Article Text

PDF

Understanding the role of mitochondria in the pathogenesis of chronic pain
  1. Bing-dong Sui1,2,
  2. Tian-qi Xu2,
  3. Jie-wei Liu2,
  4. Wei Wei2,
  5. Chen-xi Zheng2,
  6. Bao-lin Guo2,
  7. Ya-yun Wang1,2,
  8. Yan-ling Yang3
  1. 1Department of Human Anatomy, Histology and Embryology, K.K. Leung Brain Research Centre, School of Basic Medicine, The Fourth Military Medical University, Xi'an, China
  2. 2School of Stomatology, The Fourth Military Medical University, Xi'an, China
  3. 3Department of Hepatobiliary Surgery, Xi-Jing Hospital, The Fourth Military Medical University, Xi'an, China
  1. Correspondence to Dr Ya-yun Wang, Changle West Road 169, Xi'an 710032, China; wangyy{at}fmmu.edu.cn, or Dr Yan-ling Yang, Department of Hepatobiliary Surgery, Xi-Jing Hospital, The Fourth Military Medical University, Xi'an 710032, China; yangyanl{at}fmmu.edu.cn

Abstract

Chronic pain is a major public health problem. Mitochondria play important roles in a myriad of cellular processes and mitochondrial dysfunction has been implicated in multiple neurological disorders. This review aims to provide an insight into advances in understanding of the role of mitochondrial dysfunction in the pathogenesis of chronic pain. The results show that the five major mitochondrial functions (the mitochondrial energy generating system, reactive oxygen species generation, mitochondrial permeability transition pore, apoptotic pathways and intracellular calcium mobilisation) may play critical roles in neuropathic and inflammatory pain. Therefore, protecting mitochondrial function would be a promising strategy to alleviate or prevent chronic pain states. Related chronic inflammatory and neuropathic pain models, as well as the spectral characteristics of current fluorescent probes to detect mitochondria in pain studies, are also discussed.

  • Chemical Pathology
  • Neurophysiology

Statistics from Altmetric.com

 Introduction

Mitochondria play important roles in a myriad of cellular processes including ATP production via oxidative phosphorylation, biosynthetic pathways, cellular redox homeostasis, ion homeostasis, oxygen sensing, signalling, and regulation of programmed cell death. Since their discovery by Altmann1 in 1894 and the first documented mitochondrial disease by Luft in 1962,2 mitochondrial dysfunction has been implicated in neurological disorders of the central and peripheral nervous systems.3

Chronic pain is defined by the International Association for the Study of Pain as that which persists after injury when healing has ceased.4 Chronic pain is a major public health problem, with epidemiological studies reporting that about a fifth of the general population is affected in the USA and Europe.5 Chronic pain is generally classified into two broad categories: inflammatory pain and neuropathic pain. Reduced productivity, compensation costs, and treatment of comorbid conditions related to chronic pain contribute to the substantial financial burden of chronic pain management worldwide. Hence, elucidating the mechanisms of chronic pain is an urgent task.

Mitochondrial dysfunction has recently been linked to the aetiology of pain. Concurrently, protecting mitochondrial function has been suggested as a promising therapeutic strategy to alleviate or prevent chronic pain states.6 ,7 This review shows that the five major mitochondrial functions (ie, mitochondrial energy generating system, reactive oxygen species generation, mitochondrial permeability transition pore, apoptotic pathways and intracellular calcium mobilisation) have been implicated in the development and maintenance of chronic pain. All chronic inflammatory and neuropathic pain models cited in the present review, as well as the spectral characteristics of current fluorescent probes to detect mitochondria in pain, are discussed. This review aims to provide an insight into advances in understanding of the role of mitochondrial dysfunction in the pathogenesis of chronic pain.

Chronic pain models

Chronic pain is classified into inflammatory pain and neuropathic pain. Inflammatory pain refers to pain caused by tissue damage or inflammation, while neuropathic pain refers to pain caused by damage to the peripheral or central nervous system.8 The peripheral or central neural changes that underlie chronic pain manifest themselves behaviourally in humans and animals as hyperalgesia (an exaggerated response to a stimulus that normally elicits pain) and allodynia (a pain response to a stimulus that normally does not produce the experience of pain) and often persist and outlast the initial damage to peripheral or neural tissues.9 The goal of developing therapeutics to alleviate the experience of chronic inflammatory or neuropathic pain, then, is to diminish or reverse the behavioural manifestations of hyperalgesia and/or allodynia without altering the normal and important physiological role that pain stimuli play in protecting individuals from tissue or nerve damage.

