Applied information retrieval and multidisciplinary research: new mechanistic hypotheses in complex regional pain syndrome.

Background

Early in the history of western medicine, the physician and the basic scientist were one and the same person. However, over the past century, clinical research developed as a separated branch from the basic sciences such as biology, molecular biology, biochemistry, and physiology. The main objective of clinical research is the collection and analysis of clinical data concerning symptoms of the disorders and responses to treatments. Based on these observations new theories about etiology and pathogenesis can be developed. However, detailed information regarding the molecular mechanisms underlying a certain disease process often remains elusive. One avenue into possible mechanisms of complex disorders is through the use of bioinformatics. Bioinformatics applies informatics techniques to organize bio-molecular data on a large scale [ 1 ].

The combination of bioinformatics and biomedical approaches is expected to result in significant advantages in both understanding mechanisms of disorders and individual susceptibility, which in turn will open many possibilities in individualized medical health care [ 2 ]. INFOBIOMED is a Network of Excellence funded by the European Union that aims at enforcing European biomedical informatics as an integrative discipline [ 3 , 4 ]. One of the main objectives of INFOBIOMED is to enable pilot applications in several medical fields that demonstrate the benefits of a synergetic approach in biomedical informatics. An example of such a multidisciplinary project is the use of bioinformatics tools in the investigation of the relationship between clinical and molecular data. Of course, the current literature contains information from these different domains. However, the amount of information has become so large that it is very difficult for a single individual to draw conclusions across the various disciplines. Literature based discovery support tools have been developed to bridge these interdisciplinary gaps, and novel scientific hypotheses have been generated and tested [ 5 , 6 ]. This approach, wherein the clinician and the basic scientist collaborate, should be beneficial in the investigation of complex disorders, where clinical research alone is not sufficient to unravel the entire disorder process.

Case studies can be useful in exploring new ways to advance multidisciplinary biomedical research. The Complex Regional Pain Syndrome (CRPS) is an example of a complex disorder from which the etiology and pathogenesis remain unelucidated for a considerable part, despite intensive research in the medical field. For this reason CRPS was chosen as a case study on how text mining techniques could be used in multidisciplinary biomedical focused research. The results should not be regarded as answers to the long unsolved questions regarding CRPS, but rather as hypothetically new insights in the molecular mechanisms underlying the disorder. The main purpose of this exercise was to assess the benefit of a new approach on hypothesis discovery, based on the use of text mining tools by a multidisciplinary team of researchers.

A brief introduction will be provided on the disorder CRPS in the next section, including a short description of the current theories about its pathogenesis.

The Complex Regional Pain Syndrome

CRPS is a painful syndrome affecting one or more extremities of the body, marked by a wide variety of symptoms. The most prominent feature is pain, including spontaneous pain, allodynia, hyperpathia, and hyperalgesia. Additionally, the affected extremity can display changes in color and/or temperature (vasomotor disturbances), edema, alterations in transpiration, hair and nail growth (sudomotor disturbances), and muscular atrophy and/or dysfunction (motortrophic disturbances) [ 7 , 8 ]. It is usually described after a specific initiating event, in most cases a trauma or an operation, but sporadically it is observed after a stroke, myocardial infarction, infection or even without an obvious inciting event in a rarity of the cases [ 9 ]. The course of the disorder varies from patient to patient, but often ends in diminished function of the affected limb which impacts the quality of life of the patient. In rare cases, the disorder progresses to the point where amputation is necessary.

The pathogenesis of CRPS evolves from disturbances in both the peripheral nervous system (PNS) and the central nervous system (CNS) (figure 1 ). Regarding the initial phase of the disorder, recently the interest has increased towards the role of inflammatory responses. Inflammatory signs such as swelling, redness, warmth and pain are common features in the early stage of CRPS. Classic inflammation is marked by the presence of proinflammatory cytokines and in CRPS a local increase of the cytokines TNFα, IL-6 and tryptase (a product of mast cell degranulation) was observed in blister fluid derived from the affected extremity [ 10 , 11 ]. IL-1 and IL-6 were also found to be increased in spinal fluid [ 12 ]. Additional to classic inflammation, a process called neurogenic inflammation has been demonstrated in CRPS [ 13 - 17 ]. Neurogenic inflammation resembles classic inflammation, but it is initiated by neuropeptides instead of lymphocytes and cytokines [ 18 ]. Those neuropeptides include Substance P (SP), Calcitonin Gene Related Protein (CGRP), neuropeptide Y (NPY), Bradykinin (BK) and Vasoactive Intestinal Protein (VIP). Important modulators of neurogenic inflammation are Neutral Endopeptidase (NEP) and Angiotensin Converting Enzyme (ACE) [ 19 ].

