Advancements in phosphodiesterase (PDE)-targeted therapies have shown promise in recent years for treating patients with a variety of autoimmune

diseases. This review summarizes the development of PDE4 inhibitors and the associated literature with a focus on treatments for autoimmune diseases. After the initial investigations of the prototypic PDE inhibitor, rolipram, more selective inhibitors targeting the PDE4 isozyme have been developed. With phase II and phase III clinical trials currently underway to evaluate the safety and efficacy of the latest generation of PDE4 inhibitors, namely apremilast, a new class of treatments may be around the corner for patients suffering from chronic, autoimmune diseases.

Keywords:
Autoimmune; Apremilast; Crohns; Dermatitis; PDE4; Psoriasis; SLE

Introduction
Our earliest understanding of phosphodiesterase (PDE) inhibitors began with a series of publications by Sutherland and Rall in the 1950s, describing the properties of cyclic adenosine monophosphate (cAMP). Various cellular pathways and inflammatory responses are mediated by cAMP, an essential intracellular second messenger made up of phosphodiester bonds. Evidence showed formation of cAMP was induced by substances such as epinephrine and glucagon, and the suppression of the enzymes hydrolyzing cAMP, including PDEs, by sodium fluoride and caffeine[1,2]. By the 1960s, the role of cyclic nucleotide second messengers, such as cAMP, in cell signaling and homeostasis was established, and regulation of this pathway by PDE inhibitors arose as a field of considerable interest. However, the immunomodulatory properties of cAMP and the anti-inflammatory potential of PDE inhibitors were not demonstrated until the early 1970s [3-5]. Additional research would later demonstrate the expression of the PDE isoenzyme PDE4 almost exclusively within inflammatory cells [6]. Inhibition of PDE4 leads to the reduction of intracellular cAMP levels, and decreases in T cell and monocyte derived cytokine and chemokines including tumor necrosis factor (TNF) [7-11]. Targeting PDE4 has enormous clinical potential because it targets a central pathogenic process that bypasses complex antigen receptor-specific immunoregulatory mechanisms. Indeed, selective PDE4 inhibitors have generated substantial interest as a treatment for several autoimmune conditions including ankylosing spondylitis, Alzheimers disease, psoriasis, psoriatic arthritis, sarcoidosis, systemic lupus erythematosus, inflammatory bowel disease, atopic dermatitis, rheumatoid arthritis, and multiple sclerosis.

Mechanism of action
PDEs are a family of enzymes responsible for the hydrolysis and subsequent inactivation of cyclic nucleotides, and have been organized into at least 11 families based on sequence homogeneity, inhibitor sensitivity, and biochemical properties [12]. Each enzyme within the PDE4 family specifically targets cAMP for degradation and consists of four subtypes (PDE4A to PDE4D). These enzymes are located within brain and immunocompetent cells such as neutrophils, T lymphocytes, macrophages and eosinophils [13]. PDE4 inhibition results in the accumulation of the intracellular second messenger cAMP, downstream activation of protein kinase A (PKA), and subsequent phosphorylation of the transcription factor cAMP-response element binding protein (CREB). Activation of this pathway modulates gene transcription of numerous cytokines, and results in suppression of TNF production and eventual inhibition of their proinflammatory and destructive properties [14].

Pharmacokinetics
The newest and most promising of the PDE4 inhibitors, apremilast, has been evaluated for its pharmacokinetic properties and disposition following oral administration. Multiple daily doses showed rapid absorption (T max=2 h) and a moderately long half-life (8.2 h) [15]. A separate study monitored healthy male subjects following a single, 20 mg, oral dose and found that apremilast was extensively metabolized via multiple pathways, with unchanged drug representing 45% of the circulating radioactivity and <7% of the excreted radioactivity. Analysis of total radioactivity suggests rapid absorption, with plasma T max values also at 2 h. Mean Cmax and area under the curve (AUC) values in plasma were 333 ng/ml and 1,970 ng*h/ml, respectively. Metabolic clearance of apremilast was the major route of elimination with the key metabolites demonstrating at least 50-fold less pharmacologic activity than apremilast [16]. Man et al. optimized the structures of a series of 3-(1,3-dioxo-1,3-dihydroisoindol-2-yl)-3-(3,4-dialkoxyphenyl) propionic acid analogues to enhance PDE4 and TNF inhibitory activity. So far, oral and intravenous administration of these analogues in female rats has shown good pharmacokinetics with low clearance, a moderate volume of distribution, and a 64% oral bioavailability [17].

Adverse effects
PDE4 is also one of the major phosphodiesterase isoenzymes expressed in the central nervous system, and therefore nausea and emesis are common adverse effects of drug administration[18]. Early PDE4 inhibitors actually failed in clinical trials due to the high prevalence of nausea and emesis [19]. Other adverse effects associated with repeated administration of PDE4 inhibitors include headache, diarrhea, fatigue, dyspepsia, nasopharyngitis, and gastroenteritis [20]. Mesenteric vasculitis is a more worrisome toxicity

that may be associated with the PDE4 inhibitors. Studies performed in rodents have demonstrated medial necrosis of the mesenteric arteries after administration of the second generation PDE4 inhibitor cilomilast. However at a meeting convened by the US Food and Drug Administration (FDA) in 2003 to discuss cilomilast in phase III studies, the committee unanimously agreed that the risk of mesenteric vasculitis is not a safety concern based on human studies [21]. The newer PDE4 inhibitor, apremilast, has been well tolerated with few side effects in phase I and II studies. Phase III clinical trials are currently underway and will provide more insight into its dosing and side effect profile. The most frequently reported adverse events have been headache, nausea and pharyngitis (15,34). Researchers used a recognized pharmacophore from the PDE4 inhibitors rolipram and roflumilast in the development of apremilast, and added it to a series of thalidomide analogs in efforts to optimize activity and reduce side effects classically seen with earlier PDE4 inhibitors [22].

Rolipram
The discovery of the anti-inflammatory actions of PDE4 inhibitors arose from early studies with the prototypic PDE4 inhibitor, rolipram. This was the first selective PDE4 inhibitor investigated, and has been used numerous times for drug comparator studies [23]. Rolipram was also studied as an antidepressant several years before the discovery of its potent PDE4 inhibitory activity [24,25]. Despite its potent anti-inflammatory effects in vitro, clinical trials were associated with unacceptably high rates of adverse events, specifically nausea and vomiting [26].

Roflumilast
Roflumilast was the first, and currently only drug in the PDE4 inhibitor class to be FDA approved. In several countries, this highly selective PDE4 inhibitor is licensed for oral, once-daily treatment of severe chronic obstructive pulmonary disease (COPD). In a pooled analysis of over 6,000 patients receiving roflumilast, higher rates of diarrhea, weight loss, nausea, headache, back pain, insomnia, decreased appetite, and dizziness were reported as compared to those receiving placebo. However the overall rate of adverse events was comparable to that among patients receiving placebo [27]. Pooled results from the pivotal COPD studies M2-124 and M2-125 showed that the weight loss for those in the roflumilast group was generally small (<3% of baseline weight) and typically occurred in the first 6 months of treatment. By the end of the 6 months, almost two-thirds of the weight loss could be attributed to a reduction in fat mass [28]. Prior investigations have revealed a link between PDE4 and lipolysis, possibly through regulation of cAMP pools in human adipocytes and increased plasma glucagon-like peptide 1 (GLP-1) concentrations in rats [29,30].

Flavonoids
Many flavonoids have been reported to inhibit PDE4 [31,32], and also demonstrate additional anti-inflammatory effects through other pathways. For example, in addition to PDE4 inhibition, the flavonoid dioclein suppresses the production of the inflammatory mediators interleukin (IL)-6, TNF, a chemokine ligand (CXCL)1/KC, CCL2/JE (monocyte chemotactic protein 1) and nitric oxide (NO), and acts as a scavenger of reactive oxygen species [33]. Studies on dioclein have also demonstrated a synergistic antiinflammatory effect by targeting multiple pathways [34]. However in contrast to rolipram, dioclein does not selectively inhibit PDE4 with additional activity against PDE1, which may lead to unwanted side effects [35].

Psoriasis
Psoriasis is a chronic inflammatory autoimmune disorder characterized by inflammatory cell infiltration into the dermis and epidermis accompanied by keratinocyte hyperproliferation [36]. Current treatments, including biological therapies, downregulate cytokine cascades and chemokine production. While these interventions can be highly efficacious, limitations include side effects, intravenous or subcutaneous administration, quality control, and production costs. AN-2728 is a topically administered, boron-containing compound developed for the treatment of psoriasis. This compound was found to reduce cytokine production, such as TNF and interferon (IFN)y, and demonstrated activity against the PDE4 enzyme [37]. Several clinical trials of AN-2728 have reported significant effects on markers of efficacy, such as TNF, in addition to being well tolerated [38]. The determination of cytokines in skin homogenates revealed that both T helper (Th)1 as well as Th2 cytokines are suppressed by PDE4 inhibitors, further proving its utility in the treatment of T cell mediated diseases, such as psoriasis [39]. An open-label, single-arm pilot study investigated the biological and clinical effects of oral apremilast 20 mg once daily in patients with severe plaque-type psoriasis. Of the 19 patients enrolled, 17 completed the study. Of the 19 subjects, 14 (73.7%) showed improvement in their Psoriasis Area and Severity Index (PASI) scores following a 29-day treatment phase. Among these responders, T cells were reduced by 28.8% and 42.6% in the dermis and epidermis, respectively. Epidermal thickness was also reduced by a mean of 20.5% from baseline [15]. More recently, a phase IIb, randomized, multicenter, placebo-controlled, dose-ranging trial for the treatment of plaque-type psoriasis with oral apremilast was completed. Patients were randomly assigned to apremilast 10 mg twice daily, apremilast 20 mg twice daily, apremilast 30 mg twice daily, or placebo. By week 16, a 75% improvement in PASI scores (PASI75) was achieved in 6% (5/88) assigned placebo, 11% (10/89) assigned apremilast 10 mg, 29% (25/87) assigned 20 mg, and 41% (36/88) assigned 30 mg. Significant differences from placebo were seen with apremilast 20 mg and 30 mg (P <0.0001), but not 10 mg. Reported adverse

events were most frequently mild to moderate and included nausea, upper respiratory tract infection, diarrhea, nasopharyngitis, headache, gastroenteritis, and dyspepsia. Of the eight serious adverse events, none were judged to be related to apremilast [40]. Currently there are two phase III, double-blind, placebo-controlled, multicenter trials (ESTEEM 1 (NCT01194219) and ESTEEM2 (NCT01232283)) investigating the use of oral apremilast 30 mg in adults with moderate to severe plaque psoriasis. These trials include a 52 week randomized, blinded, placebo-controlled phase, with primary endpoints measured at week 16, in addition to a 4year extension phase [40].

Psoriatic arthritis
A phase II, multicenter, randomized, double-blind, placebo-controlled study enrolled 168 subjects with psoriatic arthritis (PsA) during a 12-week treatment phase. Subjects were randomized to 20 mg apremilast twice daily, apremilast 40 mg once daily, or placebo. After completing the initial 12-week phase, subjects receiving placebo were given a 12-week course of apremilast. Following the treatment phase in both groups, subjects participated in a 4-week observation phase. The primary endpoint was the proportion of subjects achieving the American College of Rheumatology criteria for 20% improvement (ACR20) at week 12. In total, 44% of actively treated patients achieved the primary endpoint of ACR20 compared with 12% of the placebo cohort (P <0.001). The study revealed promising results for the treatment of PsA with oral apremilast, but was limited by the relatively short duration and unclear long-term efficacy and safety data. Additionally, the 90% of subjects enrolled were white and therefore the study may lack generalizability. Lastly, prior systemic therapy for PsA may alter the efficacy of apremilast and was not examined in this study. The most common adverse events (AEs) were diarrhea, nausea, headache, fatigue, and nasopharyngitis with 84.3% of subjects in the treatment phase reporting at least one AE. However most events were mild to moderate and no clinically relevant laboratory or electrocardiographic abnormalities were reported [41]. Results of this study are encouraging, and phase III clinical trials are currently underway. The efficacy and tolerability of apremilast in patients with psoriatic arthritis are now being studied in four independent phase III studies (PALACE 1 (NCT01172938), PALACE 2 (NCT01212757), PALACE 3 (NCT01212770), and PALACE 4 (NCT01307423)) [40]. These studies include both patients who have received disease-modifying antirheumatic drugs and those who have not.

