Pathology: Hub and Integrator of Modern, Multidisciplinary [Precision] Oncology

Pathologists, who undergo years of training perfecting the skill of analyzing cells and tissues to provide accurate diagnostic and prognostic information to clinicians, are often referred to as ‘the lab’, which includes analytical assessment of biomarkers in blood and bodily fluids. This implies that diagnostic tests are performed by machines without human input. The goal of pathologists is to obtain from cells and tissues information that is interpreted in its clinical and anatomic context and integrated with molecular and digital data to determine the best treatment.

Pathological analyses, defining the parameters of human variation, characterize the effects of disease on the body and provide insight into the effects of therapeutic interventions. In fact, pathologists produce the majority of quantitative data in the medical record. It is estimated that 75% of clinical decisions are based on laboratory tests (1). While technological tools to perform these assessments have become exponentially more powerful, traditional assessment remains an essential first step in cancer diagnosis and prognosis, refined but not substituted by more sophisticated and modern techniques. For instance, identical mutations in different tissues have different effects and are treated differently. Thus, information derived from these techniques is interpreted by the pathologist in the context of the overall traditional assessment. Nearly every aspect of a cancer patient's treatment—surgery, chemotherapy, radiotherapy, immunotherapy, or treatment with small-molecule inhibitors – is dictated both by the tissue diagnosis and the molecular changes in the cancer. These are all determined by pathologists, who analyze, interpret, and integrate complex visual and digital data from tissue, laboratory, and clinical sources.

Pathology subspecialty areas include anatomical, forensic, laboratory medicine, molecular and computational. In addition, increasingly pathologists specialize in a single organ or disease. This is particularly evident in the field of oncology. As a consequence, oncologic pathologists are consistently part of multispecialty teams. Beyond clinical practice, the role played by pathologists in elucidating pathogenesis and thus affecting therapy, cannot be underestimated. In fact, the first chemotherapy (folic acid antagonists for treating leukemia) was developed by a pathologist, Sidney Farber. Pathologist Robin Warren, working with internist Barry Marshall, discovered the causative role of Helicobacter pylori in peptic ulcer disease. This discovery, earning them the 2005 Nobel Prize in Physiology or Medicine, changed our understanding of the pathogenetic mechanism of this common disorder and revolutionized treatment.

In addition to their essential role in diagnosis, pathologists, as gatekeepers of patient samples, ensure the quality of all laboratory operations based on Good Laboratory Practice and standards enforced by international organizations. In the United States, these standards are detailed in Clinical Laboratory Improvement Amendments (CLIA), which are regulated by the Centers for Medicare and Medicaid Services. In Western Europe, the European Union has a similar process in Mutual Recognition Agreements. Good Laboratory Practice is monitored by OECD (Organisation for Economic Cooperation and Development) Guides for Compliance Monitoring Procedures.

There are three basic parts to the legal and professional responsibility of pathologists for the quality and fidelity of the services that they perform and oversee for patients under CLIA – preanalytics, analytics, and post-analytics (the results). Preanalytics encompass responsibility for maintaining the quality of the specimens from the time they are removed from the patient until they are analyzed. This aspect of pathology practice has become more complex with the growth of molecular analyses. In addition, there is an overriding need to protect the molecular integrity of specimens for tests of a wide variety of molecular species on multiple different platforms. The preanalytics for Precision Medicine project team of the College of American Pathologists (CAP) has undertaken a 6-year initiative to develop evidence-based standards that will be enforced through the CAP Laboratory Accreditation Program. The goal of this initiative is to assure that the molecular data derived from patient specimens is not compromised by artifacts created by tissue procurement, handling or processing prior to analysis of any type, and that the data can be used with confidence to make patient management decisions.

In this commentary, we describe the pivotal roles pathologists have in both the care of patients and in understanding disease through the story of a hypothetical patient. And we describe future trends in pathology practice. In closing, we challenge the field of pathology to embrace new, evolving methods (artificial intelligence, image analysis, genomics, and computational biology) and to educate the next generation of pathologists. These methods are poised to revolutionize personalized medicine.

