Cancer Biology and Implications for the Perioperative Period

Introduction

In spite of the evident cytoreductive, potentially curative advantages of surgery, there has long been a suspicion that resecting primary tumors carries an intrinsic, paradoxical risk with respect to disease progression. ^, Anecdotal evidence prompted surgeons in the second half of the 19th century to draw comparisons between operative dissemination of cancer and of tuberculosis, seeking to understand how an apparently benign neoplasm might become malignant and disseminate so “astonishingly swiftly” after surgery. Like tuberculosis, the phenomenon was assumed to have a mechanical etiology, resulting from “notoriously forceful” manipulations of the tumor, as it was examined and dissected out. However, analysis of autopsy data from 735 women with breast cancer soon revealed that other factors were also in play, since the distribution of distant growths could not be explained simply by chance or anatomical drainage patterns. Tumors appeared to show a predisposition to metastasize to certain organs over others, leading Stephen Paget to postulate in 1889 that metastasis involves favorable interactions between disseminating tumor cells and the sites to which they colonize. This enduring “seed and soil” hypothesis continues to form the framework for exploring the biology of metastasis today and seems to be a particularly suitable metaphor for describing the risks of surgical dissemination and postoperative disease progression.

Except for some notable examples, much of our current insight into cancer pathogenesis has been shaped over the past 40 years–a revolutionary period propelled by remarkable developments in experimental tools and technology and a global will to invest in tackling this ever-more prevalent disease. Since the milestone discovery by Varmus and Bishop in 1976 that so-called “proto-oncogenes” within the normal cell genome have the capacity, when corrupted, to trigger the transformation of a healthy cell into the beginnings of a tumor, a wealth of knowledge and detail has amassed regarding the molecular mechanisms that underpin the origins of cancer, drive its progression, and determine its response to therapy. This has enabled pioneering advances in diagnostics and targeted therapeutics that have translated into a number of clinical successes, where certain cancers are now considered largely curable and others now carry survival rates measured in years rather than months. Yet, the persistence of high mortality across many types of cancer clearly indicates that major challenges remain, particularly in terms of overcoming metastatic disease and therapy resistance.

Reductionism prevails in cancer research as a logical, pragmatic approach to managing its complexity, yet the pitfalls encountered over recent decades have reinforced the need to continually reevaluate the way in which we think about cancer. In the years following the Varmus-Bishop discovery, it became widely held that cancer was a disease of identifiable genes and that a logical solution would be found in deciphering a set of genetic rules common to all mammalian cells undergoing neoplastic transformation. Yet, as the inventory of recognized oncogenes and tumor suppressor genes grew longer, it became clear that tumors follow variable and unpredictable genetic paths, even within the same tissues of origin. ^, Contemporary paradigms portray a more nuanced, less tumor-centric perspective of disease progression, where cancer genetics only partially determine the clinical course. Accordingly, cancers are no longer viewed as insular masses of genetically aberrant, incessantly proliferating cells, rather as a diverse catalog of diseases whose individual characteristics and clinical course are influenced by heterotypic and dynamic interactions among mutant genes, microenvironmental landscapes, systemic physiology, and host defenses.

With such changes in perspective, one might reason that the “cut, burn, and poison” approach that has formed the cornerstone of cancer treatment for much of the last century is outdated; after all, criticism is often leveled at such techniques for the indiscriminate way in which they are deployed against cancers of very different molecular background. However, while there have been notable and promising developments to improve clinical outcomes by moving toward so-called “personalized medicine”—where precision therapies are tailored to a tumor’s individual molecular complexion and vulnerabilities, rather than simply its tissue of origin and broad histological subtype—the promise of many targeted biological therapies to drastically reform long-term survival outcomes has yet to be realized. The inconvenient reality of cancer heterogeneity, even within a single tumor, challenges the notion that drugs with narrow molecular targets may achieve lasting efficacy, particularly in advanced disease, while the associated cost burden to health economies is enormous. Surgery, radiotherapy, and cytotoxic chemotherapy therefore retain their place as essential, often highly effective tools in modern cancer care and are likely to remain so for the foreseeable future. Consequently, alongside ongoing efforts to pioneer the next revolutions and drug discoveries in clinical oncology, there is also clear impetus to continue to deliver evolutionary improvements to current clinical practices, especially where these are a common or even ubiquitous component of disease management.

