landscape of the entire sample. The sections that are not used for mapping are saved for IHC staining or nucleic acid extraction. If the sample is destined for IHC staining, she must perform serial sectioning—maintaining the sequence and orientation of consecutive sections—for downstream digital analysis, whereas that information is not as important for nucleic acid extraction.

If an investigator wants to characterize further a specific cell population in the tissue, such as tumor infiltrating lymphocytes (TILs), additional slides are prepared for staining with IHC or multiplex immunofluorescence (IF). With IHC staining, each slide accommodates one to three chromogenic stains. Multiplex IF can stain a single sample with up to eight fluorescent markers, maximizing use of finite tissue samples and allowing investigators to see how biomarkers may relate spatially to each other and together contribute to cancer progression. “‘Hyper-plex’ staining is coming soon,” explained Dr. Golubeva, “which will make it possible to detect dozens or even hundreds of markers at once, possibly as early as the end of 2020.”

Two sets of before and after slides, one of macrodissection (A), and the other of laser capture microdissection (B).

A. Bladder tumor tissue annotated in yellow for macrodissection (top). Tissue macrodissected by hand (bottom) for nucleic acid extraction. B. Prostate tumor tissue dissected with laser-capture microdissection.

Once the incoming material has been assessed for quality, sectioned, and stained, Dr. Golubeva sends H&E slides for pathology review, often performed by Dr. Lenz. If the sample is destined for nucleic acid extraction, Dr. Lenz will mark areas for Dr. Golubeva to dissect. In macrodissection, the excision of tumor material is performed by hand. Laser-capture microdissection can also be utilized to excise individual tumor cells and is performed by specialized instrumentation. These tissues or cells are then delivered to the extraction lab. The lab is also expanding its capability to accommodate fresh-frozen samples, a sample type that can yield a greater amount of genetic material for analysis.

Scott Lawrence, M.S., an associate scientist who most recently worked in pharmacodynamics, and Lisa Garland, B.S., research associate, digitize slides for upload into the MDPL image database—accessible to internal and external collaborators alike—for pathology evaluation, and/or automated image analysis. The instrument can hold up to 400 slides and can scan up to 400X magnification, making cellular details visible. Jill Koshiol, Ph.D., Earl Stadtman tenure-track investigator in the Infections and Immunoepidemiology Branch, worked with Mr. Lawrence to set up a web-based analysis suite for a pilot within the Chile Biliary Longitudinal Study, allowing a Chilean pathologist to review images from thousands of miles away. Currently, the MDPL image repository supports over 30 DCEG studies with 85 collaborators and hundreds of thousands of images.

Scott Lawrence manually assesses tumor infiltrating lymphocyte (TIL) density in a digitized breast cancer biopsy stained with H&E (left). Lisa Garland loads slides into scanner to be digitized (right).

Mr. Lawrence collaborates with investigators to develop algorithms that analyze images for the presence, quantity, and spatial orientation of certain proteins. He is working with Mustapha Abubakar, M.D., Ph.D., a postdoctoral fellow in ITEB, to measure TILs in tumor and adjacent normal tissue with the hope of determining immunological activity in the tumor microenvironment. “Understanding the activity at the tumor border provides significant insights into clinical outcomes, such as gauging patient responsiveness to certain classes of drugs,” Mr. Lawrence explains.

Digital analysis can also offer objective quantification of target proteins by computationally measuring the intensity of a stain. Traditionally, pathologists would score IHC stains with the naked eye, a laborious task that can sometimes lead to inconsistent or inaccurate results. Mr. Lawrence also performs digital area assessment, a method to estimate the number of sections needed to yield enough genetic material for nucleic acid extraction for sequencing DNA and RNA. This helps to ensure that all samples are used efficiently, without waste.

MDPL provides support for digital image analysis to address a wide variety of research questions, including marker density, cell membrane expression, tumor/stromal ratio, multiplex IF, chromogenic ISH, vascular density, spatial relationships, and tumor microarray dearraying and analysis.