A 57-year-old male never-smoker presented to the ER with subacute chest pain. Cardiac troponin I assays were performed in the clinical laboratory at 5, 10, and 14 hours after symptom onset to address the possibility of a myocardial infarct. Troponin I values did not exceed the upper reference limit value (2). This result, with the physical exam, clinical history, and EKG, ruled out myocardial infarction. Consequently, additional workup was begun. A cardiac CT scan detected a lung mass in the left lower lobe. PET/CT demonstrated increased uptake in this mass and in multiple mediastinal lymph nodes. The patient underwent endobronchial ultrasound (EBUS)-guided biopsies to obtain tissue for diagnosis and staging. To increase the diagnostic yield of this procedure, onsite cytologic evaluation was performed to assess tissue adequacy.

The cytopathologist examined Giemsa-stained smears of EBUS fine needle aspirates of the mediastinal nodes, identifying clusters of malignant epithelial cells amongst numerous lymphocytes, confirming the diagnosis of metastatic carcinoma. This procedure, performed in real time, assured the endoscopist that diagnostic tissue was present and provided staging information. IHC stains of lymph nodes showed expression of cytokeratin 7, an epithelial marker, and TTF-1, which is expressed by cancers of thyroid and bronchial origin. These stains confirmed the diagnosis of non–small cell lung adenocarcinoma (NSCLC). The patient was staged as stage IIIB, thus not a surgical candidate. Based on discussion of treatment options at the multidisciplinary tumor board, the patient received chemoradiation. A year after completion of therapy, the patient developed ataxia. A cerebellar lesion was resected, followed by stereotactic radiation. Portions of the lesion were biobanked. The tissue specimen, which was analyzed using Next Generation Sequencing (NGS) by the molecular pathologist, had an EGFR p.E746_A750del (NM_005228.3:c.2235_2249del) mutation (4). No T790M mutation was noted. This deletion in exon 19 is associated with sensitivity to the tyrosine kinase inhibitor erlotinib, which was administered (5). A year into therapy, the patient developed systemic disease. Since this was not amenable to biopsy, a blood sample was sent to the clinical laboratories to assay circulating tumor DNA (ctDNA), following ASCO CAP guidelines for liquid biopsies (6). The molecular pathologist reported a new T790M resistance mutation. Osimertinib, an oral tyrosine kinase inhibitor active against this mutation, was administered. Initially responding to this therapy, the tumor ultimately progressed after 18 months of treatment. A new liver lesion was biopsied. An amplification of the MET gene was detected by NGS and the patient was enrolled in a clinical trial targeting this resistance mechanism.

Normal human specimens of high quality are needed both to develop diagnostics and to further our understanding of human biology. The value of normal tissues is exemplified by the Genome Tissue-Expression (GTex) project and more recently by the Human Tumor Atlas Network (HTAN) effort (https://humantumoratlas.org; ref. 7). Pathologists facilitate the acquisition of high quality normal human tissues from rapid autopsies and surgical specimens. Understanding the complexities of normal tissue function informs our understanding of cancer biology.

Pathologists are the “virtual eyes”, guiding in real time the surgeon's hand in complex resections, to make sure margins are free of tumors. Similarly, the pathologist assures the interventional radiologist that intact cancer tissue is obtained at biopsy for subsequent analysis. Finally, pathologists are an integral member of disease management teams that decide in a multidisciplinary setting the optimal therapy and/or clinical trial accrual of specimens.

The accuracy of high throughput (“omics”) analyses depends on high quality input of tissues and body fluids. Pathologists can control some of these variables, that is, cold ischemia time, sample transport vehicle, and analyte stabilization method (8). Consultations between the clinician obtaining the sample and the pathologist handling it can minimize the effect of preanalytic variables and subsequent interpretation of results. In addition, when archival tissue of variable age and unknown preanalytic effects is used for research, the pathologist can use assays to ensure the adequacy of the sample and/or to provide a means to normalize data for analysis, that is, using housekeeping genes for RNA and protein or spike-in for RNA and DNA (9).