Ankylosing spondylitis
Manifestations of ankylosing spondylitis (AS) include axial and peripheral skeletal inflammation, fat infiltration, and new bone formation. Therapeutic response centers on patient-reported outcomes such as pain, mobility and function as well as objective measures such as inflammation, and new bone formation that can be visualized by magnetic resonance imaging (MRI) and conventional radiography [42-44]. Moreover, the degree of clinical response with treatment may also correlate with fluctuations in biomarkers [45-47]. Recently, updated management guidelines published by the Assessment of SpondyloArthritis (ASAS) and the European League Against Rheumatism (EULAR) report that there is no evidence for the efficacy of disease-modifying antirheumatic drugs (DMARDs) such as methotrexate and sulfasalazine for the treatment of axial disease, leaving patients with limited treatment options. The two classes of drugs that have been shown to reduce the signs and symptoms of AS include non-steroidal anti-inflammatory drugs (NSAIDs) and TNF blockers [48,49]. Nevertheless, at the 2011 ACR meeting, results from a small pilot study were presented showing that apremilast may be efficacious in patients with longstanding AS. This double-blind, placebo controlled phase II unpowered pilot study included 36 subjects with longstanding AS who had not sufficiently responded to NSAIDs over 12 weeks. Of these subjects, 17 received apremilast 30 mg twice daily compared to 19 who received placebo. The apremilast group also saw a significant change from baseline (mean percentage) in levels of receptor activator of nuclear factor B (NFB) ligand (RANKL) and sclerostin [50].

Rheumatoid arthritis
Rheumatoid arthritis (RA) is another chronic, inflammatory autoimmune disease and primarily targets the synovial tissues of joints. Local production of cytokines and chemokines leads to leukocyte infiltration, and eventual erosion of cartilage and bone [51,52]. TNF has been shown to promote cytokine and chemokine production, as well as cellular activation and articular destruction in RA [53]. Given the pathophysiologic characteristics of RA, a study was performed to assess the anti-inflammatory effects of apremilast in human synovial cells collected from rheumatoid arthritis patients, as well as two murine models of arthritis. These synovial cells were cultured in the presence of increasing concentrations of apremilast for 48 h and enzyme-linked immunosorbent assay (ELISA) was used to analyze spontaneous TNF production. Results from this study showed that apremilast led to a dose-dependent inhibition of spontaneous production of TNF from human rheumatoid synovial membrane cultures. Additionally, both murine models showed a significant reduction in arthritis clinical scores over a ten-day treatment period with apremilast. Healthy joint architecture was also maintained in a dose-dependent manner. Unlike

the first generation PDE4 inhibitor rolipram, apremilast did not demonstrate any adverse effects in treatment-nave mice, possibly due to enhanced selectivity of apremilast [54]. An interim analysis of the data from a phase II pilot study investigating the use of apremilast in combination with methotrexate reported that the primary endpoint of ACR20 was not reached [55]. A placebo-controlled phase II trial using apremilast as monotherapy for RA is currently underway[56].

Systemic lupus erythematosus

Systemic lupus erythematosus (SLE) is a chronic autoimmune disease that can affect a variety of organs and is predominantly seen in women. Treatment is focused on controlling symptoms and often involves the use of corticosteroids and other systemic immunosuppressant therapies [57,58]. A recent study targeted increased PDE4 activity in lupus conditions using MRL/lpr mice (a mouse model developing severe lupus disease). Four groups of female MRL/lpr mice were injected at 5, 7, 9 and 13 weeks with one of ethanol, pentoxifylline, denbufylline, or NCS 613 (a novel PDE4 inhibitor). Results showed that both the survival time and the appearance of proteinuria of NCS 613-treated mice are significantly delayed, both with P values of 0.005 [59]. While study size was limited, the results demonstrate potential for the use of PDE4 inhibitors in patients with SLE.

Sarcoidosis
Previous reports have described the use of PDE4 inhibitors in patients with the systemic inflammatory disease, sarcoidosis. While pentoxifylline has been shown to be effective, use of this drug is limited by associated adverse events [60]. A small study evaluated the use of apremilast in 15 subjects who had failed systemic therapies for sarcoidosis. Patients received oral apremilast 20 mg twice daily, with a reduction to once daily dosing following the onset of any adverse event. Only two patients needed dosage reduction due to jitteriness in one patient and nausea in the other. No further adverse events were reported following this change in dose. Active lesions were assessed during 12 weeks of therapy with the Sarcoidosis Activity and Severity Index (SASI) as well as with photographs taken at baseline and at week 12. Photographs were presented in random order to three assessors and were scored from 1 to 5 (1 much better, 5 much worse). Results showed a significant reduction in SASI induration scores at weeks 4 and 12. The normalized mean score given by the assessors after 12 weeks of therapy was 2/5 (somewhat better after therapy), with good inter-reader consistency. Of note, three patients developed significant worsening of their cutaneous lesions within 3 months after discontinuation of apremilast[61].

Inflammatory bowel disease

The group of inflammatory conditions affecting the colon and small bowel known as inflammatory bowel disease can present in patients with life-altering symptoms lasting weeks to months at a time. Both Crohns disease and ulcerative colitis (UC) present with diarrhea, bleeding, fecal urgency and incontinence, abdominal pain, and fever caused by inflammation in the gut. The long-term therapeutic goal for these patients focuses on inducing and maintaining remission of symptoms to improve patient quality of life [62]. Existing anti-inflammatory agents such as 5-aminosalicyclates and other immunosuppressants have limitations due to adverse drug reactions, loss of therapeutic response, or lack of response in some patients [63]. Like many autoimmune diseases, the inflammation in inflammatory bowel disease (IBD) has been linked to upregulation of proinflammatory cytokines, such as TNF, and nuclear translocation of the proinflammatory transcription factor complex NFB [64]. TNF is thought to damage the gut via upregulation of matrix metalloproteinase (MMP) production by gut myofibroblasts, leading to breakdown of extracellular matrix, tissue damage, and ulcer formation [65]. The quantity of NFB activation has also been shown to correlate with the degree of mucosal inflammation and disease activity, as well as TNF upregulation. NFB activation works in a positive feedback to induce TNF, perpetuating further inflammation and disease processes [66]. Given the suppression of TNF and NFB associated with a variety of PDE4 inhibitors, two phase III clinical trials (FACT I and FACT II) were designed. These studies investigated the safety and efficacy of the PDE4 inhibitor, tetomilast, in treating moderately severe UC. Both studies were multicenter, randomized, double-blind, placebo-controlled, parallel-arm, dose comparison studies of tetomilast in subjects with active UC. Tetomilast was not found to have a significant effect on individual symptom or sigmoidoscopy scores. However, there was a trend for tetomilast to improve severity of rectal bleeding from baseline when compared to placebo ( P=0.017). By week 8, the efficacy scores had improved for those patients receiving tetomilast (with or without concomitant 5-aminosalicylic acid) compared to those on placebo, although these results were not significant statistically. One potential reason for the lack of difference in the therapeutic

response may be the very high placebo response rates seen in patients with IBD. Both 25 and 50 mg of tetomilast were generally well tolerated in subjects with active UC with no major adverse events [63].

Atopic dermatitis
Atopic dermatitis is a chronic inflammatory disease characterized by eczematous lesions and intense itching. Inflammatory infiltrates in these cutaneous lesions consist of T lymphocytes, neutrophils, eosinophils, monocytes, macrophages, and mast cells [67]. High levels of PDE4 activity are also found in leukocytes of these patients [68]. A study in Japan looked at the effects of PDE4 inhibitors, cilomilast, roflumilast, and rolipram on induced dermatitis in mice models. Cilomilast, roflumilast, and to a lesser extent rolipram, suppressed myeloperoxidase (MPO) activity, a quantitative index of neutrophils accumulating in skin associated with chronic inflammation. After 18 days of treatment, cilomilast and roflumilast showed a 47 and 36% recovery in skin severity score, respectively. This effect was more potent than the 25% recovery seen with cyclosporine A, especially in the earlier stages of treatment [69]. A subsequent study by Harada et al. used the PDE4B inhibitor, KF66490 to treat induced AD in mouse models. KF66490 significantly inhibited increases in ear thickness, IL-4 and IL-1B levels, and proliferation of fibroblasts and CD3-positive T cells. Compared to the first generation PDE4 inhibitor, rolipram, KF66490 also produced less potent emetic effects [70]. More recently, an open-label prospective trial of apremilast in 16 patients with moderate to severe AD was conducted to assess the safety, efficacy, and possible mechanism of action of apremilast in AD. One cohort consisted of six subjects treated with apremilast 20 mg twice daily for 3 months, while the second cohort consisted of ten subjects treated with apremilast 30 mg twice daily for 6 months. Participants in the study were required to remain on triamcinolone acetonide 0.1% for 2 weeks prior to the start of the study as well as throughout the trial. Nausea, the most common adverse event, was rated as mild and improved over the course of the study in all patients. After 3 months of treatment, a significant reduction of itch from baseline (VAS) and improvement in quality of life (assessed via Dermatology Life Quality Index (DLQI) score) was seen in cohort 1 ( P=0.02 and P=0.003, respectively), while Eczema Area and Severity Index (EASI) and quality of life (DLQI) scores improved in cohort 2 ( P=0.008 and P=0.01, respectively). At 6 months, statistically significant improvement was seen in all outcomes in cohort 2, including VAS ( P=0.03), DLQI (P=0.03), and EASI (P=0.002) [71].

Alzheimers disease
PDE4 inhibitors have also been investigated in the treatment of patients with Alzheimers disease. The neuropathological chan ges seen in Alzheimers are closely linked to chronic inflammation and apoptosis, with elevations of biomarkers seen even i n early stages of disease [72]. Accumulation of amyloid beta (A) peptides has been shown to produce inflammatory responses[73], activate the apoptotic pathway [74], inhibit hippocampal synaptic plasticity, and impair memory [75]. Similar responses were induced by infusion of aged Ab25-35 into the hippocampus, with reversal of memory deficits seen following repeated treatment with rolipram. These positive findings may be at least partially attributed to the blockade of inflammatory responses and apoptosis mediated by cAMP/CREB signaling. In fact, increased levels of pCREB were reported in the hippocampus following treatment with rolipram. These results may suggest a potential role for the use of PDE4 inhibitors in treatment of memory loss in patients with Alzheimers [75].

Multiple sclerosis
The demyelinating autoimmune disease, multiple sclerosis (MS), has often been studied in animal models by inducing experimental autoimmune encephalomyelitis (EAE) in genetically susceptible animals [76]. The EAE model also mimics the relapsing-remitting presentation of MS seen in humans. One study demonstrated a reduction in the clinical signs of EAE in mice models during the administration of rolipram. Improvement was seen during both the initial presentation of disease as well as subsequent relapses [77]. Matrix metalloproteinases (MMPs) are a gene family of zinc-dependent endopeptidases involved in the proteolytic modeling of the extracellular matrix as well as the pathogenesis of several autoimmune disorders of the peripheral and central nervous systems, such as MS [78]. In vitromodels have shown that rolipram inhibits NFB, a key regulator of inflammatory processes and gene expression related to EAE and MS, including MMP-9 [79]. When rats primed to EAE were treated with rolipram, the high levels of NFB activation in freshly obtained cells were prevented. Furthermore, inhibition was also seen after incubation of encephalitogenic cells with rolipram, indicating that interference with NFB activation is a direct effect of the drug. Inhibition of NFB was also accompanied by a decrease in MMP-9 gene expression [80].