Major animal models for the study of neuropathic and inflammatory pain cited in the present review are described in online supplementary table S1. Intradermal injection of nerve growth factor (NGF) which sensitises the peripheral terminal of the nociceptor by activating the high-affinity receptor, tropomyosin receptor kinase A (TrkA), induces neuropathic pain through both thermal and mechanical hyperalgesia.10 Intravenous injection of 2′,3′-dideoxycytidine (ddC, a highly active antiretroviral pharmaceutical)11 ,12 or streptozotocin sulphate (STZ, which produces diabetes mellitus by removing pancreatic β cells and inducing insulin deficiency),12–14 drugs used for cancer chemotherapy, including oxaliplatin,15 ,16 paclitaxel17 ,18 and cisplatin,15 as well as intraperitoneal injection of the antitumoural agent vincristine sulphate (VCR),12 ,13 induce neuropathic pain in rats and mice. Spinal nerve ligation (SNL),19 ,20 spared nerve injury (SNI),21 partial sciatic nerve ligation (PSNL)22 ,23 and chronic constrictive nerve injury (CCI)13 ,24 -induced neuropathic pain have served as powerful tools that mimic the clinical symptoms of chronic neuropathic pain. Rats fed with the Lieber–DeCarli liquid diet with added ethanol11 or with chronic post-ischaemia pain (CPIP),25 experience neuropathic pain. Intradermal injection of the inflammatory mediators, such as tumour necrosis factor-α (TNFα),12 prostaglandin E2 (PGE2),12 ,14 epinephrine (Epi),12 ,14 complete Freund's adjuvant (CFA)21 and carrageenan,26–28 are useful experimental models to evoke inflammatory pain. Intradermal injection of protein kinase Cε (PKCε) activator (ψεRACK),14 capsaicin19 ,23 ,29 ,30 and formalin,13 ,21 ,31 are useful experimental models to evoke hyperalgesia. Intracolonic injection of zymosan induces visceral hyperalgesia.29 ,32 Long-term potentiation (spinal cord LTP) is usually considered to be the electrophysiological model of chronic pain.33

Mitochondrial energy generating system and chronic pain

Mitochondria play a vital role in cellular energy metabolism. Mitochondria produce more than 90% of the cellular energy by enzymatic coupling of oxidative phosphorylation through the mitochondrial energy generating system.34 The mitochondrial energy generating system encompasses all of the enzymatic reactions involved in two closely coordinated metabolic processes—the tricarboxylic acid cycle and mitochondrial electron transport chain (mETC).34 The mETC is a series of five molecular complexes through which electrons are transported (by complexes I–IV) and ATP is generated (by complex V). Dysfunction in each complex in the mETC has been identified in connection with chronic pain. ATP, the end product of the mETC, is also a target in pain studies.

The role of the mETC in neuropathic and inflammatory pain has been evaluated through the use of selective mETC I–V complex inhibitors in different pain models (see online supplementary table S2). Rotenone is considered to act specifically as an inhibitor at NADH dehydrogenase in mETC complex I. 3-Acetylphenyl N-(p-Tolyl) carbamate (3-NP) is reported to act specifically as an inhibitor at succinate dehydrogenase in mETC complex II. Antimycin selectively acts to inhibit cytochrome b-ubisemiquinone in complex III. The mETC-IV inhibitor sodium cyanide and a specific inhibitor of the ATPase (the mETC-V inhibitor), oligomycin, are also used. Intradermal injection of the inhibitors of mETC complexes I, II, III, IV and V into a hind paw significantly attenuated neuropathic pain-related behaviour in models of neuropathy induced by HIV/AIDS therapy (ddC), cancer chemotherapy (VCR and oxaliplatin) and diabetes (STZ).10–12 ,14 ,15 Moreover, the inhibitors of all five complexes have attenuated TNF-α and PKCε-induced inflammatory pain behaviours.12 ,14 Interestingly, intrathecal injection of an mETC inhibitor, antimycin A or rotenone, in normal mice resulted in a slowly developing but long-lasting and dose-dependent mechanical hyperalgesia.7 The levels of mechanical hyperalgesia after injection with antimycin A, a complex III inhibitor, were higher than those following injection with rotenone, a complex I inhibitor. In addition, the results of a mitochondrial respiration assay found significant deficits in mETC complex I and II-mediated respiration in oxaliplatin and paclitaxel-induced neuropathic pain.35 These results, taken together, support the involvement of the mETC in multiple models of neuropathic and inflammatory pain. Inhibition of multiple mETC chain complexes may have implications for the development of analgesic therapies since drugs that partially inhibit multiple steps in signalling pathways can be more effective than those that nearly inhibit one step in the pathway.12 This is especially important since several of the mETC complexes participate in other signalling mechanisms in the cell (eg, EC-II is also part of the tricarboxylic acid cycle and EC-III and IV are part of the mitochondrial-dependent apoptotic signalling pathway). Thus, the design of ‘dirty drugs’ that partially inhibit multiple mETC complexes, may prove to be safer as well as more effective for the treatment of painful conditions.