Analysis of patient samples using hematoxylin and eosin sections, DNA, RNA, epigenetic, and protein-based assays provides insight into the evolution of the patient's tumor, the tumor microenvironment, and the response to therapy. In clinical trials, pathologists confirm that tissue samples contain nonautolyzed tumor cells. Correlative analyses, biomarker development, and validation are largely performed by pathologists, who can assure that these assays are performed on adequate tumor areas by careful microscopic analysis with or without microdissection.

Rapid autopsy services enable an autopsy to be done within hours of death, minimizing tissue autolysis and preserving analytes- proteins and nucleic acids. Fresh tissue from primary and metastatic sites can also be used to generate “living tissues” (xenografts and organoids). Studies of separate metastases have revealed the extent of histologic, IHC, and molecular heterogeneity and clonal evolution in patients with cancer (10–13). By histologically confirming that masses are cancer and by macrodissecting tumor, the pathologist enriches tissue samples for tumor cells. By separating supporting tissue, the pathologist aids in the interpretation of the role of cell-autonomous alterations such as mutations as well as cross-talk with the microenvironment, for example, the immune cell infiltrate.

Digitizing entire slides allows pathologists to more efficiently characterize tissue histology for diagnosis. Once the images of whole slides are converted into millions of pixels, they can be viewed by multiple people at multiple locations, simultaneously. The images can be paired with radiologic images and clinical information to give pathologists and the patient's health care team a unified picture of each person's cancer. As the universe of molecular analysis of cancer expands, we have learned that each cancer is defined not only by location and histologic type, but also by genomics. What has become apparent, through artificial intelligence (AI), is that the histomorphologic appearance of a tumor is the repository of extensive information. For this reason, AI holds the potential for a higher level of understanding of tumor biology, only partially and nonquantitatively perceivable by the human eye.

These benefits pale in comparison to applications that will transform digital pathology into computer-aided diagnosis by AI. The pathologist can automate manual, labor-intensive tasks, such as detecting “rare events,” that is, tiny tumor metastases within lymph nodes. AI will also improve consistency in grading and subtyping tumors and/or other diseases. With recent advancements in deep learning, computers will focus on informative histologic patterns. Deep learning can extract complex abstractions as data representations through a hierarchical learning process. AI can provide robust predictions about the disease and create risk profiles “signatures.”

For example, molecularly defined cancer subtypes or specific tumor microenvironments can be detected with accuracy using AI, whereas these features are not apparent to the human eye (14–17). Prognostic features of a tumor can also be detected (15–17). The predictive algorithms acquired by deep learning can be downloaded and used on a variety of platforms, even an iPhone. In a process not unlike a diagnostic test, pathologists can run the algorithm to ask whether their patient's digitized slide represents a specific cancer type or not. AI is likely not to replace the pathologist, but to serve as an ancillary test in the pathologists' tool kit.

In addition to digital pathology, novel imaging technologies, such as 3D microscopy, tumor scanning microscopy, multiplex immunophenotyping, optical coherence tomography, light CT, and ultraviolet-surface excitation. Quantitation and spatial location of targets is important for tissue-based biomarker analysis and spatial genomics. Instruments that perform these high-throughput, simultaneous assessments of proteins and transcript levels will soon transition to routine clinical practice. Recently developed technologies can resolve up to 40 protein biomarkers on a single section of fresh/frozen human or murine tissue, FFPE tissue, or TMA blocks using bar code-based multiplex fluorescence microscopy (18) or secondary ion mass spectrometry (19). The spatial distribution of up to 30 protein targets using in situ imaging of cellular antigens or spatial gene expression profiling of >1,800 mRNA targets can be determined (20). Software used to analyze quantitative data with a spatial component can only be analyzed by pathologists with knowledge of cell types and tissue structure. Likewise, a pathologist is needed to assign cell categories to single-cell sequencing studies.

