Cancer - Disunfied Kind, Unified Explanation Part 4

Limits of Mechanistic Decomposition ¨ ¨ ¨ Cells divide or fail to divide in response to intra- and intercellular signaling pathways, pathways that extend far beyond the cell. Mechanistic explanations of subcellular pathways in cancer thus have this core limitation: they are always characterizations of only one pragmatically bounded subset of a wider system of regulatory controls on cell division. Evolution has designed this redundancy exactly because break down in one element will call upon regulatory controls in other signaling pathways. Robustness ¨ ¨ What cancer illuminates is how systems subject to multiple sources of breakdown find stable solutions, or how robustness evolves. Indeed, cancer itself can be viewed as the evolution of a robust system: a network with its own cooperative functional organization, that continues to perform its function despite variations in environment and organization… Conservation & Evolvability There’s a tension in evolution, however, between conservation of entrenched structures (robust structures), and quasiindependence of parts. ¨ Wimsatt and Schank (drawing upon H. Simon, 1967) say this leads to the “GE paradox” ¨ The GE Paradox ¨ ¨ The paradox is that there are fitness advantages due to entrenchment, and fitness advantages to modularity, but the two cannot be achieved simultaneously: “Modularity, duplication, and functional redundancy should each decrease entrenchment by reducing interdependence between or dependency upon specific system components.” Modularity Entrenchment Victims of our own success? ¨ ¨ Arguably, cancer is in some ways a product AND byproduct of both entrenched elements (or, scale free networks; like the functional centrality of p53), AND, modularity: the capacity for parts to vary independently without affecting the whole: two features that contribute to robustness. Modularity makes it possible for a tumor (or, its components – subpopulations of cells) to develop gradually in a specific organ without necessarily affecting the the organism as a whole, which is why many tumors are not discovered until it is too late. I.e., cancer coopts the very same features (modularity, plasticity, robustness) that makes us so highly evolvable as a lineage. We are a victims of our own success. Interlevel integration? No, here’s why: ¨ ¨ Failure of the same mechanism can occur in different ways, or in the same way, with different effects in different cancers. In fact, this is what cancer researchers have come to expect. I.e., it’s mistaken to think that integration or explanation and thus subtype classification in complex systems is “complete” when we identify ‘the’ mechanism, because (a) there is not one, and (b) the same “mechanism” (or failure of mechanism) will have different effects across different genetic and epigenetic backgrounds. Conclusions ¨ ¨ ¨ ¨ ¨ There can be no classification of cancer that hierarchically subsumes each subtype in light of genomic features alone. Nonetheless, cancer science is predictive and explanatory. Success is piecemeal: due to diverse predictive and explanatory goals, and the massive heterogeneity & complexity of cancer, as well as its evolutionary dynamics. There are general theories (with wide scope but little precision or predictive power), and local integrative explanations. Integration can come in different varieties. Scope trades off with predictive power. Interlevel “mechanistic” explanation is of little relevance in complex systems with robustness & evolutionary dynamics. Extra slides Cancer a unified kind? ¨ There is no essence shared all and only with cancer, nor a single causal basis. Nonetheless, there are common features: ¤ incidence (by and large) rises with age ¤ more common in tissues with high rates of turnover of cells ¤ involves differentiation (arrest in or dedifferentiation) ¤ involves incapacity to regulate cellular growth ¤ all and only in metazoans (organisms with a mesoderm and the capacity to EMT (i.e., byproduct of modularity, plasticity) ¤ evolutionary process ¤ cancer cells have “typical” capacities: resistance of apoptosis, loss of control of cell division, acquisition of a blood supply, many associated with mutations affecting the cell cycle. Kinds of kinds ¨ ¨ ¨ ¨ Essentialism: members of a natural kind share a real essence Causal theories: natural kinds are groups of individuals with shared properties that cluster in virtue of a common cause or mutually stabilizing “homeostatic mechanisms” (i.e., 'homeostatic property clusters’) Functional kinds: kinds whose characteristic similarities must consist only in the activities, processes, or roles that category members perform. Stable Property Cluster kinds: when a cluster of properties is “sufficiently stably coinstantiated to accommodate the inferential and explanatory uses to which particular sciences put such categories.” Lange, the “end of disease”? (2007) ¨ ¨ Disease as “incapacity” to regulate some function. [Cancer]…is the incapacity to regulate the growth and reproduction of certain cells. But this incapacity, though unifying all cancer, is rather vague because the proper regulation of cell growth and reproduction is achieved by a combination of many separate capacities, each of which must be disabled in some (perhaps independent) way for a tumor to form and to metastasize… Although they might all be lumped together as constituting a single incapacity (to regulate cell growth and reproduction), this incapacity is not especially unitary… It might better be described as a conjunction of various incapacities… (Lange, 2007, p. 284) “(Dys)functional” kind? ¨ ¨ ¨ ¨ But, (a) other diseases involve dysregulation of cellular growth… And, (b) what system does cancer compromise the function of…? A single cell, a population of cells, an organ, or the body as a whole? Where we target our analysis of function will shift as we move from asymptomatic cancers to more advanced cases. Generative Entrenchment ¨ Wimsatt and Schank (2004): “a dynamic structural feature” of certain kinds of “generative structures.” The generative structure of a system: ¤ …has a characteristic set of causal interactions which could be variously represented. One of the simplest representations is a directed graph, where nodes are parts, processes, or events, and arrows are consequences of the presence or operation of nodes on other nodes. For each node, consider how many other nodes can be reached from it by following the arrows. This indicates how much of the phenotype is downstream of, causally dependent upon, or affected by a given node. We define the generative entrenchment (GE) of a node as the magnitude of its downstream dependency… ¨ ¤ Lazebnik (2010) noted: benign tumors share 5 of the 6 properties: sustained angiogenesis, evasion of cell death and insensitivity to antigrowth signals, limitless replicative potential, self-sufficiency in growth signalling… ONLY tissue invasion and metastasis distinguishes cancerous tumors… “after producing nearly two million papers on cancer, we are yet to learn the underlying mechanisms sufficiently well to have a sizable impact on cancer mortality. Otherwise… with the war on cancer about to enter its fifth decade, we might keep winning the battles, but losing the war.” (Lazebnik, 2010) Patchwork taxonomy? ¨ “The image of the scientific enterprise that emerges from these considerations is of a patchwork of overlapping theories or taxonomies, some of which inquire into different spatiotemporal levels, whereas others inquire into the same level relative to different interests. It is futile to look for a single system of categories that uniquely classifies a certain set of phenomena. For any given spatiotemporal level, there does not seem to be a limit on the number of taxonomic systems that may categorize it, so uniqueness is not a promising construal of what makes a system of categories “natural.” …” (Khalidi, 1998) HPC Kinds? Shifts the burden… ¨ ¨ ¨ “one can be led to lump or split the same putative kind in different ways depending on which mechanism one consults in accommodating the taxonomy to the mechanistic structure of the world” (Craver, 2009, 583). But… “human perspectives and conventions enter into judgments about how mechanisms should be typed and individuated” (591), It follows that what natural kinds there are on the HPC view will depend on those perspectives and conventions… what many will regard as an unacceptably conventionalist pluralism about what kinds of cancer kinds there are.