While the mETC has contributed to multiple forms of mechanical hyperalgesia, this role seems to differ from its involvement in some models of neuropathic and inflammatory pain. It is particularly noteworthy that the inhibitors of the mETC complexes I–V have no effect on mechanical hyperalgesia in some forms of chronic pain, such as cisplatin and oxaliplatin-induced neuropathic pain, as well as PGE2 and Epi-induced inflammatory pain.11 ,12 ,14 ,15 Thus, the results support the suggestion that the mETC-independent signalling pathways in primary afferent nociceptors may contribute to some forms of neuropathic and inflammatory pain.

Since the end product of the mETC is ATP and the mitochondrion is the major source of intracellular ATP, the role of ATP in neuropathic and inflammatory pain has been investigated through the use of some ATP analogue reagents. Adenosine 5′-(β,γ-imido) triphosphate (AMP-PNP) is a non-hydrolysable ATP analogue which inhibits ATP-dependent cellular mechanisms. P1,P4-di(adenosine-5′) tetraphosphate (Ap4A) is a stable nucleotide which antagonises the actions of ATP. Pentachlorophenol (PCP) is a potent uncoupler of mitochondrial phosphorylation which is known to inhibit ATP synthesis. Intradermal injection of AMP-PNP, PCP and Ap4A significantly attenuated the mechanical hyperalgesia induced by ddC, VCR and STZ-induced neuropathic pain.11 ,12 Although the antagonists of ATP-dependent mechanisms blocked pain-related behaviours in three models of neuropathic pain, they could not inhibit the nociceptive effects of TNFα, PGE2 and Epi-induced inflammatory pain.11 ,12 These findings provide further support the involvement of mETC in chronic pain mechanisms including a role for its end product, ATP. Also, we are only able to demonstrate that ATP just contributes to neuropathic pain. The mechanism downstream of the mETC (eg, an ATP modulated kinase or ion channel) remains to be elucidated. It should be mentioned that dysfunction of the mitochondrial energy generating system underlying pain is mainly associated with the mETC rather than abnormality of the tricarboxylic acid cycle, possibly because the mETC complexes are located on the inner membrane of mitochondria making the mitochondria vulnerable to toxic factors.20

Mitochondrial reactive oxygen species and chronic pain

Reactive oxygen species (ROS), for example superoxide radical O2 and hydroxyl radical OH, are by-products of oxidative phosphorylation and are usually decomposed by specialised cellular enzymes, for example superoxide dismutases and peroxidases. Under normal physiological conditions, production of ROS is balanced by several cellular antioxidant mechanisms.29 In certain conditions, however, levels of ROS rise to the point that may endanger the functional and structural integrity of cells, sometimes leading to irreversible damage.29 Recent studies indicate that ROS are also involved in persistent pain. For example, removal of excessive ROS by free radical scavengers, such as phenyl N-tert-butylnitrone (PBN) and 4-hydroxy-2, 2,6,6-tetramethylpiperidine 1-oxyl (TEMPOL), produced a significant analgesic effects in both neuropathic and inflammatory pain (see online supplementary table S2).17 ,19 ,20 ,23–32 Antinociceptive effects of other antioxidants, such as vitamins C and E and olesoxime, have also been reported in animal models of inflammatory or neuropathic pain (see online supplementary table S2).18 ,21 ,35 α-Lipoic acid, first identified as a co-enzyme in the tricarboxylic acid cycle, is also a potent antioxidant which relieved neuropathic pain behaviours.11 ,14 ,15 In contrast, intrathecal administration of a ROS donor, tert-butyl hydroperoxide (t-BOOH), produces transient pain behaviours in normal mice in a dose-dependent manner.16 ,30 ,31 ,33 Also, mitochondrial ROS-producing profiles are increased in the spinal cord following peripheral nerve injury or an inflammatory stimulus.30 ,36 Thus, these investigations into the analgesic properties of antioxidants have produced data supporting further examination of this class of compounds for use in pain treatment.