Endothelial dysfunction, hypoxic changes, and free radical damage have also been suggested as important processes in the pathogenesis of CRPS [ 20 - 24 ]. Ischemia, together with inflammation, can result in the formation of free radicals, lactate acidosis and altered nitric oxide (NO) synthesis. The NO metabolism in its turn has an effect on the microcirculation and thus can influence peripheral oxygen supply [ 25 ]. An impaired microcirculation might underlie a cold extremity that is observed in some cases of CRPS.

In time, the peripheral alterations give rise to disturbances at the level of the CNS, and the clinical picture of CRPS evolves more towards that of a neuropathic pain syndrome. The mechanisms behind the development of neuropathic pain have been investigated extensively in animal models [ 26 , 27 ]. The pathogenesis of neuropathic pain is marked by a phenomenon called neural plasticity, which is described by Woolf and Salter as the capacity of central and peripheral neurons to change their function, chemical profile, and their structure in reaction to activation of peripheral afferent nerve endings [ 28 ]. This conducts towards a state of hyperexcitability of the peripheral C- and A-fiber transducers, and of the neurons in the dorsal root ganglia (DRG) of the CNS, referred to as peripheral and central sensitization respectively. Sensitization results in a painful response to a normally not painful stimulus, leading to features as allodynia and hyperalgesia. A prominent factor in the evolvement of sensitization is the interaction between glutamate and the N-methyl-D-aspartate (NMDA) receptor, from which functional status Mg2+ is a modulator [ 28 ]. Following central sensitization, alterations at the supraspinal level might evolve, resulting in an altered sensory mapping in the cerebral cortex. This pays an additional contribution to the sensational disturbances in chronic CRPS [ 29 ].

Dysregulation of the sympathetic nerve system was classically supposed a main feature of CRPS. This was based on the observation that pain relief could be obtained by performing a sympathectomy, although only in a subset of CRPS patients. Vasomotor disturbances (warm or cold limb) [ 30 ] and sympathetically maintained pain (SMP) [ 31 - 34 ] are features that have been ascribed to sympathetic dysregulation. The painful responses may result from the expression of α-adrenoreceptors on sensory fibers [ 34 ]. Due to sympathetic sprouting new communication pathways are formed between sympathetic terminals and sensory neurons [ 35 ]. Additionally, the α-adrenoreceptors might have developed a hypersensitivity for normal stimulation by cathecholamines [ 30 ].

The authors have decided in this paper to focus mainly on the peripheral processes in the initial phase of CRPS. The first reason for this is that it is very feasible that the neuropathic pain component of CRPS does not form the initial pathogenesis of the disorder, but that it is preceded and sustained by the presence of the peripheral inflammatory and hypoxic reactions in the affected extremity [ 35 ], The second reason is that, until now biomolecular research on CRPS in humans and animals concerned mostly the peripheral inflammatory aspect of the disorder. Sensitization and neuronal plasticity have been studied broadly in models for neuropathic pain in general. However, they have never been demonstrated in models for CRPS in particular, although is it highly reasonable to assume that they occur in CRPS in a similar manner.

Methods

The INFOBIOMED Network of Excellence is organized into various work packages with different aims [ 3 ]. The aim of the so called "pilot applications" is to analyse the impact of biomedical informatics in specific fields (Pharmainformatics, Genomics and Microbiology, Genomics and Chronic Inflammation, and Genomics and Cancer). The team of researchers behind the study outlined in this paper was part of the pilot application Pharmainformatics, which aims at assessing the mutual impact of BMI and pharmaceutical research. Research in this area focuses on establishing the information continuum pathology – pathway – target – ligand.