The attention focused toward inadvertent surgical cooperation in disease progression now extends to a range of factors beyond the physical effects of tumor handling. These include the activation of evolutionarily conserved responses to tissue trauma, such as sympathetic nervous system activation and inflammation—aggravated further by postoperative infective or wound-healing complications—as well as a postoperative period of impaired immunological competence, in which antitumor immunity may be temporarily compromised. ^, There are also concerns about the impact of perioperative pharmacology, most prominently centering around anesthetic and analgesic drugs and their purported influence over cancer cell biology and host immunity. ^, In light of an improved awareness of the cancer cell–extrinsic factors that contribute to cancer pathogenesis, and the manner in which they do so, it is increasingly conceivable to appreciate how the inflammatory, immunological, and metabolic state of the surgical patient might relate to and impact upon conditions so prominently associated with tumor evolution and metastasis ( Fig. 3.1 ).

Much of the detail concerning individual components of perioperative care, their potential interplay with cancer pathogenesis, and their potential role in influencing disease outcomes will be explored in the chapters that follow. The goal of this chapter is to introduce the common biological themes applicable to solid cancers and to construct a conceptual framework that begins to relate these themes to the potentially influential events taking place within the perioperative period.

Development of a Tumor

The origins of more than 200 types of human cancers are diverse, but fundamentally are thought to concern a series of genetic, environmental, and host interactions that drive healthy somatic cells through a multistep process toward a neoplastic state. The defining, transformative event involves the corruption and unfaithful propagation of a cell’s genetic code, with the chances of such events occurring now clearly understood to be influenced by a combination of hereditary and environmental factors. The principles of Darwinian natural selection govern the likelihood as to whether these mutations are carried forward through subsequent rounds of cell division, ^, with those attributable to phenotypes that confer some degree of survival, functional, or proliferative advantage tending to drive the emergence of a predominant clone that may eventually manifest as a tumor. The great majority of human tumors are benign; it is the acquisition of invasive or disseminative capabilities that determines malignancy, and it is the metastases spawned by these tumors that are responsible for 90% of cancer-related deaths.

Cancer is a disease of clonal evolution ^, that explains both the process of carcinogenesis and the tendency for most advanced cancers to eventually acquire therapy resistance. For many years, the prevailing model of tumor development traced neoplasms to a single ancestral cell of origin, which acquired the necessary initiating genetic lesion(s) to transition from health to a cancer cell. In a linear fashion, its progeny would sequentially acquire and accumulate mutations that enabled an ever-more autonomous, inimical existence, culminating with the host-compromising capabilities of invasion, dissemination, and growth in distant sites. We now understand that most cancers exhibit a considerable degree of clonal and subclonal heterogeneity, comprising numerous genetically and phenotypically distinct subpopulations of cells that both compete and cooperate with each other ( Fig. 3.2 ). These arise from heritable and stochastic genetic and epigenetic changes over time, driven locally by microenvironmental variation across the three-dimensional architecture of a developing tumor, and systemically by factors such as nutrition, hormones, infection, and environmental exposures. Furthermore, while metastasis has long been described in terms of a late event in the evolution of a primary tumor, there is increasing evidence that dissemination occurs early, in some cases even before the discernible manifestation of a primary tumor, potentially leading to the parallel progression of secondary growths at distant sites that are remarkably distinct from each other and the primary tumor ( Fig. 3.3 ).

These observations have transformed thinking in cancer biology in recent years, lifting the horizons of research beyond cancer cell-autonomous paradigms of oncogenes and tumor suppressors towards the dynamic forces at play within the intratumoral and organismal ecosystem. They also point to the troubling reality that many tumors will have seeded distant organs with thousands of cancer cells by the time of diagnosis, where disparate ecologies provide the pressures for further clonal diversification. Thus, while disease may appear clinically localized, the likely existence of invisible micrometastases means that the systemic impact of surgery and anesthesia following complete resection of the primary tumor should never be discounted.

Nature of a Tumor

The biological complexities of tumors have been rationalized by Hanahan and Weinberg’s widely acknowledged “Hallmarks of Cancer,” ^, which sets out the unifying themes and overarching phenotypic characteristics of human tumors, as they evolve from healthy, somatic cells to malignant neoplasms capable of unrestricted and potentially disseminated growth. In keeping with the reductionist nature of cancer research, these hallmarks are often studied and therapeutically targeted in relative isolation from one another, although their codependency and complementarity is essential to bear in mind.