With rapid advances in genomics, proteomics, metabolomics, informatics, computational biology, image analysis, and artificial intelligence, we are on the brink of a truly transformational period where the promise of personalized medicine and precision oncology can be realized. These novel technologies, largely developed by people working outside of pathology - engineers, molecular and cellular biologists, mathematicians, statisticians and computer scientists, will provide immense opportunity to pathologists. Our training programs must therefore adequately prepare medical students, as well as residents and fellows of all disciplines for the medicine that will be practiced. Pathology trainees need to be fully versed in bioinformatics, genomics, and AI in addition to the core principles of anatomic and clinical pathology. The pathology community should not assume that the skill set that got the discipline here is the same one needed to move forward (21).

Rudolf Virchow in the 19th century expanded pathology from an autopsy discipline to one that served both biomedical research and evidence-based clinical care. His work led to the first reproducible categorization of disease diagnoses. This development catalyzed many subsequent advances in clinical medicine.

Medicine today is at a point analogous to its position in the mid-19th century where many rapidly evolving advances in technology are set to transform the enterprise. As documented in our case history, and in our vision for the future, pathology will continue to play an important role in the diagnosis of disease and the selection of specific therapies. Novel technologies, such as molecular diagnosis and image analysis, are necessary adjunct to traditional, morphology-based pathology. By embracing new technologies, making them part of the present pathology repertoire, pathology, can and will expand its scope to provide more precise and essential information for disease diagnosis, classification, prognosis and therapy. The pathologist has evolved to become the medical professional who not only “guides the surgeons' hands”, but also decision- making by oncologists.

With sequencing of the human genome and rapid advances in genomics, proteomics, metabolomics, informatics, computational biology, image analysis, and artificial intelligence, we are on the brink of a truly transformational period where the promise of personalized medicine and precision oncology can be realized. These novel technologies, largely developed by people working outside of pathology - engineers, molecular and cellular biologists, mathematicians, statisticians and computer scientists - will provide immense opportunity to pathologists. To play a leadership role in this revolution pathology must evolve, incorporating these technologies. Pathology residency curricula will continue to expand teaching bioinformatics, genomics, and AI training, to supplement the core principles of anatomic and clinical pathology. These skills will characterize the pathologist of the future (21).

Carolyn C. Compton, School of Life Sciences, Arizona State University, Scottsdale, Arizona.

Angelo M. De Marzo, Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, Maryland.

Jayanta Debnath, Department of Pathology, University of California, San Francisco, San Francisco, California.

Keith D. Eaton, Department of Medicine, University of Washington, Seattle, Washington.

Kojo Elenitoba-Johnson, Center for Personalized Diagnostics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania.

Michelangelo Fiorentino, Dipartimento di Medicina Specialistica Diagnostica e Sperimentale DIMES, University of Bologna, Bologna, Italy.

Christopher A. French, Department of Pathology, Brigham and Women's Hospital/Harvard Medical School, Boston, Massachusetts.

Thomas J. Fuchs, Hasso Plattner Institute for Digital Health, Icahn School of Medicine at Mount Sinai, New York, New York.

Felice Giangaspero, Department of Radiological, Oncological and Anatomic Pathology Sciences, University Sapienza of Rome, Rome, Italy; and IRCCS Neuromed, Pozzilli (Is), Italy.

Jiaoti Huang, Department of Pathology, Duke University School of Medicine, Durham, North Carolina.

A. John Iafrate, Massachusetts General Hospital, Boston, Massachusetts.

Michael M. Ittmann, Department of Pathology, Baylor College of Medicine, Houston, Texas.

Roy A. Jensen, Department of Pathology and Laboratory Medicine, University of Kansas Cancer Center, Kansas City, Kansas.

Annette S. Kim, Department of Pathology, Harvard Medical School/Brigham and Women's Hospital, Boston, Massachusetts.

Massimo Loda, Department of Pathology, Weill Cornell Medicine, New York, New York.

Robin G. Lorenz, Research Pathology, Genentech, Inc., South San Francisco, California.

Tamara L. Lotan, Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, Maryland.

Anirban Maitra, Sheikh Ahmed Bin Zayed Al Nahyan Center, UT MD Anderson Cancer Center, Houston, Texas.