The pathogenesis of CRPS was one of the subjects that were chosen for a case study. The purpose was to investigate how to gain further insight into pathogenesis behind a complex disease and to identify possible pathways, targets and ligands for improving pharmacological therapy, using a multidisciplinary approach. The team of researchers had varying backgrounds mirroring the biomedical informatics research area (one physician with domain expertise on CRPS that was asked to participate in the work package only for this specific case study; one bioinformatician; one specialist in text mining). No formal leader was chosen for the group. All researches provided input based on their background and a plan of action took form by mutual agreement. The physician provided biological and medical concepts that are possibly linked to CRPS, the bioinformatician provided the appropriate software tool to be used in the study, and the text mining expert provided knowledge on how to best perform the extensive literature analysis. The analysis took place in September 2004, at the premises of AstraZeneca in Molndal, Sweden.

The selection of the CRPS related concepts was based on current (but not always objectified) opinions about the pathogenesis and treatment of CRPS and on different pathogenic mechanisms described in a selection of articles concerning the pathogenesis of CRPS [ 33 , 36 - 39 ]. The focus was on chemical and biochemical identities and mainly, but not completely, on the peripheral components of the disorder. The collection of concepts and their synonyms can be found in Table 1 .

The text mining/entity relationship tool PathwayAssist™ (Version 2.5) was used to visualize the connections between the CRPS concepts in a network and to search for new concepts taking part within these relations. PathwayAssist is a software application developed for navigation and analysis of biological pathways, gene regulation networks and protein interaction maps. It has been used before in studies to identify genes that are involved in autism [ 40 ] and in regulation of human primordial follicle development [ 41 ]. The application has been described in more detail by Nikitin and colleagues [ 42 ]. PathwayAssist finds connections between concepts, henceforth referred to as 'nodes', by searching through a database of interactions derived from literature, using the natural language processing (NLP) based software MedScan. MedScan performs a grammatical and semantic analysis of the complete MEDLINE ® database of life sciences and biomedical bibliographic information. The MedScan software has been described by Novichkova et al. [ 43 ] and more recently by Daraselia et al. [ 44 ]. A brief overview of the technology is presented below.

When parsing a MEDLINE document, a semantic interpreter of the NLP component transforms the syntactic structure into a semantic structure. The syntactic structure and main constituents (surrounded by square brackets) of a sentence can be exemplified using the general sentence Protein X inhibits protein Y , which has the syntactic structure of [ Protein X [ N ] ] [ NP ] [ inhibits [ V ] Protein Y [ N ] ] [ VP ] ] [ S ] . The phrasal category is shown immediately following each constituent ( NP designates noun phrase, N designates noun, V designates verb, VP designates verbal phrase, S designates sentence). The syntactic structure will be transformed to a semantic frame of inhibition that has an 'agent' protein X and a 'patient' protein Y. The output of the semantic parse is the input for an ontological analysis that was developed by Daraselia et al. [ 44 ]. In this an 'entity' is represented either as a protein, a cellular object, a cellular process, or a small molecule and 'controls' describe functional relationships between these entities. Relations between entities are stored in a relational database. These relations can be displayed and explored through a graphical interface.

The list of concepts provided by the physician was regarded as current knowledge about CRPS (table 1 ) and was used as input for PathwayAssist. PathwayAssist views these concepts as nodes. The option "find only direct interactions between selected nodes" was used to search the underlying interaction database for direct connections between the CRPS concepts. Based on these connections, the system builds a network. This procedure would only detect direct relationships between two nodes; mechanisms that require a new intermediate node (a node not present in the original list of nodes that was provided as input) would not be detected. Such intermediate nodes, however, might represent knowledge that the scientists entering the concepts were not aware of at the time of input (such as factors acting as enhancers for a critical step in a pathway, or co-factors needed for a transcription factor to bind to DNA). Therefore, an option was used that incorporates a new node in the network if that node is the intermediate node that allows the creation of a triplet connecting two input nodes. When a new node is added to the network, all links to and from that node to all other nodes in the network are displayed.