Conclusions
Advancements in phosphodiesterase-targeted therapies have shown promise in recent years for treating patients with a variety of autoimmune diseases. After the initial investigations of the prototypic PDE inhibitor rolipram, more selective inhibitors targeting the PDE4 isozyme have been developed. With phase II and phase III clinical trials currently underway to evaluate the safety and

efficacy of the latest generation of PDE4 inhibitors, namely apremilast, a new class of treatments may be around the corner for patients suffering from chronic, autoimmune diseases.

Abbreviations
A: Amyloid beta; ACR20: American college of rheumatology criteria for 20% improvement; AE: Adverse event; AS: Ankylosing spondylitis; ASAS: Assessment of spondyloarthritis; cAMP: Cyclic adenosine monophosphate; CREB: cAMP-response element binding protein; COPD: Chronic obstructive pulmonary disease; CXCL: Chemokine C-X-C motif ligand; DLQI: Dermatology life quality index; DMARD: Disease-modifying antirheumatic drug; EAE: Experimental autoimmune encephalomyelitis; EASI: Eczema area and severity index; ELISA: Enzyme-linked immunosorbent assay; EULAR: European League Against Rheumatism; GLP-1: Glucagon-like peptide; IFN: Interferon; IL: Interleukin; MMP: Matrix metalloproteinase; MPO: Myeloperoxidase; MS: Multiple sclerosis; NFB: Nuclear factor B; NO: Nitric oxide; NSAID: Non-steroidal anti-inflammatory drug; PASI: Psoriasis area and severity index; PDE: Phosphodiesterase; PKA: Protein kinase A; PsA: Psoriatic arthritis; RA: Rheumatoid arthritis; RANKL: Receptor activator of NFB ligand; SASI: Sarcoidosis activity and severity index; SLE: Systemic lupus erythematosus; TNF: Tumor necrosis factor; UC: Ulcerative colitis; VAS: Visual analog scale.

Competing interests
ABG is on the advisory boards of Abbott Laboratories, Actelion, Amgen, Astellas Pharma US Inc., Belersdorf Inc., BMS, Celgene Corporation, Coronado Biosciences, Janssen-Ortho Inc., Novo Nordisk A/S, Pfizer Inc., and UCB. ABG is a consultant to Abbott Laboratories, Amgen, Canfite, Celgene Corporation, Incyte Corporation, Lilly ICOS LLC, Merck & Co. Inc., Novartis Pharmaceuticals Corp., and Teva. ABG holds an investigator role for Abbott Laboratories, Celgene Corporation, Immune Control, Janssen-Ortho Inc., Lilly ICOS LLC, Novartis Pharmaceuticals Corp., Novo Nordisk A/S, Pfizer Inc., and UCB. ABG has received an institutional grant from Amgen, and honoraria from DermiPsor. NKu, AMG and NKi have no competing interests.

Authors contributions Abstract (provisional)

Atherosclerosis, the major cause of cardiovascular disease (CVD), is a chronic inflammatory condition with immune competent cells in lesions producing mainly pro-inflammatory cytokines. Dead cells and oxidized forms of low density lipoproteins (oxLDL) are abundant. The major direct cause of CVD appears to be rupture of atherosclerotic plaques. oxLDL has proinflammatory and immune-stimulatory properties, causes cell death at higher concentrations and contains inflammatory phospholipids with phosphorylcholine (PC) as an interesting epitope. Antibodies against PC (anti-PC) may be atheroprotective, one mechanism being anti-inflammatory. Bacteria and virus have been discussed, but it has been difficult to find direct evidence, and antibiotic trials have not been successful. Heat shock proteins could be one major target for atherogenic immune reactions. More direct causes of plaque rupture include pro-inflammatory cytokines, chemokines, and lipid mediators. To prove that inflammation is a cause of atherosclerosis and CVD, clinical studies with anti-inflammatory and/or immune-modulatory treatment are needed. The potential causes of immune reactions and inflammation in atherosclerosis and how inflammation can be targeted therapeutically to provide novel treatments for CVD are reviewed. Evolutionary medicine (EM) is a growing field focusing on the evolutionary basis of human diseases and their changes through time. To date, the majority of EM studies have used pure theories of hominin macroevolution to explain the present-day state of human health. Here, we propose a different approach by addressing more empirical and health-oriented research concerning past, current and future microevolutionary changes of human structure, functions and pathologies. Studying generation-to-generation changes of human morphology that occurred in historical times, and still occur in present-day populations under the forces of evolution, helps to explain medical conditions and warns clinicians that their current practices may influence future humans. Also, analyzing historic tissue specimens such as mummies is crucial in order to address the molecular evolution of pathogens, of the human genome, and their coadaptations.

Keywords:

anatomical variation; empirical; evolutionary medicine; microevolution; mortality; pathology; secular trends

Definition, history of evolutionary medicine research and present situation

Evolutionary medicine (EM), or Darwinian medicine as it is sometimes called, investigates human disease vulnerability and disease etiologies (genetics, behavior, environment, pathogens, and so on) from an evolutionary perspective. EM is a biomedical scientific concept of increasing interest since the 1990s [1,2]. It has been the topic of several textbooks [3-5] and also recently a major scientific colloquium [6]. The intellectual beginnings of evolutionary medicine stemmed from the recognition that past evolutionary events can explain present-day conditions of the human body. Thus, by applying the concept of nature's evolution to human morphology, physiology and pathophysiology, a better understanding of the etiology of present-day human ailments can be achieved. Early applications of poorly understood Darwinian concepts to human biology led to eugenic theories [7,8]. However, EM (as we strongly emphasize here) does not deal with eugenic approaches. It espouses approaches to population biology that do not deal with individuals, but with intergenerational manifestations of biological processes that have no value attached to them. Humans still evolve, in terms of anatomical structures and physiological processes as well as disease patterns and prevalence. The platonic, essentialist view that Homo sapiens, once formed, remains the same biological entity throughout the centuries is patently incorrect. Irrespective of the disparate views on the origin of humans held by adherents to different religions and scientific theories, changes in human genes and phenotypes from generation to generation do occur. Microevolutionary changes in human lineages during historical times are clearly understandable in the evolution of immunity to diseases, but also in the appearance of new metabolic processes such as lactose tolerance [9] or in the widespread acquisition of genetic variations in the capacity to process ethanol [10]. They have occurred in anatomical structures, too; such significant changes in morphological characteristics include: decrease in the robusticity of the musculoskeletal apparatus (gracilization) [11,12], weight and height [13], microcranialization and brachycephalization (reduction in braincase size and change of its shape) [14], reductions in the size and number of teeth [15] and spinal morphology alterations [16]. These alterations are all likely to be at least partially the result of structural reductions in response to technology diminishing the need for the use of physical strength and introducing extraoral food processing. Aside from genetic changes, such alterations may occur due to environmental changes such as a reduction of chewing effort in the processing of food, leading to a mechanically caused reduction in jaw size. Microevolution is observable as a process of changes occurring in phenotypes of successive generations. These changes may result from changing, under the operation of forces of evolution, gene frequencies, or from adaptive phenotypic responses to changing living conditions. The changes of gene frequencies are a part of the general evolutionary process involving mating systems, drift, gene flow, mutations and selection. They can only occur through the process of reproduction that requires the genetic endowment of one generation to be passed on to the next generation. During that process frequencies of alleles or of genotypes can be altered leading to permanent alterations of immune responses, physiological processes and anatomical structures. Phenotypic adaptive responses are modifiable through alteration of living conditions during the lifespan of one generation, but only within the limits of genetically determined plasticity of individual responses to environmental stimuli. Theoretically, the minimum time span required for microevolutionary change of the gene pool is that of two generations, while there is no constraint on the minimum time span for an adaptive phenotypic change. Since, due to a long human fertile lifespan of some 30 years, generations are widely overlapping in living populations, while living conditions in modern economies change rapidly, it is not always easy to distinguish between a truly evolutionary change and a phenotypic secular trend if the specific genetic determination of changing functions or structures is not known. Although classic descriptions of evolutionary processes refer to long time spans, there is no reason to expect that a change in gene frequencies may not occur during the time span of a century. The average age of parents at the time they produce offspring is approximately 20 to 40 years and thus on average three generations can be turned over during a century, each providing an opportunity for change of gene frequencies. This change may be rapid if a particular force of evolution operates strongly. For example, gene flow resulting from mass migration may profoundly alter the gene pool of a given geographic region within several decades. Human microevolution has recently accelerated due to the rapid growth of human population numbers facilitated through cultural development and technologies [17]. Phenotypic manifestations of these changes are sometimes referred to as secular (derived from the Latin term saeculum, for 'a generation') or microevolutionary alterations. A distinction can be made between secular changes and microevolutionary alterations, based on their causes as explained earlier. Secular changes, such as increases in stature or weight are usually alterations of phenotypic expression of genetic potential without any changes in gene frequencies, while true microevolution involves change of gene frequencies, like in the case of accumulating mutations. Since for many morphological and physiological characters the exact mode of inheritance is not known, the distinction between phenotypic adaptive trends and true microevolution

may be made by observing whether the magnitude of a particular change exceeds the range of adaptive phenotypic responses of the same genetic potential. If generation-to-generation changes exceed full phenotypic expression of the same genetic potential they can be considered microevolutionary ones, since they must reflect the changing genetic endowment of successive generations. Most microevolutionary alterations have medical implications for individual patients (for example, knowledge of current anatomical variations for surgeons) as well as at the population level (for example, sociospecific public health measures). Studies of microevolutionary changes require time depths of at least a few generations, thus EM research specifically uses historic samples, where the time periods investigated extend over a number of centuries or even a few millennia. The value of such studies of ancient tissues has become more and more accepted even for clinical research, particularly as a crucial reservoir to study the evolution of infectious diseases [18-21]. The aim of this review is to highlight the potential of novel directions in EM empirical research for current and future biological and medical applications rather than discuss pure theoretical understanding of the origin of humans. Thus, it discusses present-day public health activities and biomedical practices from a perspective of future generations. Furthermore, the value of ancient tissue samples such as mummified bodies and archaeological bones and teeth to study recent evolution of human disease is addressed, as well as the possible impact of EM on academic curricula.

Current EM research
Major fields of primarily non-clinical EM research to date have included aspects of demography[22], evolutionary genetics [23], sex [24], and socioanthropological issues [25]. The value of EM has been recognized particularly for clinical research [26,27]. Presently, EM concepts have been applied in clinical settings with a major focus on disease-provoking morphology, for example, of the human spine [28], on the changes in infectious diseases through time [3], explanations of psychiatric diseases such as depression, schizophrenia, anxiety disorders, and personality disorders [29-31], metabolic disorders such as iron deficiency [32] or nutritionbased pathological effects [33,34].