Active Projects and Early Accomplishments

In just a few years, MDPL has produced critical resources and provided an infrastructure that has made innovative studies possible across the Division. Below are short summaries of selected projects.

Studies of Lung Cancer

Maria Teresa Landi, M.D., Ph.D., senior investigator in ITEB, and colleagues are conducting a study on lung cancer among never-smokers. Sherlock-lung is an international collaboration to collect tumor and adjacent normal tissue from 2,500 cases. Utilizing the unique capabilities of the MDPL, the investigators will perform genomic analysis and analyze the tumor microenvironment. They will integrate the molecular landscape with histological and radiological features in order to develop—for the first time—a more refined classification of lung cancer in never-smokers that may provide insights into prognosis and treatment strategies.

In a study of lung cancer among workers with occupational exposure to diesel exhaust (DE), Stella Koutros, Ph.D., M.P.H., tenure-track investigator in the Occupational and Environmental Epidemiology Branch (OEEB), and collaborators are combining exposure data with tumor molecular profiles to evaluate whether DE exposure results in unique genomic changes. These efforts could provide important mechanistic insights into DE-induced lung cancer and new information about the role of DE on cancer sites other than lung.

A set of before and after images displaying the same tumor sample in its H&E stain (before) and after digital analysis to detect epithelial, stromal, and adipose tissue components.

Supervised machine-learning used on H&E breast biopsy slides (left) detects epithelial (red), stromal (green), and adipose (yellow) tissue components (right).

Studies of Breast Cancer

Dr. Abubakar is utilizing digital pathology algorithms and multiplex staining to characterize the samples from over 5,000 breast cancer cases collected from five studies. By combining that information with epidemiologic, demographic, and clinical data, he hopes to understand better the role of the tumor microenvironment as an intermediary between risk factors, tumor behavior, and clinical outcomes. This study is supported by an Intramural Research Award, and utilizes data from the Prostate, Lung, Colon, and Ovary Study, Ghana Breast Health Study, Polish Breast Cancer Study, Asian Breast Cancer Study, and the BREAST Stamp Project.

Using the Hong Kong Breast Study, Xiaohong Rose Yang, Ph.D., M.P.H., senior investigator in ITEB, is evaluating TIL levels and investigating associations with germline and somatic factors, as well as gene expression and methylation, to determine risk and prognostic factors.

Brittny Davis Lynn, Ph.D., M.P.H., an independent research scholar in ITEB, conducted a study investigating the association between terminal duct lobular unit (TDLU) involution and molecular subtypes of breast cancer in black women, using data from the Black Women’s Health Study. TDLUs are structures that give rise to most breast cancers, and involution is the process by which TDLUs decrease in number and size. Dr. Davis Lynn and colleagues found that women with triple negative breast cancer—an aggressive, early-onset subtype more common in black women—had reduced TDLU involution compared to those with luminal A tumors, which may reflect diverse etiology and pathology of breast cancer in black women.

H&E images from patients with (A) luminal A and (B) triple negative breast cancer.  In the images above, the normal tissue adjacent to a triple negative tumor (B) shows a larger TDLU, and therefore reduced involution, compared to the smaller TDLU in normal tissue adjacent to a luminal A tumor (A).

A TDLU in the normal tissue adjacent to a luminal A tumor (A) is smaller than in the normal tissue adjacent to a triple negative tumor, showing reduced involution (B).

Gretchen Gierach, Ph.D., M.P.H., deputy chief and senior investigator in ITEB, and her team collaborated with Dr. Lenz to show that microvessel density measured in breast tissue is associated with breast cancer, independent of mammographic breast density.

Together with MDPL, Dr. Gierach and Dr. Abubakar are assessing whether quantitative and objective digital histologic approaches for characterizing breast tissue composition may provide high-throughput methods suitable for epidemiologic applications to further our understanding of breast cancer etiology. Drs. Gierach, Abubakar and Clara Bodelon, Ph.D., M.S., staff scientist in ITEB, are also conducting molecular (e.g., gene expression profiling) and morphometric analyses (e.g., TDLU involution metrics) to provide mechanistic insights into the biologic underpinnings of breast cancer risk and progression.