Gerrit A. Meijer, Department of Pathology, Netherlands Cancer Institute, Amsterdam, the Netherlands.

Richard N. Mitchell, Department of Pathology, Harvard Medical School/Brigham and Women's Hospital, Boston, Massachusetts.

Anil V. Parwani, Anatomical Pathology, The Ohio State University Wexner Medical Center, Columbus, Ohio.

Francesco Pezzella, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom.

Eli Pikarsky, Department of Pathology, The Hebrew University of Jerusalem, Jerusalem, Israel.

Jorge S. Reis-Filho, Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, New York.

Andrea L. Richardson, Department of Pathology, Johns Hopkins University School of Medicine, Washington, DC.

Mark A. Rubin, Department for BioMedical Research (DBMR) and Bern Center for Precision Medicine, University of Bern, Bern, Switzerland.

Lawrence D. True, Department of Pathology, University of Washington, Seattle, Washington.

A.M. De Marzo reports grants from Janssen R&D and personal fees from Cepheid Inc and Merck outside the submitted work. J. Debnath reports personal fees from Vescor and Merck outside the submitted work. K.D. Eaton reports grants from Mirati Therapeutics and personal fees from Lilly outside the submitted work. K. Elenitoba-Johnson reports grants from Thermofisher outside the submitted work. C.A. French reports grants, personal fees, and other support from Boehringer-Ingelheim and Glaxo-Smith-Kline, grants from c4 Therapeutics, and other support from Daichi-Sanyo and Vertex outside the submitted work. T.J. Fuchs reports personal fees and other support from Paige.ai Inc. outside the submitted work. J. Huang reports personal fees from Kingmed, MoreHealth, OptraScan, Genetron, Omnitura, York Biotechnology, Genecode, VIVA Biotech, Sisu Pharma, and Vetonco and grants from Zenith Epigenetics, BioXcel Therapeutics, and Fortis Therapeutics outside the submitted work. A. Iafrate reports other support from ArcherDx and personal fees from Kinnate Therapeutics, Repare Therapeutics, PAIGE.AI, and Oncoclinicas Brasil during the conduct of the study; in addition, A. Iafrate has a patent for Anchored Multiplex PCR licensed and with royalties paid from ArcherDx. M. Loda reports other support from Methylex outside the submitted work. R.G. Lorenz reports other support from Genentech, Inc. outside the submitted work. T.L. Lotan reports grants from Roche/Ventana, DeepBio, and Myriad outside the submitted work. A. Maitra reports other support from Cosmos Wisdom Biotechnology and Thrive Earlier Detection, an Exact Sciences Company, and personal fees from Freenome outside the submitted work. G.A. Meijer reports other support from Hartwig Medical Foundation, Sysmex, and Exact Sciences and grants from CZ Health Insurance outside the submitted work; in addition, G.A. Meijer has several patents pending and is co-founder and board member (CSO) of CRCbioscreen BV. G.A. Meijer also has research collaborations with Sysmex, Sentinel Ch. SpA, Personal Genome Diagnostics (PGDX) and Hartwig Medical Foundation: these companies/foundations provide materials, equipment and/or sample/genomic analyses. E. Pikarsky reports personal fees from Roche and MSD outside the submitted work. J.S. Reis-Filho reports personal fees from Paige.AI during the conduct of the study, as well as personal fees from Repare Therapeutics, Goldman Sachs, Grupo Oncoclinicas, Roche Tissue Diagnostics, Novartis, Genentech, Roche, In Vicro, Eli-Lilly, and Personalis outside the submitted work. A.L. Richardson reports a patent for NtAI for Myriad Genetics for MyChoice HRD test licensed and with royalties paid from Myriad Genetics. L.D. True reports being co-founder of LightSpeed Microscopy, Inc. and holding equity in the company. No disclosures were reported by the other authors.

We wish to thank Margaret Foti and the AACR for their generous support of this project. We also wish to acknowledge AACR Pathology Task Force members, Andrew H. Beck, Michael J. Kluk, David L. Rimm, and Ignacio I. Wistuba for their insightful observations during early discussions of the topics in the manuscript.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.