Altered second messenger signalling systems are involved in chronic pain conditions. Studies have emerged that focus on the various effects of ROS on signalling pathways. Investigations into the mechanisms through which second messenger systems like mitogen-activated protein kinases (MAPK), NF-κB and second messenger molecules, including protein kinase A and C, and nitric oxide, contribute to pain conditions and how ROS in turn affect these signalling pathways could help explain the mechanism by which oxidative stress exacerbates chronic pain conditions.20 ,23 ,24 ,26–31

There is evidence to suggest that ROS-induced pain enhancement can be the result of increased glutamate receptor activation.19 ,20 Glutamate will activate both ionotropic and metabotropic glutamate receptors, including NMDA receptors, which leads to massive Ca2+ influx, thus increasing intracellular Ca2+ levels. High levels of intracellular Ca2+ will overload mitochondria, destabilise mitochondrial membrane potentials and accelerate the mitochondrial electron transport chain resulting in increased mitochondrial ROS generation.30 These increased ROS can then affect redox sensitive signalling pathways toward sensitisation of neurons. Another possible mechanism underlying the involvement of ROS in pain is through heightened neuronal excitability of the spinal dorsal horn by reducing spinal GABA neurotransmission.16

Mitochondrial permeability transition pore and chronic pain

The mitochondrial permeability transition pore (MPTP) plays an important role in cell function since it opens, leading to mitochondrial swelling and release of cytochrome c, which initiates apoptosis. By inhibiting the opening of MPTP, cyclosporine A may contribute to maintaining mitochondrial homeostasis. Previous results have shown that there has been a significant increase in the ratio of cytochrome c at the spinal cord level in PSNL-induced neuropathic pain and cyclosporine A has attenuated the decrease in mechanical threshold and the increase in the ratio of cytochrome c. These findings suggest that the release of cytochrome c from mitochondria is the result of PSNL.

Cholest-4-en-3-one, oxime (TRO19622) has been shown to bind to two outer mitochondrial membrane proteins, the peripheral benzodiazepine receptor (also called translocator protein (18 kDa)) and the voltage-dependent anion channel.13 It has been reported that daily oral administration of TRO19622 can reverse neuropathic pain behaviour in STZ and VCR-induced neuropathic pain.13 It is interesting to note that TRO19622 did not have analgesic activity in animal models of pain produced by formalin injection, noxious thermal or mechanical stimulation, or CCI-induced neuropathic pain, indicating that painful diabetic or chemotherapy-induced neuropathies share a common mechanism that is distinct from acute, inflammation driven or lesion-induced neuropathic pain. These results support the potential use of TRO19622 to treat painful diabetic and chemotherapy-induced neuropathies. TRO19622-related mitochondrial proteins are part of the MPTP. This result suggests that abnormality in axonal mitochondria directly contributes to chemotherapy-induced pain.

The mitochondrial apoptotic pathways and chronic pain

Whereas small-fibre sensory neuropathies might ultimately lead to cell death and loss of sensation, they first progress through a phase, which might last for years, characterised by the presence of analgesia-resistant neuropathic dysaesthesias and pain. Much previous research has addressed the fact that activity in signalling pathways that ultimately leads to apoptosis plays a critical role in the generation of neuropathic pain, before the death of sensory neurons becomes apparent. In two models of painful peripheral neuropathy, neuropathic pain induced by ddC and VCR which inhibit activator (1, 2, 8 and 9) and effector (3) caspases, attenuates neuropathic pain-related behaviour, but has no effect in STZ-diabetic neuropathy and control rats.11 The non-specific caspase inhibitor Z-Val-Ala-Asp (OMe)-fluoromethyl ester (Z-VAD) inhibited ddC hyperalgesia and the second but not the first plateau in oxaliplatin hyperalgesia.15 ,37 Therefore, it seems that during a latent phase, before apoptotic cell death is manifest, the caspase signalling pathway can contribute to pain in small-fibre peripheral neuropathies, and that inflammatory/immune mediators also activate these pathways. This suggests that these pathways are potential targets for novel pharmacological agents for the treatment of inflammatory as well as neuropathic pain. However, previous results have not detected a significant contribution of caspase signalling to PKCε, PGE2 or Epi-induced inflammatory pain, which implies that caspase signalling may play different roles in neuropathic and inflammatory pain.14