In collaboration with Dr. Lenz, Drs. Gierach and Abubakar, Cody Ramin, Ph.D., a postdoctoral fellow in the Radiation Epidemiology Branch (REB), and Lea Widemann, a summer student in ITEB, are developing a standardized protocol for characterizing histological features in breast adipose tissue called crown-like structures, which are thought to influence local estrogen biosynthesis. This project follows up on prior reports from ITEB investigators (Mullooly M, et al. Relationship between crown-like structures and sex-steroid hormones in breast adipose tissue and serum among postmenopausal breast cancer patients. Breast Cancer Res 2017). They aim to determine the role of these structures in breast carcinogenesis.

Dr. García-Closas is leading a study of the etiology and prognosis of molecular subtypes of breast cancer in the Breast Cancer STratification Study (B-CAST), a large-scale international pooling project. B-CAST has tissue specimens from over 20,000 cases that have been stained for 15 different IHC biomarkers and digitized by collaborating centers in Cambridge, United Kingdom. These samples will enable a better understanding of the etiologic heterogeneity and prognosis of breast cancer subtypes, including the role of immune infiltration in tumors. To advance these efforts, Dr. Abubakar and MDPL staff are developing automated, high-throughput image analysis strategies to evaluate specimen quality and perform large-scale unsupervised scoring of biomarkers, while ensuring data integrity.

Drs. García-Closas and Abubakar are working on a breast cancer study with collaborators in Ghana to characterize the molecular landscape of breast cancer in African women and to determine the relation of environmental and genetic risk factors in this understudied population. This project uses breast biopsies collected prior to treatment from the population-based case-control study.

Studies of Urinary Cancers

Mark Purdue, Ph.D., senior investigator in OEEB, has recently expanded a study on renal cancer to determine whether certain subtypes have distinct risk factor profiles. MDPL will perform RNA extraction on tumor tissue to assist this investigation.

Similarly, Dr. Koutros is characterizing somatic mutations across 44 frequently mutated genes in non-muscle invasive bladder cancer tissue from the New England Bladder Cancer Study. This work provides some of the first data to evaluate comprehensively the relationship between risk factors and somatic mutations in non-muscle invasive bladder cancer, which makes up approximately 70 percent of tumors.

Maximizing a Resource for the Future

Over the last three years, MDPL has established the necessary expertise and infrastructure to support a wide range of tissue-based studies, with increasing capacity for the future. These studies address important etiological and clinical questions, and have begun to yield scientific publications, listed at the bottom of this page, with several other manuscripts under review. Moving forward, MDPL will continue to incorporate new technologies, such as analyzing protein biomarkers and nucleic acid expression and variation on the same slide, which promises unbounded potential for functional studies, among others.

Additionally, they are rapidly building a resource of digital images to be evaluated with emerging technologies. Jonas S. Almeida, Ph.D., chief data scientist, and experts in the NCI Center for Biomedical Informatics and Information Technology are collaborating to expand cloud-computing capabilities for the application of artificial intelligence to image analysis.

The greatest challenge will remain the variety and quality of samples. “The technologies coming down the road are tremendous,” Mr. Lawrence explains, “but the quality of samples has to be there.” MDPL is establishing methodologies and protocols that will yield greater confidence in the ability to assess the quality of a wide variety of samples and continue to meet the demands of tissue-based studies in cancer epidemiology research.

Scientific Publications

< Older Post

Bone and Soft Tissue Sarcoma Risk Patterns Among Survivors of Hereditary Retinoblastoma

Newer Post >

Conversation with Former Fellow, Diana Merino, Ph.D., M.Sc.

If you would like to reproduce some or all of this content, see Reuse of NCI Information for guidance about copyright and permissions. In the case of permitted digital reproduction, please credit the National Cancer Institute as the source and link to the original NCI product using the original product's title; e.g., “Molecular and Digital Pathology Laboratory Expands Capacity for DCEG Studies was originally published by the National Cancer Institute.”