Possible approaches in future EM studies

Relaxed natural selection and microevolution of human morphology
One major field for future evolutionary research with a particular biomedical perspective is the study of alterations of natural selection, understood as differential reproductive success of carriers of different genes, and its impact on human morphology and pathology. Over the entire evolution of humankind, there was a very significant opportunity for the process of natural selection (Figure 1). It mostly occurred due to high levels of differential mortality that allowed less than half of individuals born to pass on their genes to next generations, eliminating the other half [35]. Until the mid-19th century infant and child mortality was so high that the survivorship to age 15 years was around 50% or somewhat less, even in countries presently considered to be 'developed' [36,37]. Although some deaths happened without a link to individual genetic endowment, many were linked to varying physical strength, levels of immunity, metabolic disorders (for example, type 1 diabetes, phenylketonuria), vision defects [38] and less common congenital defects. Differential fertility contributed much less to the overall opportunity for selection since there was little genetic variation in this characteristic [39]. This situation has changed drastically during the last approximately 150 years with the most welcome advent of sanitation and generally available medical treatments. The opportunity for natural selection through differential mortality has been so severely reduced that, at the end of the 20th century, more than 90% of newborns had an opportunity to fully participate in the reproduction of the next generation [36], while fertility became dependent on the conscious decisions of individuals and couples in both the sense of avoiding births and giving birth by infertile couples. For the first time in the evolution of humanity, the majority of natural selection pressures were relaxed to the apparent benefit of all of us. The increase in variability of heritable traits is a predictable outcome of such a relaxation of selection as its stabilizing effects are diminished [40]. This might not be true for psychiatric disorders, where social pressure may still influence reproductive success [31]. There is also evidence that, at least for some disorders, psychiatric disorders may be linked to allele variations that predispose to differential susceptibility and adverse effects in terms of developing a disorder [41]. At the genetic level, alleles do not have an absolute adaptive or maladaptive value, they assume it by interaction with the rest of the genome and the epigenetics determines their Darwinian fitness [42]. A gene producing pathological effects in the past (for example, predisposition to type I diabetes mellitus) may not be considered maladaptive in an environment where there is an effective treatment for diabetes.

Figure 1. Changes in the Biological State Index over the last 15,000 years of human evolution. The index value is a probability that an average person will be able to fully participate in reproduction of the next generation. The lower the index value the greater the opportunity for natural selection. Labels in the formula are: d x = a fraction of dying people of age x; sx = reproductive value of a person of age x (for example, s65 = 0, while s15 = 1). For further explanation and data see [35,70]. One can multiply such clinically relevant examples of relaxed natural selection. For example, an increase in the range of human biological variation has been already documented for a plethora of anatomical structures. Some 'anomalous' arteries have more than doubled their prevalence (for example, the median artery of the forearm is now present in around 30% of individuals in different populations, while at the beginning of the 20th century it was present in only around 10% of individuals [43]) (Figure 2), and the thyroidea ima branch of the aortic arch had disappeared completely by the end of the 20th century [44].

Figure 2. Frequencies of individuals with median arteries of forearms by date of birth in a sample of 284 South African dissection cadavers. Trend of increase in the incidence is significant (2(1) = 11.90, P <0.001 z = 3.94, P <0.0011 (z test for linear trends in proportions)). For further data see [71]. Climatic factors have been proposed to influence the altered prevalence of the internal thoracic artery [45]. In the skeletal system, opening of the sacral canal (spina bifida occulta) became more common in cohorts born in the second half of the 20th century than it was before [46], and tarsal coalitions appear more often in more modern times, too [47]. Skeletal pathologies such as ossification of the posterior longitudinal ligament of the spine have increased [48] as have diffuse idiopathic skeletal hyperostosis [49]. Many other rather short-term changes of body morphology, such as alterations in body dimensions and proportions (for example, body mass index, skeletal robustness or bone density) have also been shown. The widely reported secular increase in stature, which occurred with varying speeds (from 0 to over 150 mm per century [50]) in various populations, has affected body proportions since most of the stature increase, where it occurred, was due to the growth of the lower extremities [51]. Also, the current epidemic of obesity may in part result from increasing variation in the size of the body frame that reflects a greater variation in the size of the gastrointestinal tract [52] rather than just caloric imbalance. The range of variations of hormones regulating human appetite, for example, leptin and ghrelin and enzymes regulating carbohydrate and fat metabolism in past and present populations may differ, thus adding to the evolutionary explanation for part of the obesity problem. Although it can be argued that short-term changes in body height and body weight are not the result of changes in gene frequencies, but simply adaptive, non-heritable responses to changing living conditions, the ability of the human body to respond to such changes is a product of its earlier evolution. The response, especially in the case of increasing body weight, seems to be harmful and needs to be treated by interventions based on the understanding of human heritable adaptations to past diets, the so called thrifty genotype hypothesis debate (for example, modern diabetes-causing genes were advantageous in the past) [53]. The economic impact of such body shape alterations on 'biological standards of living' has been addressed previously [54]. Besides direct economic costs, obesity is linked to increased mortality and morbidity and thus any short-term alteration in obesity rates will have huge public health implications. Finally, it is also not clear if the entirety of body height increases that occurred during the 20th century are adaptive rather than of a genetic nature [55]. Regulation of postnatal growth and development has undergone a significant transformation during the last century. This change has become most obvious in the adolescent period [56]. Sexual maturity accelerated, while the rates of growth at puberty became much higher than before, resulting in problems in adolescence [56]. It remains to be determined to what extent acceleration of sexual maturation and increases in peak growth velocity are the result of alterations in socioeconomic conditions, and to what extent microevolution of human growth regulation has occurred. Studies of the occurrence of skeletal manifestations such as hyperostosis frontalis interna, may possibly further elucidate the recent evolution of the human endocrine system [57]. Other examples might be the alterations in prevalence and the etiology of metabolic syndrome, and the introduction of biologically active substances (for example, xenoestrogens or endocrine disruptors) into the food chain [58]. Finally, even within short time periods disorders of unknown cause, such as Paget's disease, can show a notable yet etiologically enigmatic alteration in prevalence [59]. Thus, recording these secular alterations is the very first step to explore possible environmental cofactors of such disorders. It is evident that our biological properties are changing even within very short historical time frames. More research

elucidating what changes occur, with what intensity and to predict their biomedical consequences is needed, and should be a major future field of EM research.

Lessons from paleopathology: Evolution of diseases and genomic studies

Of special importance to EM is the subdiscipline of paleopathology, which attempts to describe diseases in the past and to trace changes of those diseases in response to the historical development of humans, especially during the last several thousand years. From diagnoses of individual cases observed in ancient skeletons and in mummies, the discipline has evolved into palaeoepidemiological studies [60,61], even though meta-analytic standards known from clinical studies can hardly ever be met. Studies have discussed the impact of recent genetic sweeps such as the positive selection of Tay-Sachs disease-affected people versus tuberculosis [62]. Also, sex differences in genetic vulnerability to cancer or arteriosclerosis can be addressed by EM research. Many attempts at explaining host-pathogen coevolution in relation to major infectious diseases such as leishmaniasis or plague have been made [63,64]. Epigenetics is another field for future EM research. Epigenetic factors mediating gene expression such as early-in-life stress ('fetal programming') would be one such example. Poor intrauterine conditions are predictive for somatic and psychiatric disorders, including maternal adversity [65]. Since it has been suggested that micro-RNA is linked with human pathologies such as cancer, molecular evolutionary studies may solve certain etiological enigmas. Another example, the lively and still continuing debate about the origin of syphilis [66,67], has stimulated closer scrutiny of pathogens, the study of their impact on the health of populations, systems of public health and ways of handling the recent resurgence of treatment resistant forms of the disease. For genomic studies in particular, new technological advances will allow more sensitive and specified research.

Outlook
Medically oriented empirical research with an evolutionary focus may help to redesign public health policies and public awareness of science. A 'morphological anomaly' may become more frequent or even 'normal' in a given population and, thus, it should be no reason for concern for a particular individual. This needs to be realized and communicated accordingly (for example, by general practitioners to their patients). Accepting variation as normal is an important issue in clinical medicine. To summarize, human biological traits still evolve. We are not simply 'stone age bodies in a modern world', but we are both at the same time adapted and adapting; biological compromises in a rapidly changing environment, with the latter also being full of coevolving pathogens. Therefore, future clinical studies in EM should focus particularly on the genomic evolution of bacterial and viral diseases and responses in the evolution of human immune system. For the latter, DNA viruses are easier to extract than RNA viruses. Issues such as viral pandemics or evolution of strain-dependent virulence can be explored by using a temporal and thus historical perspective. As highlighted previously [68], the imminent conflict of our short-term and long-term evolutionary genetic endowment is etiologically linked to the major causes of death in first-world countries such as cardiovascular or oncological disease. Thus, any progress in fighting disease based on evolutionary insights would be most welcome in the medical community, as well as in the general community. With the prospect of improved ancient DNA and proteomic analyses, we are now only at the edge of an entirely new era that will allow us to unravel the mysteries of human disease evolution. Furthermore, the incorporation of principles of human evolution and its forces into the knowledge of future medical practitioners is needed. A general practitioner may not directly heal a patient using only EM principles, but without any evolutionary knowledge he/she certainly will not be able to provide the best, individualized diagnosis, medicosocial advice and prescribe optimal personal treatment [69]. The same is true for any biomedical researcher; not applying EM principles may restrict the true scientific impact and applicability of a particular research result. Thus, the introduction of EM topics into medical (and science) curricula is recommended.

Abbreviations
EM: evolutionary medicine.

Competing interests
FJR has received research funding from the Mxi Foundation. No other relevant conflicts of interest exist. MH has no conflicts of interest.

Authors' contributions
FJR conceived the idea for this paper and discussed it with MH. MH produced the first draft, which was substantially expanded and modified by FJR. Both authors contributed examples from their earlier studies and incorporated them into the text. Both authors edited the text and approved the final manuscript.

References

Despite substantial heritability in pubertal development, much variation remains to be explained, leaving room for the influence of environmental factors to adjust its phenotypic trajectory in the service of fitness goals. Utilizing evolutionary development biology (evo-devo), we examine adolescence as an evolutionary life-history stage in its developmental context. We show that the transition from the preceding stage of juvenility entails adaptive plasticity in response to energy resources, other environmental cues, social needs of adolescence and maturation toward youth and adulthood. Using the evolutionary theory of socialization, we show that familial psychosocial stress fosters a fast life history and reproductive strategy rather than early maturation being just a risk factor for aggression and delinquency. Here we explore implications of an evolutionary-developmental-endocrinological-anthropological framework for theory building, while illuminating new directions for research.

Keywords:
adolescence; evolution; juvenility; puberty; youth

Introduction
Evolutionary development biology (evo-devo) is concerned with how developmental systems evolved, while probing the consequences of these historically established systems for organismal evolution [1]. Research in evo-devo has formed around comparative embryology and morphology, evolutionary developmental genetics, and experimental epigenetics. Here we examine adolescence from an evo-devo perspective, treating this life-history stage of rapid growth and maturation in its ecological and developmental context [2]. Modern lifestyle and medicine have influenced nutritional and infectious constraints on puberty, resulting in the secular trend in pubertal development over the past 150 years. In girls, more than in boys, the change in pubertal age has been intriguing; in the last decades, the rate of precocious sexual maturation in girls has been high and increasing, and the mechanism for the 'epidemic' has been much debated, pointing fingers at toxins and perhaps other chemical products. Evidence that the timing of boys' somatic maturation is changing has recently been reviewed and remains inconclusive [3], though one Danish study documented a 3-month acceleration in male pubertal onset across a 15-year period (from 11.92 years in 1991 to 11.66 years in 2008) [4]. Evolutionary analysis highlights the fact that it is the female who is reproductively constrained in terms of the maximum number of offspring she can generate over her reproductive years. In consequence, early maturation affords a potential fitness advantage for females more than males, allowing more time to reproduce. Thus, evolutionary life-history thinking challenges the prevailing notion that early puberty is exclusively or primarily pathological in origin, viewing it rather as an adaptive response to changing life conditions. Indeed, as we hope to show, evidence indicates that since the emergence of homo sapiens there has been much change in the timing of pubertal maturation - and not just in a singular direction - and a variety of contextual factors appear to regulate pubertal development. To our way of thinking, it is a mistake to focus only on environmental toxins or even simply cast changes in pubertal timing in disease terms. We challenge the pathological view by advancing an evolutionary perspective on the issue of juvenile transition and timing of pubertal development, as drawn from life-history theory. Toward this end we consider the anthropological record, showing that adolescence as a stage was a new development in primate life history and that over time there have been many changes in pubertal timing, both accelerating and delaying it. While acknowledging heritable individual differences in pubertal timing, we emphasize developmental plasticity and the role of the environment in regulating pubertal timing in the service of fitness goals, using hormonal and developmental mechanisms. A central claim will be that the transition from the preceding stage of juvenility to adolescence entails adaptive developmental response to energy resources, other environmental cues, social needs of adolescence and maturation toward youth and adulthood, with the latter defined as the life-history stage of reproduction (Table 1). The plasticity which we argue characterizes adolescence is regulated by hormonal processes. We explore implications of this evolutionary-developmentalendocrinological-anthropological framework for theory building, while illuminating new directions for research. Table 1. Developmental tasks for adolescents and young adults [67].