Proliferative Signaling and Cell-Cycle Deregulation

Probably the most prominent feature of a cancer cell is its capability to sustain proliferation. Proliferation is normally tightly controlled by a concert of growth signaling molecules and checkpoints in order to maintain normal tissue function, architecture, and repair capabilities throughout the mammalian lifespan, but defects in one or more nodes of these cellular systems can lead to progressive deregulation and autonomy of cell-cycle progression. There are a number of ways in which cancer cells have been shown to exploit the enabling mechanisms of proliferation, including overexpression of cell surface receptor proteins to render cells hyperresponsive to relatively low ligand bioavailability and autocrine stimulation via the self-production of growth factor ligands, as well as by engaging in reciprocal signaling with resident and infiltrating cells of the local stroma. However, many tumors acquire growth factor independence through somatic mutations of key ligand receptors, enzymes, or transduction molecules comprised within mitogenic circuits ( Fig. 3.4 ). Such pathways exist to mediate signals from cell surface receptors to the nucleus in order to permit the interpretation of and response to specific extracellular cues, such as the presence of growth factors, cytokines, and microenvironmental stress; however, their corruption may lead to inappropriate, constitutive activation in the absence of such cues. For example, activating mutations that affect the structure of the B-Raf serine/threonine kinase—which stimulates the extracellular signal-regulated kinase (ERK)/mitogen-activated protein kinase (MAPK) cascade—are known to exist in approximately 8% of all human cancers, especially melanomas (50%). Mutations of Ras—a binary molecular switch that cycles between active guanosine triphosphate (GTP)-bound and inactive guanosine diphosphate (GDP)-bound states, also upstream of the ERK/MAPK cascade—underlie approximately 90% of pancreatic cancers and 50% of colon cancers. In such cases, Ras GTPase activity is compromised, leading to the impairment of an intrinsic negative feedback mechanism operating to ensure signal transmission is transient.

Phosphoinositide 3-kinase (PI3K) signaling is one of the most frequently dysregulated pathways in cancer and exerts influence on many hallmark phenotypes besides proliferation. Aside from responding to increased oncogenic signals upstream, it may be hyperactivated directly by malignant transformations. These include gain-of-function mutations to the PI3K catalytic subunit (most commonly the PIK3CA oncogene encoding p110α) that phosphorylates phosphatidylinositol 4,5-bisphosphate (PIP ₂ ) to generate phosphatidylinositol 3,4,5-trisphosphate (PIP ₃ ), as well as inactivating mutations or loss-of-heterozygosity in tumor suppressor genes such as PTEN . PTEN is a 3′ phosphatase that counteracts PI3K by dephosphorylating PIP ₃ back to PIP ₂ . The net result of an overabundance of PIP ₃ is the hyperactivation of multiple downstream effectors, including most notably, the pleiotropic serine/threonine kinase Akt (also known as protein kinase B). ^, Among the most highly conserved functions of Akt are its roles in promoting cell growth, via activation of mTOR complex 1 (mTORC1), and in supporting cell proliferation by the complementary phosphorylation of GSK3, TSC2, and PRAS40, which drive cell-cycle entry and progression, and the inactivation of the p27 ^Kip1 and p21 ^Cip1/WAF1 cyclin-dependent kinase inhibitors. Reflecting the advantageous nature of amplified PI3K-Akt signaling in tumorigenesis, a recent meta-analysis of cancer genome data from nearly 5000 tumor samples revealed that PIK3CA and PTEN represent the second- and third-most frequently mutated genes in human cancer.

The most commonly mutated gene, accounting for approximately 50% of all human cancers, is the tumor suppressor, p53, ^, which illustrates the gain that also comes from circumventing antiproliferative safeguards. Accordingly, TP53 is the most studied human gene in history and is frequently referred to as the “guardian of the genome” in recognition of its central role in DNA damage response. In response to cellular insults and abnormalities, including genotoxic, metabolic, and replication stress, the stabilization and subsequent activity of this DNA-binding protein can arrest cell-cycle progression, instigate a raft of reparative and adaptive pathways, and govern cell-fate decisions such as apoptosis and senescence, with the overriding purpose of conserving genomic integrity. Consistent with mediating its tumor-suppressive function through a transcriptional mechanism, the vast majority of cancer-associated TP53 mutations occur in its DNA-binding domain. ^, Its role in cancer biology is appearing increasingly context-dependent, but in elementary terms, loss of p53 both lifts a major restriction on cell proliferation and promotes genome instability and phenotypic evolution by permitting the accumulation of oncogenic mutations through successive rounds of cell division.