Results

The resulting network based on the imported CRPS concepts is shown in Figure 2 . In addition to the original concepts that were provided as input nodes, six new nodes have been added by the algorithm (nodes that were not in the original list of concepts known to be associated with CRPS). Of these six new nodes, the node NFκB was the one connecting to most of the original CRPS concepts and was pivotal in the final network. NFκB appeared as new node because it was part of a triple connection between mannitol and TNFα with the actual sentence being 'High glucose or mannitol also enhanced TNFalpha-stimulated NF-kappaB activity': TNFα stimulates NFκB, and mannitol acts as an enhancer of this process. The sentence describing this particular relationship between mannitol, TNFα, and NFκB had been parsed from the article by Hattori et al [ 45 ].

The transcription factor NFκB is known as a mediator in many different physiological processes [ 46 , 47 ]. It is also related to 7% of all the nodes that exist in PathwayAssist, which mirrors its versatility. However, even though NFκB has a very high connectivity in PathwayAssist and is involved in many physiological processes, the calculated p-value using a one-tailed Chi-square test with Yates's correction was highly statistically significant (p-value less than 0.0001). This demonstrates the significance of NFκB in the CRPS network in comparison to the whole PathwayAssist database of interactions.

To investigate the possibility that NFκB appeared in the network simply because it is a highly connected node in the database, the in PathwayAssist equally highly connected node "protein tumor protein p53" was manually included into the network and tested for its connectivity with the CRPS concepts. The p-value was found to be equally significant as for NFκB (p-value less than 0.0001). The reason for TP53 not appearing in the expanded CRPS network is related to on the algorithm that was chosen to find new nodes. In order for a new node to be incorporated into the network by PathwayAssist it has to be part of a triplet that connects two input nodes. This was the case for NFκB but not for TP53. There are also other algorithms available in PathwayAssist for incorporating new nodes in a network. These include an algorithm to find the shortest paths between the nodes in the network, an algorithm to find common targets, an algorithm to find common regulators, and also an option to expand the network by finding all nodes in the database that are connected to the nodes in the network. The shortest paths algorithm yielded 20 new nodes, the common target algorithm 46 new nodes and the common regulators algorithm yielded 38 new nodes. The option of finding all nodes connected to the nodes in the CRPS network (the expanded network algorithm) yielded 3872 nodes and it was considered to be practically infeasible to analyze all these nodes separately for their connectivity in the CRPS network. Neither NFκB nor TP53 were found by using any of the other algorithms except for the expand network algorithm. Thus, it might be possible that NFκB emerged in the CRPS network simply because it is a highly connected node in the whole PathwayAssist database. However, NFκB was still connected to more nodes in the CRPS network than the reference node TP53. Therefore it was still regarded as a candidate for a key role in the pathogenesis of CRPS

As far as the authors of this paper know, NFκB has never before been mentioned together with CRPS. For this reason NFκB was considered interesting enough to be explored further by manual research of literature, using advanced PubMed queries concerning the relation between NFκB and the CRPS concepts. Based on the results of this literature research, and on discussions about this topic within the Pharmainformatics group of the INFOBIOMED network of excellence, theories for the involvement of NFκB in the described pathogenic mechanisms in CRPS were developed. A summary of these theories is described table 2 and further explained below.

Role of NFκB in CRPS related mechanisms

1) NFκB involvement in neurogenic and classic inflammation

NFκB has been demonstrated to be an essential transcription factor in the mediation of the effects of neuropeptides that are also involved in CRPS: 1) SP induced expression of IL-6 and TNFα (cytokines that are locally increased in CRPS) is regulated by NFκB [ 48 , 49 ]; 2) Glutamate, released in the dorsal horn together with SP induces neuronal apoptosis through NFκB induction [ 50 ]; 3) Expression of CGRP (an important neuropeptide in CRPS) is induced by an NFκB mediated pathway, initiated by IL-1β stimulation [ 12 ]; 4) Bradykinin activates NFκB and cyclooxygenase-2 via more than one pathway [ 46 , 51 ]; 5) VIP inhibits NFκB mediated chemokine production by macrophages [ 52 ]. It also prevents NFκB from binding to the promotor site for nitric oxide synthethase iNOS, thereby affecting the microcirculation in CRPS [ 53 ]; 6) For the NPY receptor, potential NFκB binding sites have been found in the promoter regions, suggesting a role for NFκB in the expressional regulation of this receptor [ 54 ]; 7) The important inflammatory modulator ACE reduces NFκB activity [ 55 ]. The overall effect appears to be that NFκB activity is upregulated by the neuropeptides involved in CRPS, resulting in a pro-inflammatory response mediated through NFκB.