Conceptual foundations
Life-history theory

Evolutionary life-history theory deals with the strategic allocation of an organism's energy toward growth, maintenance and reproduction, including raising offspring to independence, while avoiding death [2,5]. It predicts that selection will promote fitnessenhancing physiological, psychological and behavioral mechanisms that make strategic tradeoffs involved in the allocation of energetic resources to influence the three foundations of natural selection: survival, sexual selection and fertility fitness. Relative to other species, human life-history strategy includes a long period of postnatal growth, including dependency to sexual maturity, rapid adolescent growth and delayed reproduction [5]. Consideration of intermediate growth stages and the transitions between them from a life-history perspective affords insight into strategic objectives that include the age of pubertal onset, pubertal tempo, ultimate size and cognitive targets.

A matter of definition
The terms puberty and adolescence are often used interchangeably and thus incorrectly. Whereas puberty refers to the activation of the neuroendocrine hypothalamic-pituitary-gonadal axis that culminates in gonadal maturation and the biological effects of sex steroids, the package we call adolescence includes such pubertal development plus the growth spurt, cognitive and brain maturation, and social aspects in learning, intimacy and mutual support, intensification of pre-existing friendships, development of new relationships, and the attainment of biosocial skills needed for successful reproduction. The collective endpoint of the adolescence package is the socially and reproductively mature adult. To promote reproductive and parenting success in the service of reproductive fitness, hormonal and mental maturations are intimately coupled through iterative transactions between the nervous system and endocrine systems, with the latter involving gonadal steroid hormones [6].

Adolescence as a unique life-history stage

As late as 3,000,000 to 4,000,000 years ago the early homininae Australopithecus afarensis had only three postnatal, pre-adult lifehistory stages, just like the chimpanzee - 5 years of infancy, 5 years of juvenility and 2 years of youth - before the onset of reproduction [2]. During the evolution of the Hominidae, childhood and adolescence were inserted as new life-history stages ofHomo sapiens [7]: infancy, lasting 30 to 36 months; childhood, lasting an additional 2 to 4 years; a juvenility stage of 3 to 4 years of semi-independence, followed by adolescence, which lasts 3 to 5 years; and a youth stage, which lasts an average of 4 years [2]. The uniquely human adolescent growth spurt is often regarded by physical anthropologists and ape biologists as the operational definition of adolescence [7], even though it starts before the emergence of secondary sexual characteristics in girls, and much later than the onset of genital changes in boys. There is no evidence for such a human-like adolescent growth spurt in any living ape. With obvious limitations as to what can be inferred from fossilized skeletal remains, there is some suggestive evidence that 1,800,000 years ago hominids may have had a pattern of growth indicative of an adolescent stage of development [8]. In addition to being a period of rapid growth, adolescence is a time of subcutaneous fat deposition, especially in girls. Whereas subcutaneous fat is evenly spread over the female chimpanzee's body, the human adolescent has striking fat deposits in the thighs, buttocks and breasts, even if she is thin overall. These enable her to get through periods of scarcity, signal sexual maturity and facilitate sexual attraction of a mate, and allows others to continuously monitor her nutritional state[9].

Boys and girls coming of age

Boys and girls embark on different come-of-age strategies to achieve their fertility goals (Table 2). The onset of puberty in girls is generally considered to take place when breast buds erupt (thelarche), but even during childhood and juvenility girls have active ovaries that generate estrogens [10]. It is now recognized that thelarche is not the first sign of maturation of the female hypothalamic-pituitary-gonadal axis. Very much like boys, whose gonads are assessed via direct palpation and show testicular growth before sex-steroid concentrations increase, the ovaries start to grow discretely about two years before thelarche; estradiol levels increase during this period also [11]. Growth acceleration in the girl occurs some 6 months before the budding of breasts, and menarche about a year after the peak height velocity. Table 2. Adolescence in boys and girls manifests differently with regard to their actual fertility. From this time forward, girls have an apparent womanly body form, yet are not fertile; they will develop an adult cycle of ovulation and adult size of the birth canal much later, around 18 years of age. They gain knowledge of their adult social roles while still infertile, but perceived by adults as mature. The perception of fertility in girls facilitates their entry into the social-economic-sexual world of adult women, allowing them to practice reproductive skills without conceiving [12]. Boys show a pattern of gradual maturation of the hypothalamic-pituitary-gonadal axis similar to that of girls, becoming fertile at 14 to 15 years on average, about two years after their peak height velocity. But they are still young in outward development, body size, voice and facial features. Boys will learn their adult social-economic-sexual roles while already sexually mature but not yet perceived as such by adults. This allows them to interact and learn from older adolescents and adults without seeming to compete for status and other important resources, including fertile females [7,13]. Testosterone, which plays a central role in male peak growth velocity, appears to be important for activation of the courtship behavior that leads to the formation of sexual pairing bonds [14].

Before and during adolescence

Preceding adolescence - the juvenile stage

All mammals, including the great apes, transit directly from infancy to juvenility without passing through the childhood stage except humans (Figure 1). Comparison with African apes suggests that the timing of the transition to juvenility, as measured by adrenarche, may be similar to that in humans, although the full course of age-related changes in dehydroepiandrosterone sulfate (DHEAS) and their relationship to reproductive and brain maturation are not clear [15].

Figure 1. Evolution of the hominidae life history during the first 20 years of life. During evolution, childhood and adolescence have been added as new life-history stages and compared with apes and the presumably early hominidae. The chimpanzee serves as a living representative of the assumed Australopithecus afarensis life history. As childhood emerged and prolonged, infancy has gradually become shorter, and the latest introduced adolescence came at the expense of a shorter juvenility. Detailed accounts of these stages are given in[2]. We and others have defined juvenility as a distinct life-history stage in humans, characterizing it in terms of endocrine and body composition changes resulting in changes to social assignments and psychological maturation [16,17]. Developmental psychologists refer to this period as 'middle childhood', 'the five-to-seven-year shift' and 'the age of reason and responsibility' [18]: the brain reaches its final size - even if neuronal development is not complete - and primates, equipped with adult molars, forage independently for food and care for themselves. In modern societies the transition to juvenility coincides with the age when children go to school and compete to some extent with adults for food and space, while working out their social standing among age-mates. Coinciding with participation in adult social activities, juveniles develop a strong odor during the juvenile period; intriguingly, olfactory aversion emerges in the case of father-daughter and brother-sister, but not other family relationships, presumably for incest avoidance [19]. Del Giudice contends that juvenility (adrenarche) represents a 'switch point', a time when the environment can reprogram nascent reproductive strategies established earlier in life [17,20]. Indeed, he argues that sex differences in attachment relationships emerge in middle childhood and have adaptive significance for sexually selected life-history strategies. Early psychosocial stress and insecure attachment during juvenility direct development towards mating-oriented reproductive strategies; insecure males tend to adopt avoidant reproductive strategy, whereas insecure females tend to adopt an anxious/ambivalent strategy (which maximize investment from kin and mates). Strategies such as those involving bearing few or many offspring are passed to future generations [17]. In social terms, juvenility offers opportunities to prepare for the social complexities of adolescence youth and adulthood - in part by assaying one's social status and standing in the competitive world of peers [17]. Transition from childhood to juvenility is marked by the onset of adrenal androgen generation (adrenarche), adiposity rebound, deceleration of growth [21], and the eruption of the first molar teeth [16]. Whereas humans and chimpanzees exhibit adrenarche, other primates such as the baboon and rhesus monkey do not, and the adrenals of most other mammals produce little or no DHEA [22]. Thus, adrenarche is a recent evolutionary event. The human and chimp DHEA-generating enzymes, 17,20-lyase, differ at only two amino acids, whereas the human/chimp enzyme differs from the baboon or rhesus enzyme by 25 to 27 residues (95% identity) [22]. Serum DHEA and DHEAS rise progressively throughout juvenility [23], with effects on a wide variety of physiological systems, including neurological [24]and immune [25], as well as somatic growth and development [21,26]. DHEA in humans operates as a neurosteroid, affecting neurological functions and modulating mood [27,28]. The age at transition from childhood to juvenility has been remarkably constant, especially when compared to the plasticity that characterizes other life-history features such as age of sexual maturation [29]. Comparison with the African apes suggests that the timing of adrenarche and the sex difference in chimpanzees timing of transition from infancy to juvenility may be similar to that of humans moving from childhood to juvenility [15,30]. Assuming an important role for adrenarche in human brain maturation, Campbell argued that the increased brain size and extended lifespan of humans relative to the great apes imply changes in the timing and impact of adrenarche [15]. Thus, increases in body size evident among Homo erectus imply increases in lifespan and delayed adrenarche and reproductive maturation, and as such are a natural point at which to consider the potential role of delayed transition from childhood to adrenarche in human evolution.

Transition from juvenility to adolescence

Age and size at adolescence have strong effects on an individual's fitness because they affect reproductive potential, schedule and efficiency [31]. Emphasizing fitness goals, early transition to adolescence and the abbreviation of its duration increase the likelihood of reproduction before death, so may prove especially adaptive under conditions of ecological risk (provided nutrients are sufficient to foster maturation). Accelerated pubertal development also reduces generation time, while potentially lengthening reproductive lifespan. Alternatively, late transition into adolescence lengthens preadolescent growth and the opportunity to embody or internalize

the various resources to which the individual is exposed, be those resources nutritional, social or psychological. At the same time, delayed maturation prolongs the preadolescent hazard period, which may be compensated for by continuing parental care. Ultimately, individuals face a tradeoff between maturing to reproduction young and small and maturing at large body size, since for any given growth rate earlier maturation implies smaller size at transition.

Adaptive developmental plasticity

The changing age of adolescence
Substantial physiological variation of some 4 to 5 years is evident for age at onset of sexual maturation across individuals under varying life conditions [32]. Despite substantial heritability in pubertal development [33,34], much variation remains to be explained [34], leaving room for the influence of environmental factors to adjust the phenotype in the service of fitness goals [32]. The human population has grown exponentially from a brink of extinction 80,000 years ago, with a world population of several thousands, to that of several billion today. Such population growth increased opportunity for genetic mutations, thereby accelerating the pace of human evolution[35]. Estimates indicate that the age for menarche around 20,000 to 12,000 years ago - at the beginning of the agrarian period - was 7 to 13 years, and that full reproductive competence in Neolithic females (New Stone Age, 12,000 to 5,000 years ago) occurred at age 9 to 14 years (Figure 2) [36]. This would place menarche at 7 to 12 years, assuming a 2- to 4-year gap between menarche and reproductive competence [37]. This suggests that menarche in Neolithic times occurred even earlier than observed in modern Western countries [36]. This claim is consistent with data on the Aeta of the Philippines, who reproduce as early as age 10 to 14 years [38].