Evading Cell Death

In addition to directing cell proliferation, both PI3K signaling and p53 (among many other players) are critically involved in determining cell survival. After sensing overwhelming stress or irreparable DNA damage, p53 transcriptionally activates a group of BCL-2 family proteins, including BAX, NOXA, and PUMA, which initiate apoptosis —an orderly cascade beginning with mitochondrial outer membrane permeabilization (MOMP) and culminating in the proteolysis and self-destruction of a cell. Evidently, the acquisition of a loss-of-function mutation to p53 constitutes a major mechanistic opportunity for a renegade cell to evade death, as it navigates the numerous physiological stresses associated with hyperproliferation, tumorigenesis, or anticancer therapy. In tumors with functional p53, similar ends may also be achieved by alternative means to inhibit the activity of proapoptotic proteins or by overexpressing the counterbalancing, antiapoptotic members of the BCL-2 family, including BCL-2 itself, as well as BCL-X _L and MCL-1. For instance, Akt can directly phosphorylate the proapoptotic BH3-only protein BAD, thereby sequestering it from its target in the mitochondrial outer membrane and preventing its action in MOMP. Akt also phosphorylates the Forkhead Box O (FOXO) transcription factors, leading to their displacement from the nucleus and suppression of FOXO target expression, including proapoptotic molecules such as BIM, PUMA, and Fas ligand (FasL). A third Akt-mediated survival mechanism returns us once more to p53, as Akt phosphorylates and promotes the nuclear translocation of MDM2—an E3 ubiquitin ligase and the main negative regulator of p53. ^,

The Tumor Microenvironment

Studying the genetic basis of cancer growth reveals the impact of chronically dysregulated signaling in mediating cancer cell phenotypes. Perhaps of greater interest to the perioperative physician is to know how, and to what consequence, these same oncogenic signaling pathways might be acutely perturbed by extrinsic events. After all, little can be done about tumor genotype by the time of presentation, whereas there may be an opportunity to modulate certain external influences that commonly cooperate in disease progression. Over the past decade, an abundance of work examining the conditions and other constituents of the tumor microenvironment (TME) has demonstrated the impact of such extrinsic factors on the growth and malignant potential of cancer cells, many of which can be closely compared with conditions and physiological processes engaged during the perioperative period.

In contrast to previous tumor-centric perspectives of cancer pathogenesis, we now know that an array of other specialized, nonmalignant cells—which comprise more than 50% of the mass of tumors—actively collaborate in promoting tumor growth and malignant progression ( Fig. 3.5 ). These include fibroblasts and myofibroblasts, immune inflammatory cells, mesenchymal stem and progenitor cells, and vascular endothelial cells and pericytes. They are recruited and activated by danger-associated molecular patterns (DAMPs), cytokines, chemokines, and angiogenic factors emanating from emerging neoplasias, in much the same fashion as an acute inflammatory response to tissue trauma or infectious pathogens. Over time, transformed and nontransformed cells act in concert to form an increasingly reactive stroma, rich in soluble growth and inflammatory mediators, which sustains the inflammatory response, dynamically remodels the extracellular matrix, and reciprocally reinforces the malignant behavior of cancer cells.

Catecholamines emanating from the systemic circulation and infiltrating sympathetic neural fibers modulate transformed and nontransformed cells alike through activation of beta-adrenergic signaling. Adrenoceptors are often abundantly expressed at the sites of primary and metastatic tumors. When stimulated by adrenaline or noradrenaline, the resulting cyclic AMP (cAMP) flux upstream of protein kinase A (PKA) and ERK/MAPK signaling pathways leads to an array of phenotypic responses in cancer cells and supporting stromal cells. ^, In addition to the aforementioned effects of MAPK signaling on cell growth and proliferation, these responses include cell metabolic, morphology, and motility changes, as well as inducing and sustaining angiogenesis, differentiation, and inflammation, which collectively remodel the TME. ^, Accordingly, in vivo cancer models show heightened sympathetic nervous system and beta-adrenergic signaling activation to accelerate tumor growth and metastasis, while those models and epidemiological data from cancer patients point to the therapeutic potential of beta blockers.

