E.5 Tumor Types and Subtypes that will be studied by the ICGC

The aim of the ICGC is to provide a comprehensive description of the somatic genomic abnormalities present in the broad range of human tumors. Given our current knowledge of the heterogeneity of tumor types and subtypes, the ICGC set a goal of coordinating approximately 50 projects, each of which will generate the genomic analyses outlined elsewhere in this document on approximately 500 cancer samples of each class. It is well recognized, however, that cancer is highly heterogeneous and hundreds of types/subtypes can be defined. Therefore, the stated goal of 50 ICGC projects is not intended to, and cannot, exhaustively cover the full spectrum of cancer types.

With the overall principles of the ICGC in mind, the Clinical and Pathologies Working Group discussed extensively the issues pertaining to selection of specific tumor types/subtypes and defining the criteria/parameters for each project. Recognizing the varying impacts of different cancers in different countries or global regions, the working group recommended that the ultimate justification for an ICGC Project on a specific cancer type will rest with the collaborative groups of pathologists, genomicists, clinical oncologists, cancer geneticists, cancer biologists and epidemiologists that will be proposing each project.

In selecting each project, ICGC members should articulate its importance and relevance based on the public health impact of a cancer type and unmet clinical need. The potential for novel insights that might broadly inform cancer research should also be considered. For example, testis cancer is unique amongst solid tumors in its high cure rate by conventional chemotherapy when widely metastatic. Depending on the rationale, the subtype or subtypes being proposed may be defined on pathological, molecular, etiological or geographical differences. The scientific merit of each definition of a tumor type for ICGC must be sound and technical feasibility should be addressed. In the case of tumor classes with multiple subtypes characterised by distinct clinical / biological behaviours, a balance may be drawn between studying 500 cases of exclusively one subtype and hence providing optimal power for that subtype, versus including multiple subtypes with lesser power for any one but enabling informative comparisons.

The ICGC aims to study cancers of all major organ systems including central nervous system, hemopietic and lymphoid tissue, head, neck and nasopharynx, skin, lung, breast, esophagus, stomach, colon, rectum, kidney, bladder and urinary tract, soft tissues, bone, pancreas, gall bladder and biliary system, liver, ovary, uterus, cervix, endocrine tissues, testis and prostate. It is also envisaged that the studies will cover adult and childhood / adolescent cancers, for example neuroblastoma, pilocytic astrocytoma, medulloblastoma, osteosarcoma and childhood leukemias.

Box 3. Guidelines for the selection of cancer genome projects

The Clinical and Pathologies Issues Working Group used the following examples to illustrate some of the issues which will affect the selection of cancer genome projects:

  • For many common cancer classes, there are well recognized, clearly defined histopathological or molecular subtypes with differing etiologies or geographical prevalence that merit separate projects, for example transitional cell and squamous carcinomas of the urinary bladder or squamous and adenocarcinomas of the esophagus. Indeed, the classical histological subtypes of some very common cancers, for example adenocarcinoma, squamous carcinoma, small cell carcinoma and large cell carcinoma of the lung are already known to harbour genetic differences. Therefore, each of these might reasonably be represented by a separate ICGC project. For other common cancers, subclassification has recently improved and requires characterization beyond conventional histopathology. For example, in breast cancer the classification of luminal A ER positive PR positive, Luminal B, triple negative, and HER2 positive cancers based on expression of molecular markers is currently believed to optimally reflect the diverse biology of this disease. Similarly in colon and rectal cancer, the presence or absence of mismatch repair defects and the high or low prevalence of methylation changes may represent informative modes of choosing subclasses for study.
  • For some cancer classes in which one subtype has a significantly higher incidence or mortality than others, it may be pragmatic to focus exclusively on this subtype, for example clear cell renal carcinoma, papillary thyroid cancer and ductal carcinoma of the pancreas. Alternatively, it may be appropriate to formulate two projects, one based on the most common subclass, for example in ovarian cancer a project on serous carcinoma and a second based on one or more of the rarer subclasses of mucinous, clear cell and endometrioid. Similarly, it may be appropriate to choose a small number of the commoner, better-defined classes of soft tissue sarcoma.
  • Among some cancer classes, pragmatic issues of tissue collections may argue for separating different stages of progression into separate projects. For example, patients with androgen independent (AI) or metastatic prostate cancers do not routinely undergo surgery. Therefore collection of AI or metastatic prostate specimens requires a different protocol from collection of primary prostate tumors. Similarly, in malignant melanoma, samples from distant metastases may be easier to obtain than from primary tumors, where collection is constrained by medico-legal issues. Under these scenarios, it is reasonable to propose independent projects on primary and metastatic stages of these cancer types. For other tumor types, it may be important to distinguish subtypes based on their patterns of progression, for example secondary glioblastomas, which evolve from lower grade gliomas, and primary glioblastomas, which arise de novo.
  • It is recognized that multiple etiologies can underlie cancer development in a particular organ system. Even with similar histopathology, the genomics of each subtype may be different. For example, many hepatocellular carcinomas are associated with viral hepatitides, while others are associated with alcohol-related cirrhosis. It would be reasonable to consider these as distinct entities for ICGC projects. Furthermore, selection and comparison of cancer cases with different prevalences in different geographical regions may be important as these may also reflect distinct underlying etiologies. Examples of this form of classification might include cancers of the oral cavity, gallbladder and biliary tract from high-risk areas and gastric, colon, rectal, lung and nasopharyngeal cancers from Asia.
  • Tumors of hematological and lymphoid tissues present particular issues. A large number of well established molecular and/or morphological subtypes have been defined, and it is not feasible to conduct full-scale projects on each one. In this context it may be appropriate to select tumor types based on their clinical impact and a relative paucity of information concerning their genetic basis. Examples of such selections might include follicular lymphoma and diffuse large-cell B cell lymphoma, myeloma, myeloproliferative disorders (which represent early stages of neoplastic change), chronic lymphocytic leukaemia, acute myeloid leukaemia with normal cytogenetics, T-ALL and B-ALL.