Figure 2. Menarche age over the last 12,000 years. The age of menarche gradually increased until the recent secular trend's decline, as shown in Figure 3. Data from [36]. Developmental and maturational tempo is flexible and responsive to environmental conditions in a presumptively adaptive manner. When immature animals experience severe environmental stresses such as malnutrition or disease, maturation is often delayed until conditions improve and normal growth can resume. By contrast, when animals are raised under ideal conditions that promote rapid growth, internal checkpoints ensure that maturation does not occur until juvenile development is complete. But when contextual stress is not so great as to challenge survival, pubertal development is accelerated, thereby increasing the likelihood of reproduction before death or disability. Collectively, these phenomena highlight a U-shaped link between contextual risk and nutritional cues as they predict pubertal development (Figure 3). Nutritional cues have clear temporal influence, including the timing of juvenility and adolescence, with a trend towards earlier maturation among those whose average body mass early in life is lower or higher than average, yet later among those with a poor childhood weight gain [32], resulting in a U-shape relationship [39,40].

Figure 3. The U-shaped link between contextual risk and nutritional cues as they predict pubertal development. When immature animals experience severe environmental stresses such as malnutrition or disease, maturation is often delayed until conditions improve and normal growth can resume. By contrast, when animals are raised under ideal conditions that promote rapid growth, internal checkpoints ensure that maturation does not occur until juvenile development is complete. But when contextual stress is not so great as to challenge survival, pubertal development is accelerated, thereby increasing the likelihood of reproduction before death or disability. Domestication of animals and agriculture altered the human environment - and, thereby, human development - in several ways, including adaptive changes in the onset of puberty. A relatively sedentary lifestyle increased local human densities, facilitated the spread of infectious diseases and was associated with recurrent famines. Later maturation to adulthood was a tradeoff to adapt to poor nutrition as well as the increasing complexity of being an adult in a society engaged in agriculture, settlement and population aggregation. Increased differentiation of social tasks and the creation of societal hierarchies in wealth-accumulating agrarian societies resulted in variation in nutritional status and family conditions, which themselves led to an overall increase in the mean age of menarche. This crucial point is further discussed below under 'Evolutionary Theory of Socialization'. Thus, by medieval times, the average age of menarche was deferred to 16.5 years, as it remains today among underprivileged adolescents in developing countries [32].

Modern hygiene and medicine have influenced nutritional and infectious constraints on puberty, resulting in the secular trend in pubertal development over the past 150 years (see below). Further evidence to this effect would seem to come from research showing that girls and, to a lesser extent, boys adopted from developing to industrial countries show accelerated sexual development[41,42]. The greater tendency of adopted girls to respond with pubertal onset to a changing environment is in line with female preponderance of idiopathic central precocious puberty. As note earlier, it is the female who has intrinsic constraints on the number of offspring she can generate over her reproductive years, and it is females more than males who may enjoy a fitness advantage from early maturation. Thus, evolutionary life-history thinking challenges the notion that earlier puberty is the result of a hypothalamic control malfunction, viewing it rather as an adaptive response to improving life conditions, similar to that witnessed in the case of the secular trend. The Philippine Aeta provide unique insight into the strategic importance of the age of adolescence. Their growth deviates from the U.S. 0.01th percentile, showing an early juvenile deceleration, early pubertal spurt and early growth termination [38]. With life expectancy of 16 and 27 years at birth and adulthood, respectively, their first reproduction occurs at age 10. Early reproduction minimizes the likelihood of death before reproduction. Thus it seems that early fecundity evolved to adapt to high-risk, highmortality life of a short duration, with shortness of stature resulting from a short period of preadolescent growth [38].

The secular trend in pubertal maturation

The secular trend provides compelling evidence that pubertal development is developmentally plastic. Over the past 150 years without any documented change in gene frequencies - the age of menarche has fallen by a full 4 years in the industrialized West (Figure 4).

Figure 4. The secular trend in puberty. Declining age of menarche in Western societies from 1840 to 2000. Data from [68]. The line does not show a saturation point; the trend is expected to continue. As much as the secular trend in human size has been an adaptive response to a nutritionally rich environment, the receding age of adolescence and pubertal development has been an adaptive response to positive environmental cues in terms of energy balance. The ever-younger age of girls' thelarche and menarche may have more than a single justification, however. In the last decade, the popular explanation has been that this phenomenon results from environmental endocrine disruptors that accelerate hypothalamic maturation [43]. Whereas endocrine disruptors may have a bearing on the earlier age of thelarche, which is a recent trend, it can hardly explain the secular trend in the age of menarche over the last 150 years. Following Belsky et al. [20] and Gluckman and Hanson [36], we explicitly challenge the concept that this has been a disease process, proposing that accelerated pubertal development reflects contextually regulated reproductive and life-history strategies. Indeed, the age at transition from juvenility to adolescence in humans has a variety of physical and social correlates. Women face a tradeoff between spending a long time accumulating resources through childhood growth, thereby improving the odds for successful pregnancy while also risking death before sexual maturation, against beginning early reproduction and increasing the number of reproductive cycles. A later first birth allows for a longer period of adolescent weight gain, and heavier women in traditional societies are more fertile; both these attributes correlate with higher birth rates. This tradeoff has been used to model the optimal age at first birth, which under such conditions is 18 years, near the observed mean of 17.5 years in such societies [44].

Evolutionary theory of socialization

Belsky et al. [20] advanced an evolutionary theory of socialization stipulating that familial psychosocial stress (for example, marital conflict, harsh parenting, father absence), itself induced by extrafamilial ecological stress (for example, limited income, unemployment), fosters a fast life-history and reproductive strategy. They claimed that pubertal maturation played a previously unappreciated role in linking early rearing experiences with subsequent mating and parenting, in line with the attachment theory of Bowlby [45,46], which is hereby expanded (Figure 5). The evolutionary reasoning was that early maturation would be selected under conditions of emotional risk and uncertainty, thereby setting the stage for earlier sexual debut, more promiscuous mating and the bearing of more offspring, along with lesser parental investment. Natural selection favors accelerated development when early life experiences suggest an insecure world in which intimate relations are not enduring [47]. Thus, slower physical maturation would risk lowering reproductive fitness and survival: in an insecure world, maturing early and breeding promiscuously would enhance reproductive fitness more than delaying development, mating cautiously and investing heavily in parenting. The latter would make evolutionary sense for reproductive fitness in a secure world, as perceived by the young child and juvenile [48]. Such theorizing is certainly consistent with evidence that earlier pubertal development is associated with greater sexual risk taking; earlier age of menarche is associated with earlier age of first dating, first kissing, first genital petting and first sexual intercourse, and higher rates of adolescent pregnancy, as reviewed[49,50].

Figure 5. Faster and slower reproductive strategies: Reproductive strategies develop in different contexts and are characterized by diverging patterns of psychological, somatic and behavioral development. More and less supportive family contexts (and broader ecologies) influence the quality and quantity of parental investment, which in turn influences psychological and behavioral development. Collectively these forces regulate the timing of pubertal development and, thereby, sexual behavior, pair bonding and eventual childbearing and parental investment. The faster strategy fits a world in which risk and uncertainty is high, whereas the slower strategy fits a world in which resources are predictably available and sufficient. The faster strategy enables the individual to reduce the risk of dying before reproducing and reflects the fact that the individual's capacity to attract and maintain a high-quality mate and provide resources to their own (eventual) offspring will be limited. The slower strategy reflects the opposite. Based on [20]. In the two or more decades since the adaptive-predictive-response theory of human development by Belsky et al. appeared, an abundance of evidence consistent with its critical pubertal-timing prediction has emerged (for review see [51]). Consider in this regard findings from longitudinal research indicating that limited family support during childhood (for instance, authoritarian parenting, negative family relationships) is associated with females' advanced adrenarche and early puberty [52], and that harsh parenting in early childhood predicts earlier age of menarche and, thereby, greater sexual 'risk taking' in adolescence [53]. Important also are data showing that younger sisters with greater earlier exposure to an absent father as a result of divorce or separation matured earlier than did their older sisters [54], and that girls evacuated from their homeland during World War II and sent to live in Sweden and Denmark reached menarche at a younger age and even bore more children than members of the same birth cohort who remained at home [55]. In line with the above, experimental manipulation of licking and grooming of newborn rats by their mothers illuminates the role of epigenetic processes in regulating the stress-response system, pubertal timing, sexual behavior and parenting [56], as reviewed [51].

Individual differences in developmental plasticity

Some individuals are more plastic and responsive to environmental cues and others less so, adopting a more fixed developmental trajectory for reproductive strategy [50,57-62]. Children who were more physiologically reactive in terms of cortisol response to a psychological challenge were more responsive to family forces in accelerating pubertal development [63]. In a recent geneenvironment interaction study, an allelic variation in the estrogen-receptor gene determined which girls' menarche age was accelerated by high levels of family conflict [64]. Such findings suggest that population estimates of environmental influences on pubertal development do not necessarily reflect individual response.

Conclusions
This review uses an evo-devo approach and life-history theory for understanding human adolescence and especially variation in the timing of reproductive maturation. Developmental and maturational traits that respond to environmental cues enhance fecunditysurvival schedules and behavioral strategies that yield the highest fitness in a given environment. Why is it that we have a unique adolescence phase, preceded by a juvenile phase and followed by a youth phase leading to such a delayed reproduction? Like other organisms, humans evolved to withstand environmental hardships by responding in ways that maintain evolutionary fitness, even if submaximal. The means to do this is a series of predictive adaptive responses that utilize the sensitive times of transitions from one life-history stage to the next, each assigned with its own domain. The transition from juvenility to adolescence is deferred when food supply is short, programming at the same time for later fecundity and fertility and for longevity [65,66]. The timing of puberty sheds light on the relationship between phenotypic adaptive plasticity and adaptive genetic changes. Whereas the gradual tendency to mature late over Homo's evolution is genetic, the gradual secular trend in industrial societies over the last 150 years is not, given the short timescale. The more recent tendency for earlier puberty reflects the overall quality of modern environments, allowing females to approach the extremes of their genetic range of reaction. Such evo-devo thinking demands reconsideration of the notion of 'precocious puberty'. This term implies pathology, whereas the vast majority of early puberty probably reflects normal physiological and adaptive developmental plasticity [36]. Thus, the term precocious puberty is to be reserved for those few with anatomical or genetic defects, without a precise definition in the present context.

The implications of disconnection of the mental and somatic components of human adolescence are underappreciated, and result in both mental and somatic consequences. Among them are the obesity and polycystic ovary syndrome epidemics, but also mental and social behaviors. In an American study, early-maturing girls displayed higher levels of self-reported criminality, drug abuse, social isolation, early sexual behavior and psychiatric problems [67]. Early-maturing girls, particularly those with a history of adolescent conduct disorder, were more likely to be depressed and to have many sexual partners in young adulthood compared with their counterparts. Early puberty may thus represent a social pathology in industrialized societies. The claim we make, however, is markedly different from the widespread assertion that early maturation is a risk factor for aggression and delinquency [68]. Rather, we contend here, that early life experience during infancy and childhood will be associated with a change in maturational tempo, such that harsh rearing conditions predict earlier age maturation and associated behavioral phenotypes, including perhaps aggression, delinquency and promiscuity because these responses were selected to promote fitness. This framework is quite distinct from the disease perspective that fails to appreciate the evolutionary wisdom of maturing early and behaving in opportunistic, advantagetaking ways under certain contextual conditions. Hereditary, environmental and stochastic factors regulate puberty in a unique environment, but their relative contribution to the phenotypic outcome and the extent of stochastic epigenetic reprogramming that is required to alter human phenotypes is not known because few data are available [69]. If the environment can influence developmental and maturational trajectories during pre-adult life-history stages, how do epigenetic events influence the transition from one life-history stage to the next, growth and puberty at the molecular level? Growth and puberty are regulated by insulin, growth hormone, the insulin-like growth factors and the sex hormones, to mention a few of the control factors. These hormones drive the rate of growth and development, but it is unclear how the environment shapes the timing of the different phases of developmental events. Epigenetic mechanisms potentially play an important role. Perhaps the most fundamental question raised by the life-history approach to adolescence concerns the uniqueness of each child in her given genetic background and current environment as they best serve her reproductive fitness. Given the evidence on the strong influence of socioeconomic conditions early in life, we have to better understand how these interact with or via endocrine mechanisms to generate signals that affect life history and adolescence.