Mitochondrial calcium homeostasis and chronic pain

It has been suggested that mitochondria play a key role in intracellular Ca2+ homeostasis and affect membrane excitotoxity, increase neurotransmitter release and impair synaptic plasticity, all of which could contribute to the aetiology of pain. The mitochondrial Ca2+ uniporter is considered a major pathway of Ca2+ influx into mitochondria, while Ca2+ is released by mitochondrial Na+/Ca2+ exchange. It should be noted that there is only indirect evidence of the contribution of mitochondrial calcium homeostasis to chronic pain. The vanilloid receptor TRPV1 is a critical molecular detector of noxious signals in primary sensory neurons and is either directly activated or modulated by a broad range of pain-producing stimuli.38 These includes noxious heat, protons, lipid-derived endovanilloids and inflammatory mediators. Electrophysiology investigations combined with measurement of mitochondrial Ca2+ concentrations ([Ca2+]mt) and cytoplasmic Ca2+ concentrations ([Ca2+]c) in axonal boutons have shown that brief stimulation of the receptor using either capsaicin or the endogenous TRPV1 agonist can induce prolonged elevation of presynaptic [Ca2+]c and a concomitant enhancement of glutamate release at sensory synapses. Initiation of this response requires Ca2+ entry, primarily via TRPV1. The sustained phase of the response elicits prolonged Ca2+ elevation in presynaptic mitochondria. The concentration of the TRPV1 agonist determines the duration of [Ca2+]mt and [Ca2+]c signals in presynaptic boutons and, consequently, the period of enhanced glutamate release and action potential firing by postsynaptic neurons.38 These data suggest that mitochondria control vanilloid-induced neurotransmission by translating the strength of presynaptic TRPV1 stimulation into duration of the postsynaptic response.

Current fluorescent probes to detect mitochondria in pain studies

To date, the main method for studying pain-related mitochondrial ROS is by using mitochondrial ROS-sensitive fluorescent probes (see online supplementary table S3), such as MitoSOX, Mitotracker Red (MT-Red), dichlorodihydrofluorescein (DCFH), dihydrorhodamine 123 (DHR 123) and hydroethidine (HET). MitoSOX and MT-Red are the most commonly used probes. MitoSOX is a ROS specific dye whose reduced form does not show fluorescence until it penetrates into cells and is sequestered in the mitochondria.39

Direct monitoring of [Ca2+]mt and [Ca2+]c in intact cells has become feasible following the recent development of a series of green or luminescent protein-based fluorescent probes.38 [Ca2+]c was measured using Fluo-3/AM, Fura-2/AM and indo-1/AM. Because buffering of Ca2+ by the endoplasmic reticulum can mask the accumulation of [Ca2+]mt, it is necessary to use 3,4,5-trimethoxybenzoic acid 8-(diethylamino)octyl ester (TMB-8), an intracellular mobilisation blocker, to prevent Ca2+ release from the endoplasmic reticulum.

Depolarisation of the mitochondrial membrane potential (MMP) is an important marker for mitochondrial function. Cationic lipophilic fluorescent probes (see online supplementary table S3) may be used for detecting the MMP, as they can easily permeate through biological phospholipid bilayers, and accumulate in the mitochondrial matrix. These fluorochromes include but are not limited to rhodamine 123 (Rh 123), tetramethylrhodamine ethyl and methyl esters (TMRE and TMRM), chloromethyl-X-rosamine (CMXRos, also known as MT-Red) and 5,5′,6,6′-tetrachloro-1,1′,3, 3′-tetraethylbenzymidazolyl carbocyanine iodide (JC-1). Rh 123 was the first probe to be discovered and widely used, but is not recommended today because it quenches at high concentrations and is toxic in certain experimental conditions.40 Its derivatives (ie, TMRE and TMRM) perform better at relatively low concentration but they still show partial quenching at high concentrations and are oversensitive to pH in mitochondria.41 JC-1 is considered the most satisfactory probe to date because of its higher sensitivity and resolution at both high and low concentrations. JC-1 is more suitable for quantitative and qualitative analysis with a flow cytometer or a fluorescence confocal microscope expressed by the ratio of red (590 nm) to green (530 nm).42 ,43