The biology of the TME is also substantially determined by its changing physical and chemical characteristics. Reflecting the increased metabolic activities of both tumor and stromal cells in conjunction with aberrant and underdeveloped vascularity, developing tumors are typified by hypoxia, acidosis, an accumulation of catabolic metabolites, and raised interstitial pressure. The extracellular matrix is also stiff and fibrotic in view of the increased collagen deposition by resident and recruited fibroblasts and frequent remodeling, ^, further exacerbating problems associated with hypoperfusion. These conditions invariably place substantial stresses upon all cells, invoking prosurvival adaptions in malignant cells and a state of exhaustion in tumoricidal immune cells, all the while perpetuating reactive changes in the stroma. In a perioperative context, it is worth considering how surgical disruption of local vasculature and lymphatics and the accompanying tissue edema, inflammation, and hypoxia may contribute to a wound microenvironment very similar to that of a growing tumor. Such stimulating conditions might support a residual fraction of cancer cells to reestablish a tumor or to disseminate, which could partially explain why surgical wound and inflammatory complications are associated with an increased risk of cancer recurrence. ^,

Tumor Immune Landscape

Infiltrating immune cells act in conflicting ways depending on tissue context or the balance of stimuli. ^, Many early tumors are rejected or kept in check by cell-mediated immune responses that sense altered-self cells and effect their elimination. In particular, the activities of natural killer (NK) cells, dendritic cells, M1-polarized macrophages, and cytotoxic CD8 ⁺ T cells are strongly associated with a favorable prognosis. CD8 ⁺ T cells are supported by CD4 ⁺ T helper 1 (T _H 1) cells that produce interleukin (IL)-2 and interferon gamma (IFN-γ); an abundance of these is also associated with a good prognosis. Conversely, as chronic inflammation ensues, a number of immune cells may be coopted to promote tumor progression. These include subpopulations of T cells, including CD4 ⁺ T helper cells with a T _H 2 orientation, as well as neutrophils, mast cells, M2-polarized macrophages, and myeloid progenitors. In keeping with their usual function in clearing and remodeling wounds, these latter types of immune cell contribute substantially to pools of growth-promoting and angiogenic factors such as EGF, FGF, and VEGF; invasion-enabling matrix remodeling enzymes such as matrix metalloproteinase (MMP)-9 and heparinase; and an abundance of cytokines and chemokines that amplify and sustain the inflammatory state through paracrine feedback loops with malignant cells and other infiltrates. Collectively, these have been shown to elicit and sustain multiple traits of high-grade malignancy and are associated with a poor prognosis. ^,

Cancer cells also exploit mechanisms to actively evade cell-mediated immunity, some of which may be driven by specific oncogenes such as BRAF and STAT3, but otherwise often arise through crosstalk with other stromal constituents. For example, production of the eicosanoid prostaglandin E ₂ impairs NK cell viability and cytolytic activity and subverts NK cell-mediated recruitment of dendritic cells into the TME resulting in immune escape ^, ; this is in keeping with data showing correlations between poor NK cell abundance and function with local recurrence, metastasis, and reduced overall survival. ^, ^, Similarly, expression of transforming growth factor (TGF)-β by both tumor cells and stromal cells promotes T-cell exclusion from tumors, inhibits the acquisition of an antitumor T _H 1 phenotype, and stimulates the differentiation of immunosuppressive regulatory T (T _reg ) cells, which at least partially explains why patients with elevated TGF-β levels tend to mount poor antitumor immune responses and derive less therapeutic benefit from immunotherapy. T _reg cells crucially operate to maintain self-tolerance and immune homeostasis by competing for IL-2, secreting immunosuppressive cytokines (such as IL-10 and TGF-β), and suppressing antigen-presenting cell (APC) function by expressing cytotoxic T lymphocyte antigen 4 (CTLA-4), but their contribution in an oncological context only serves to antagonize effector lymphocytes, undermine host defense, and worsen clinical outcomes. ^,