Abbreviations
DHEA: dehydroepiandrosterone; DHEAS: dehydroepiandrosterone-sulfate.

Competing interests
The authors declare that they have no competing interests.

Authors' contributions
The purpose of this article is to review the imaging techniques that have changed and are anticipated to change bladder cancer evaluation. The use of multidetector 64-slice computed tomography (CT) and magnetic resonance imaging (MRI) remain standard staging modalities. The development of functional imaging such as dynamic contrast-enhanced MRI, diffusion-weighted MRI and positron emission tomography (PET)-CT allows characterization of tumor physiology and potential genotypic activity, to help stratify and inform future patient management. They open up the possibility of tumor mapping and individualized treatment solutions, permitting early identification of response and allowing timely change in treatment. Further validation of these methods is required however, and at present they are used in conjunction with, rather than as an alternative to, conventional imaging techniques.

Keywords:
Bladder cancer; Diffusion-weighted MRI; Multidetector computed tomography; PET-CT; Staging; Ultrasound; Virtual cystoscopy

Introduction
Bladder cancer is a common disease with significant associated morbidity and mortality. Globally it is the ninth most common cause of cancer related death in men [1,2]. The clinical spectrum at presentation can be divided in to those with (i) superficial or nonmuscle-invasive bladder cancer (NMIBC), approximately 70% to 80% of patients; (ii) muscle-invasive bladder cancer (MIBC), approximately 20% at presentation; and (iii) metastatic disease [3].

The challenge

Anatomically the bladder is divided into several layers. Accurate local staging with imaging is dependent on reliably distinguishing these layers. Clinical staging of the primary tumor is with bimanual examination under anesthesia. In those with MIBC this has been shown to be inaccurate in 23% to 50% of cases [4,5]. Transurethral resection of bladder tumor (TURBT) also understages tumors; approximately 30% to 50% patients are understaged at the time of cystectomy [6-10]. Therefore, accurate radiological correlation is important to help guide patient management. The objective of any staging modality is to achieve adequate visualization of the primary tumor, extent of loco regional disease and to determine the presence of metastases. However, conventional imaging modalities are unable to identify microscopic disease and can be inaccurate in identifying macroscopic disease. To address this newer imaging techniques are being explored to improve pretreatment staging, predict early response to treatment and provide non-invasive alternatives to cystoscopy for those requiring long-term surveillance. This article will review the development of imaging techniques that have changed, and are anticipated to change, bladder cancer evaluation.

Computed tomography (CT)

Multidetector (64-slice) CT scanning has provided the mainstay in radiological assessment. It has a reported sensitivity of 85% and specificity of 94% for the diagnosis of bladder cancers [11]. Detection is dependent on the morphology and size of the tumor. Flat lesions, carcinoma in situ(CIS), tumors less than 1 cm and in those whom recent resection has been performed, are more likely to be falsely negative [5,11,12]. Previous biopsy, inflammation, systemic chemotherapy and intravesical drugs also interfere with interpretation [5,13]. CT remains the modality of choice for investigating hematuria and has replaced intravenous urogram (IVU) at most centers; its precise role in diagnosing bladder cancer is controversial. Although a 64-slice multidetector CT provides high spatial resolution allowing visualization of extravesical spread it is not reliable in determining the extent of locoregional disease [14]. It is limited by interobserver variability and inability to distinguish the muscle layers of the bladder[5,15].

Ultrasound (US)
US is at present not used routinely in clinical practice for the assessment of known bladder cancer, although the presence of hydronephrosis is suggestive of MIBC. It is however an important diagnostic tool in the investigation of hematuria in particular to assess large renal masses/upper renal tracts. Bladder tumors may be visualized by ultrasound but a negative test does not exclude the presence of bladder cancer. Two-dimensional transabdominal and transvaginal US have been investigated with regards to aiding bladder cancer staging but transurethral ultrasound has demonstrated the best ability to visualize depth of tumor penetration through the bladder wall [16,17]. Use of two-dimensional US is often limited by the subjectivity and expertise of the examiner. It is also unreliable in determining deeply infiltrating disease and nodal involvement [5]. Contrast-enhanced US (CE US) has been shown to be superior to conventional US in differentiating non-invasive and invasive bladder tumors [18,19]. Prior to TURBT CE US has a reported accuracy of 88.4% versus 72.1% compared to standard two-dimensional US, with 94.7% sensitivity for tumors greater than 5 mm; for lesions less than 5 mm sensitivity is reduced to 20% with a negative predictive value of 28.6%[18]. Three-dimensional US has been developed to provide reconstruction of the actual tumor with visualization of the bladder wall layers. The volume data can be retrieved and manipulated as if in real time, which increases objectivity, allows views in multiple planes to be obtained and improves the rate of primary bladder tumor diagnosis (88.9% with two-dimensional US versus 100% with threedimensional US) when identifying T3b disease [20]. The disadvantage of this technique is that the entire tumor is not visualized and detection is particularly difficult when flat, plaque like tumors are present, there is coexistent calcification, the abdominal wall is rigid or the patient has central obesity [21]. Three-dimensional CE US uses enhanced images in three orthogonal planes and reflections of microbubbles to depict blood vessels. This has been shown to be clinically useful in differentiating MIBC and NMIBC [21]. Three-dimensional CE US may also have future role in assessing treatment response of the primary tumor to guide bladder-sparing approaches. Its use in assessing treatment has been evaluated other solid tumors including primary liver cancer following local therapy [22]. Although recent developments in US techniques overcome many of the limitations of two-dimensional US, they remain under investigation and have yet not translated in to widespread clinical use.

Magnetic resonance imaging (MRI)

MRI has excellent soft tissue resolution and multiplanar capabilities, which has made it an important staging modality for bladder cancer. Fundamental to its importance in local staging is the ability to manipulate image contrast by using different sequences. On T2-weighted images the bladder tumor is usually more conspicuous. The signal from perivesical fat can be suppressed using short tau inversion recovery (STIR) sequences, allowing signal from the tumor to be highlighted by suppressing signal from adjacent surrounding normal tissue [5,23,24].

In conjunction with gadolinium-containing contrast (Gd-Ca), MRI has an accuracy of 85% in differentiating NMIBC from MIBC and 82% accuracy in distinguishing organ confined disease from non-organ confined disease [25]. However, Gd-Ca should be avoided in those with renal impairment (estimated glomerular filtration rate (GFR) <60 ml/minute), as there is an increased risk of developing nephrogenic systemic fibrosis [26]. The clearest advantage of MRI over CT is the ability to determine the presence of muscle-invasive and extravesicle disease, and is our preferred modality of local staging prior to definitive radical treatment in MIBC. The use of functional MRI imaging to provide biological information of tumor characteristics is under investigation. Dynamic contrast-enhanced MRI (DCE-MRI) enables in vivo assessment of tumor blood flow and permeability using paramagnetic contrast agents. Visualization of tumor blood flow can be used to identify areas of hypoxia and subsequently be used to predict treatment response [27]. It is an effective biomarker in predicting pathological complete response in those receiving primary chemotherapy for breast cancer and chemoradiotherapy for rectal cancer[23,28]. Intrinsic-susceptibility-weighted or blood-oxygenation-level-dependent (BOLD) MRI, exploits the difference in magnetic susceptibility of oxyhemoglobin and deoxyhemoglobin. Deoxyhemoglobin is a paramagnetic molecule that allows it to act as an intrinsic contrast agent. BOLD MRI image acquisition during high oxygen concentration inhalation (carbogen, 95% oxygen, 5% carbon dioxide) reflects improved tumor oxygenation and blood flow and may help identify patients more likely to benefit from carbogen radiosensitization [29]. Diffusion-weighted MRI (DW MRI) is a functional imaging technique dependent on the inhibitory effect of cell membranes to the random motion of water molecules (Brownian motion) to generate image contrast by applying two equally sized but opposite diffusion sensitizing gradients, characterized by their b-values. As tumors have greater cellularity than normal tissue they demonstrate higher signal intensity (that is, restricted diffusion on MRI, reflected in the low mean apparent diffusion coefficient value (ADC)). This has the potential to provide both qualitative and quantitative information to aid tumor assessment. Histogram analysis of ADC through the entire tumor volume captures the diffusivity microenvironment and may aid identification of the heterogeneity known to exist within tumors that may have prognostic and predictive value [30-34]. DW MRI is also more accurate than T2-weighted MRI in staging both organ confined (pT2) (69.7% versus 15.1%) and higher stage tumors (92.5% versus 80.1%), with a reported sensitivity of 98.1% and positive predictive value of 100% [35]. There is also evidence that the ADC value may help identify high-grade tumors, with low ADC (<110-3 mm2/s) suggesting G3 disease[36]. Following treatment the ADC value increases, reflecting decreased cellularity, consistent with response. In other tumor types change in ADC has been used to identify and quantify early treatment response that may occur before conventional assessment of response is seen (for example, the Response Evaluation Criteria In Solid Tumors (RECIST) criteria), or where evaluation of morphological change is difficult to interpret [34,37-39]. DW MRI therefore has the potential for monitoring treatment response to chemotherapy or radiotherapy with identification of early non-responders who may benefit from change in treatment approach. Conventional assessment of local response is with cystoscopy. One study has explored the role of DW MRI to assess response to chemoradiotherapy in 23 patients with MIBC. It demonstrated that the ADC was the only significant, independent predictor of chemoradiotherapy response with a sensitivity, specificity and accuracy of 92%, 90% and 91% respectively. Consistent with other studies higher ADC was associated with unfavorable response [40]. Preliminary results for lymph node evaluation using DW MRI do not appear to be accurate (sensitivity 76.4%, specificity 89.4%, positive predictive value 86.6% and negative predictive value 71.4%) [41]. The clinical use of functional MRI in bladder cancer assessment is not yet clearly defined but work to date suggests it likely to provide important information that may help guide treatment selection.

Nanoparticle-enhanced MRI
Involved lymph node detection by convention is governed by size and shape. Nodes greater than 1 cm are considered malignant on CT with a sensitivity of 85%, specificity of 67%, with a false negative rate of 21% [5,14]. However, enlarged reactive, nonmalignant lymph nodes can mimic metastatic involvement with these modalities. Techniques that evaluate nodal function rather than morphology are aimed to more accurately characterize nodal disease. Lymphotropic nanoparticle enhanced MRI exploits nodal macrophage function to detect metastases using ultra-small superparamagnetic particles of iron oxide (USPIO) (ferumoxtran-10, Sinerem). After intravenous administration, the USPIO particles reach the lymph nodes via the lymphatics. Benign nodes have functioning macrophages, which phagocytose the USPIO causing them to accumulate within the node. This causes a drop in signal intensity on T2-weighted images. Metastatic lymph nodes that are partially or completely infiltrated are unable to take up these particles effectively. These nodal regions retain their signal intensity on T2-weighted images allowing detection on post-contrast imaging [42,43]. Nodal enhancement is dependent on the tumor burden. Failure to detect microscopic foci of metastatic disease in very small lymph nodes leads to false negative results. False positives are due to reactive hyperplasia, localized nodal lipomatosis and insufficient

USPIO [24]. Despite these limitations, the reported accuracy when evaluated prospectively in 58 patients prior to surgery is 95%, with sensitivity of 96%, specificity of 95%, positive predictive value of 89% and a negative predictive value of 98% [44]. However, USPIO are no longer readily available, which limits the scope for further investigation and clinical application.