Conclusions and perspectives

In conclusion, previous studies have shown that the five major mitochondrial functions play an important role in chronic inflammatory and neuropathic pain. However, major questions remain regarding which targets downstream of mitochondrial dysfunction play a role in enhanced nociceptive neurotransmission, through which mechanisms the mitochondria influence chronic pain, and how interactions between different mitochondrial functions contribute to pain. In the future, protecting mitochondrial function would be an effective therapeutic approach for patients with chronic pain.

Current research questions

  • What targets downstream of mitochondrial dysfunction play a role in pain regulation?

  • Through what mechanisms do mitochondria influence chronic pain?

  • How do interactions between different mitochondrial functions contribute to chronic pain?

Main messages

  • Mitochondrial dysfunction contributes to the aetiology of pain.

  • It is suggested that the mitochondrial energy generating system and mitochondrial reactive oxygen species play the most important roles in the pathogenesis of chronic pain.

  • Protecting mitochondrial function would be a promising therapeutic strategy to alleviate or prevent chronic pain states.

Key references

  • Ferrari LF, Levine JD. Alcohol consumption enhances antiretroviral painful peripheral neuropathy by mitochondrial mechanisms. Eur J Neurosci 2010;32:811–18.

  • Joseph EK, Levine JD. Mitochondrial electron transport in models of neuropathic and inflammatory pain. Pain 2006;121:105–14.

  • Kim HK, Park SK, Zhou JL, et al. Reactive oxygen species (ROS) play an important role in a rat model of neuropathic pain. Pain 2004;111:116–24.

  • Joseph EK, Levine JD. Caspase signalling in neuropathic and inflammatory pain in the rat. Eur J Neurosci 2004;20:2896–902.

  • Shin CY, Shin J, Kim BM, et al. Essential role of mitochondrial permeability transition in vanilloid receptor 1-dependent cell death of sensory neurones. Mol Cell Neurosci 2003;24:57–68.

Self assessment questions

  1. To date, which four major critical functions of mitochondria in neurons have been found to participate in hyperalgesia and/or allodynia in pain studies?

    1. Energy generation

    2. Metabolism of reactive oxygen species (ROS)

    3. Calcium homeostasis

    4. Apoptotic pathways

    5. Autophagy

  2. Which of the following fluorescent probes can be used to detect mitochondrial function in pain studies?

    1. MT-Red

    2. MitoSOX

    3. Rh 123

    4. TMRE

    5. JC-1

  3. Which of the following fluorescent probes can be used to detect pain-related mitochondrial ROS?

    1. MT-Red

    2. MitoSOX

    3. Rh 123

    4. TMRE

    5. JC-1

  4. Which of the following reagents related to mitochondrial dysfunction play a role in down-regulation in hyperalgesia and/or allodynia in pain studies?

    1. Antagonists of ATP-dependent mechanisms

    2. mETC inhibitor

    3. ROS scavenger and antioxidant

    4. ROS donor

    5. Mitochondrial permeability transition pore inhibitor

  5. Which of the following reagents related to mitochondrial dysfunction has been applied in pain studies?

    1. Intradermal injection into a hind paw

    2. Intrathecal injection

    3. Intravenous injection

    4. Intraperitoneal injection

    5. Intracerebroventricular injection.

Answers

  1. A, B, C and D

  2. A, B, C, D and E

  3. A and B

  4. A, B, C and E

  5. A, B, C, D and E

Acknowledgments

This work was supported by grants from the National Natural Science Foundation of China (nos. 81272555, 81070327, 81071895 and 30971174) and the Excellent Project of the Fourth Military Medical University (to YYW).

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors BDS, TQX, JWL and WW collected the data, analysed the results and drafted the manuscript. CXZ and BLG participated in analysis and discussion. YYW and YLY conceived of the review, participated in the design of the study and verified the submission. All authors read and approved the final manuscript.

  • Funding This work was supported by grants from the National Natural Science Foundation of China (grant numbers 81272555, 81070327, 81071895 and 30971174) and the Excellent Project of the Fourth Military Medical University (to YYW).

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.