Angiogenesis

As with any other tissue, tumors need to develop a vascular infrastructure to maintain delivery of sufficient oxygen and nutrients and to drain metabolic by-products. The sprouting and assembly of new blood vessels from the preexisting vascular network is usually a transient homeostatic process in health, but is a consistent and usually sustained feature of growing tumors that has been described in terms of tripping an “angiogenic switch.” Often evident in histological slices of premalignant lesions such as dysplasia, it is also an early feature of tumor development, becoming more robust and penetrating as the growing tumor develops a fuller stromal compartment and strives to satisfy its metabolic demands. Yet tumor-associated vasculature is typically structurally aberrant and dysfunctional, reflecting the excessive and disorderly proangiogenic signaling driving the process. It is characterized by excessive branching, fenestrations, blind ends, and discontinuous cellular architecture that contribute to a state of turbulent flow, impaired perfusion, excessive vascular permeability, and acute fluxes in oxygen tension and substrate availability ( Fig. 3.6 ).

Much attention has been directed toward the prototypical angiogenic factor vascular endothelial growth factor-A (VEGF-A) and the receptor tyrosine kinases that mediate its canonical effects on endothelial cell proliferation, migration, survival, and permeability. VEGF-A is secreted by most cancer cells, as well as surrounding stromal cells, while the degree of expression correlates with invasiveness, metastasis, recurrence, and prognosis. As a transcriptional target of hypoxia inducible factor (HIF)-1, its expression is prominently influenced by tumor hypoxia. Similarly, oncogenic mutations that directly enhance HIF stability (such as VHL mutations commonly seen in clear cell renal carcinomas) or signaling pathway activity upstream of HIF-1, including PI3K/Akt signaling, also drive angiogenesis through VEGF-A. ^, However, its expression—along with other proangiogenic factors such as platelet-derived growth factor (PDGF) and fibroblast growth factor (FGF)—may also be induced and reciprocally reinforced in the stromal compartment by activated platelets, immune inflammatory cells, and fibroblasts, highlighting the importance of extrinsic regulation of angiogenesis by the TME.

Reprogramming Cellular Energy Metabolism

The rapid and sustained proliferative activities of a growing tumor must be accompanied by substantial adaptations to cellular metabolism, especially within a context of limited oxygen and nutrient bioavailability. Metabolic “reprogramming,” a term widely used in recent cancer metabolism literature, describes the upregulation or suppression of conventional metabolic pathways undertaken to satisfy the heightened energy and biosynthetic requirements of tumors, improve cellular fitness, and support cell survival under stressful and nutrient-deprived conditions. ^,

The first observations on this were made by Otto Warburg, long before the discovery of oncogenes and tumor suppressors. Enhanced glycolysis is a normal physiological response to hypoxia, but Warburg noticed that cancer cells preferentially employ glucose metabolism over oxygen-consuming mitochondrial metabolism, leading to the production of substantial quantities of lactate regardless of oxygen availability (a phenomenon now known as the Warburg effect). Indeed, a marked increase in glucose uptake and utilization is a characteristic of many tumors, and it is this very phenotype that has been exploited in diagnostics for many years through use of a radiolabeled glucose analog ( ¹⁸ F-fluorodeoxyglucose, FDG) in positron emission tomography (PET). From a cellular energetics perspective, this state of “aerobic glycolysis” is ostensibly counterintuitive, since glycolysis is 18 times less efficient in yielding ATP per mole of glucose than oxidative phosphorylation. This led Warburg to reason that perhaps it was not adopted out of choice, but rather out of necessity to compensate for defects in the mitochondrial respiratory apparatus. However, while certainly true in some instances, numerous observations have since demonstrated that defective respiration is not a ubiquitous feature of malignant cells and that many cancers exhibit the Warburg effect while retaining functional mitochondrial respiration. Moreover, enhanced expression of glucose transporters and rate-limiting glycolytic enzymes have now been extensively associated with activation of oncogenes such as KRAS and MYC, loss of tumor suppressors like p53, or malignant cooption of physiological PI3K/Akt/mTOR or HIF signaling, suggesting that such a metabolic phenotype is actively selected for its capabilities in supporting tumorigenesis. ^, Indeed, we now understand that an overriding advantage of upregulating glycolysis in cancer is to enhance the flux of glycolytic intermediates through subsidiary pathways involved in macromolecular biosynthesis and redox homeostasis ( Fig. 3.7 A).