Virtual cystoscopy (VC)

Three-dimensional surface modeling is possible using cross sectional data obtained from CT or MRI, allowing indirect visualization of the mucosa and simulation of endoscopic evaluation. This method has been used to assess a number of other hollow organs including the colon and bronchus. Once source images are obtained, VC is performed on a dedicated workstation using a variety of computer algorithms [45-48]. Previous VC studies have focused on the potential diagnostic capability of evaluating hematuria. In a meta-analysis of 3084 patients from 26 studies to determine the validity of VC by CT, MRI or US, the pooled sensitivity for bladder cancer detection using CT virtual cystoscopy, magnetic resonance virtual cystoscopy and US was 93.9%, 90.8% and 77.9% respectively. The pooled specificity for bladder cancer detection was 98.1%, 94.8% and 96.2% respectively [49]. The advantage of this technique is that it is non-invasive, making it a potential alternative for those unable to tolerate conventional cystoscopy. It also allows visualization of areas that are difficult to access such as the bladder neck and mucosa within diverticulae. The disadvantages include inability to obtain pathology and low sensitivity in identifying smaller tumors (<1 cm), flat lesions and CIS [24,46]. Although VC is unlikely at present to replace conventional cystoscopy it may be considered in conjunction allowing the possibility of minimally invasive follow-up. Figure 1illustrates appearance of tumor as seen on VC and on CT and MRI.

Figure 1. Patient with known T2 N0 M0 bladder cancer (left bladder wall): (a) contrast-enhanced computed tomography (CT) scan, (b) axial T2-weighted image performed on a 3 T magnetic resonance imaging (MRI) unit showing a hypointense lesion, (c) corresponding T1 image, (d) axial diffusion-weighted (DW) MRI at b-value 0, (e) axial DW MRI at b-value 100, (f) axial DW MRI at b-value 750, and (g) magnetic resonance virtual cystoscopy (MRVC) of same tumor with threedimensional reconstruction of tumor bed showing opening into adjacent diverticulum.

Positron emission tomography (PET) PET/CT: metabolic tracers

Composite PET/CT images provide three-dimensional whole body structural and functional information. The patient first moves through a spiral CT then a gamma camera in a single investigation. Radiotracers are used to identify the altered metabolic activity occurring within tumors. Metabolic change detectable by PET may precede anatomical changes on CT or MRI leading to greater sensitivity as compared to conventional axial imaging alone. This uptake is quantified using the standardized uptake value (SUV). Uniform radiotracer distribution throughout the body produces a SUV of 1 [24]. 18 F-Flurodeoxyglucose (18-FDG) is currently the most commonly used PET tracer in oncological imaging and has an established role in the initial staging, response assessment and recurrence detection of many cancer types [50-53]. Its use is dependent on the increase glucose metabolism occurring within the tumor. However, it cannot distinguish between increased metabolic rate occurring as a result of infection, inflammation or the normal physiological activity in some organs[24]. The use of 18-FDG-PET in staging primary bladder disease, locally recurrent and perivesical nodal disease has been difficult because the interference caused by the urinary excretion of the isotope. A number of techniques encouraging adequate washout of 18-FDG from the urinary tract have been investigated to overcome this. These include elective voiding, catheterization, bladder irrigation, and forced diuresis with intravenous frusemide prior to delayed image acquisition[54-57]. Catheterization and irrigation prior to FDG-PET imaging has a reported 40% false positive rate for detection of recurrent or residual bladder cancer [57]. These measures are invasive, making them less acceptable to patients, and continuous bladder irrigation during image acquisition increases staff exposure to radiation [58]. In those whom FDG-urine washout was encouraged by diuretic injection, oral hydration and voiding, the sensitivity and specificity for FDG-PET CT was 86.7% and 100% respectively for detecting recurrent disease within the bladder [56]. Further investigation is necessary however to evaluate the impact of radiotherapy, endoscopic intervention and intravesical chemotherapy on FDG-PET

interpretation within the bladder. When imaging is performed after chemotherapy the sensitivity decreases to 50% and therefore 18FDG PET results should be interpreted with caution following systemic treatment [59]. In a meta-analysis of the overall diagnostic accuracy of 18-FDG PET in bladder cancer, 6 studies involving 203 patients were assessed. The sensitivity and specificity of 18-FDG PET or PET/CT for staging or restaging (metastatic lesions) of bladder cancer was 82% and 89% respectively. The global measure of accuracy was 0.92 [60]. The limitations accepted by the authors include variation in the imaging technique used, one study used PET alone which meant anatomical accuracy because of the poor spatial resolution was lost, three studies were retrospective in nature and only two studies assessed detection of the primary tumor. Although there is evidence that FDG PET-CT has a diagnostic role for identifying metastatic bladder disease, in our clinical practice it is not used as principal staging modality because of the limitations discussed above. In certain circumstances, however, it provides important contributory information when CT or MRI alone raises uncertainty regarding staging. Alternative radiotracers that are dependent on cell proliferation, apoptosis, and angiogenesis, hypoxia and growth factors are also under investigation [61,62]. 11C-Choline and 11C-methionine are not excreted in the urine and may have role in future imaging of bladder cancer [24,54,63-65]. There is however limited data at present to support routine clinical use. Choline is an essential component of cell membranes. Malignant tumors have a high turnover of cellular membranes representing their increased proliferation rate [66]. The normal bladder has low uptake with 11C-choline [63]. In the preoperative staging of 18 patients, 11C-choline was highly positive for primary and metastatic bladder cancer. Uptake was seen in all primary transitional cell carcinomas (mean SUV 7.33.2 SD). In six patients, 11C-choline uptake was seen in lymph nodes as small as 5 mm; of those, four proceeded to surgery and three had pathological conformation of nodal disease [64]. 11 C-Choline has also been used to detect residual disease after TURBT. In a prospective study of 27 patients prior to radical surgery, 11C-choline PET was comparable to CT alone for detecting residual cancer after TURBT but appeared to be superior for detecting nodal involvement, with reported sensitivity and specificity of 62.5% and 100% versus 50% and 68.4% for contrastenhanced CT alone [65]. 11 C-Choline has a short half-life of approximately 20 minutes. Therefore clinical use is restricted predominantly to those centers with an on-site cyclotron. 18F-choline analogs with greater half-lives have been developed to overcome this; however, significant urinary excretion occurs as compared to 11C-choline [67]. This represents a disadvantage for pelvic imaging as previously discussed unless adequate urinary washout can be encouraged. 11 C-Methionine is a radiolabeled amino acid and is a potential tracer for visualizing protein metabolism, cellular proliferation and amino acid transport. Compared to 18-FDG PET in identifying primary tumors within the bladder, uptake is proportional to tumor stage, with a reported sensitivity of 78% in tumors greater than 1 cm but its value in local staging is not superior to conventional imaging [54]. 18 F Fluoride is a bone-seeking radiopharmaceutical that accumulates at sites of increased bone formation reflecting increased osteoblastic activity occurring within metastases. It has been shown to have increased diagnostic accuracy as compared to technetium-99 m-methylene diphosphonate (99mTc-MDP) planar or single photon emission computed tomography (SPECT) in other solid tumors [68,69]. Non-FDG tracers are not in widespread clinical use partly because of the lack of robust evidence supporting clinical benefit but also because of their cost and limited availability.

PET/CT: receptor specific radiopharmaceuticals

Imaging biomarkers using PET-CT and radioimmunotherapy opens the possibility of an individualized therapeutic and imaging approach. The rationale is that a tumor specific target is combined with a therapeutic radioactive agent. The selective accumulation within the target tissue can then be visualized on PET. These imaging techniques have the potential to permit an image and t reat approach by allowing tumor staging, estimation of radiation dose distribution prior to therapy, and early monitoring of treatment response [70]. Commonly used nuclides in other tumors types include emitters such as 131I and 90Y. They are attached to somatostatin receptor binding agents such as 90Y-DOTA-d-Phe(1)-Tyr(3)-octreotide (90Y-DOTATOC) for the treatment of neuroendocrine tumors and to antibodies in 131I-tositumomab (Bexxar), 90Y-rituximab (Zevalin) to target the CD20 antigen on B cells for the treatment of lymphoma [70,71]. Overexpression and amplification of epidermal growth factor receptor (EGFR) (HER1 or ErdB1) and, or the HER2 gene is found in bladder cancers [72,73]. It therefore represents a potential target for both molecular imaging and therapy in those with known HER2positive disease [61]. Monoclonal antibodies for example, trastuzumab labeled with 18F, allows in vivo monitoring of HER2 expression by PET as well as assessing change in HER2 expression with therapy [74,75]. Trastuzumab has also been labeled with nucleotides suitable for therapy with the future possibility of treating metastatic disease and improving the outcome of those with HER2 bladder cancer[70,76,77].

PET/MRI
Image acquisition with PET/CT occurs sequentially rather than simultaneously. This means there is loss of temporal correlation, and additionally from the patients perspective scanning time is significantly longer. The possibility of assessing different fun ctional parameters using PET, DW MRI and combining that data with high-resolution anatomical information from MRI may provides new opportunities to study pathological and biochemical processes in vivo[78-81]. Most patients in the UK will present to urology teams via a one-stop hematuria clinic, where a two-dimensional US, urine cytology and flexible cystoscopy will be carried out. On the basis of these results a TURBT will be performed if appropriate. A CT scan will be undertaken where there is evidence of MIBC, high-grade NMIBC or suspicion of an upper tract lesion. However, in view of the evidence supporting MRI (as discussed above) our preferred practice is to stage local disease using this modality. A summary of imaging modalities and their current clinical role in staging known MIBC is presented in Table 1 and Figure 2. Table 1. A summary of imaging modalities and their current clinical role(s) in staging known muscle-invasive bladder cancer

Figure 2. Anticipated clinical pathway for staging of confirmed muscle invasive bladder cancer. *Diagnostic investigations for haematuria differ from imaging to determine extent of local and distant disease in confirmed muscle invasive bladder cancer. **Based on European Association of Urology guidelines 2012, available athttp://www.uroweb.org/guidelines/onlineguidelines/ webcite andhttp://www.nccn.org/professionals/physician_gls/f_guidelines.asp#site. webcite

Patient perspectives
Image acquisition time and tolerability of any proposed scan is important. CT image acquisition is usually within minutes but some research MRI protocols may take up to 1 h. This can impact on patient compliance and predispose results to motion artifacts. Use of MRI and US are free from ionizing radiation as compared to CT and PET/CT. At our institution a 64-slice multidetector CT scan of the chest, abdomen and pelvis is associated with radiation exposure of 16 mSv during imaging; PET/CT is associated with exposure of 14 mSv (PET component 8 mSv; 6 mSv from a rapid image acquisition CT). The clinical significance of these values in terms of inducing second malignancy is small in the context of the overall poor prognosis from muscle-invasive and metastatic bladder cancer at present. MRI scanning also does not require the use of iodine contrast agents that can induce reactions potentially anaphylactic in some individuals. In addition to allergy, intravenous contrast is also omitted for CT scanning in the presence of significant renal impairment. Technology is rapidly changing so the ability to use this information to identify the most effective intervention for patients is critical. In the future we anticipate it will be routine to tailor a patients treatment plan to both the physiological and p hysical characteristics of their disease, to monitor effectiveness of the intervention allowing a more dynamic approach to treatment.

Conclusions
Accurate staging is important in determining prognostic information and identifying appropriate treatment options. CT and MRI remain central to bladder cancer staging. The role of PET-CT using current tracers in staging and guiding management of bladder cancer remains to be defined. Future developments in functional imaging are likely to be important in predicting treatment response allowing timely identification of non-responders to guide appropriate change in treatment but further studies are required to determine which techniques or combination of techniques will optimize patient care.

Competing interests
The authors declare that they have no competing interests.

Authors contributions

SH drafted the manuscript. SH and RH