2) NFκB involvement in hypoxic changes

NFκB activity is inhibited by NO [ 45 , 56 , 57 ]. However, peroxynitrite (ONOO), which is formed from NO after reaction with radical oxygen intermittents, sustains NFκB activity [ 45 ]. NFκB also is involved in the expression of inducible nitric oxide synthethase (iNOS) induced by pro-inflammatory cytokines TNFα and IL1-β, which are locally elevated in CRPS [ 10 , 58 ]. Additionally, NFκB activity is involved in and affected by free radical formation [ 50 , 59 , 60 ] and acidosis [ 22 , 61 , 62 ].

3) NFκB and neuropathic pain

Several animal models demonstrate the role of NFκB in pain induction and pain maintenance in the CNS. For example, NFκB is involved in the upregulation of some types of NMDA receptors which are important mediators in the development of central sensitization [ 63 ]. Additionally, NFκB in the CNS mediates IL-1 induced COX-2 up-regulation and prodynorphin expression [ 64 , 65 ]. The derived compound dynorphin is an endogenous opiate that causes hyperalgesia and allodynia in mice. In line with this, intrathecal (in the central spinal fluid) or neuronal injected NFκB inhibitors attenuate proinflammatory cytokine mediated pain in rats [ 66 , 67 ].

4) NFκB involvement in sympathetic dysregulation

Pro-inflammatory responses induced by the catecholamines derived from the sympathetic nervous system, which might be dysregulated in CRPS, are mediated by NFκB [ 68 - 70 ]. Additionally, NFκB up-regulates directly the expression of the β-adrenoreceptor on immune cells [ 71 ]. Moreover, the expression of the α-receptor might be regulated by NFκB in an indirect way which involves Il-1β mediated apoptosis of sympathetic neurons, leading to up-regulation and/or hypersensitivity of the peripheral α-receptors expressed on blood vessels, immune cells, and nociceptive primary afferents [ 72 ].

Discussion

This paper describes how a multidisciplinary team of investigators applied advanced information retrieval methods, in the form of a text mining/entity relationship tool, with the purpose of discovering new hypotheses concerning the pathogenesis of a complex disorder, exemplified here by CRPS. The exercise should be regarded as a "journey" to discover what benefit could emerge from this kind of collaboration between bioinformatics experts and clinical experts. The purpose was neither to assess the specificity of the used methodology, nor to discover complete underlying biochemical pathways.

The text mining/entity relationship tool PathwayAssist provided a new concept, named NFκB, which is related to a majority of the concepts described to be involved in CRPS. After manual literature search, NFκB appeared to be a link between the various, previously described pathogenic mechanisms and appeared to serve as a mediating component in their connections.

Validity of the method

The MedScan system has been manually validated by the developers through random extraction and analysis of direct physical protein-protein interactions from MEDLINE [ 44 ]. From the interactions extracted by MedScan, 91% was correct (precision). Most errors were attributed to the extracted functional interference between proteins rather than physical interaction. Coverage (recall), however, was low: 21%. This is primarily due to the low coverage rate of the NLP component (34%). The concept network on CRPS (figure 2 ) was validated manually by the authors of this paper by checking all relations leading to or from NFκB. Out of 38 relations, 34 were correct and 4 incorrect.

Naturally, it would be helpful if some sort of independent relevance score could be provided for new hypotheses generated by the method used in this report. In this view, a p-value was calculated for the new node NFκB in the network using a one-tailed Chi-square test with Yates's correction, comparing its node connectivity in the PathwayAssist database to its node connectivity in the CRPS concepts network. In addition, the protein TP53 was chosen as a reference case because of the equally high connectivity in the PathwayAssist database compared to NFκB. Unfortunately, TP53 was found to be equally significant in the CRPS network as NFκB. However, NFκB was connected to more concepts in the original CRPS network than TP53.

An automatic network analysis approach where a variety of nodes (for example all highly connected proteins in the PathwayAssist database, all nodes generated by the expanded network algorithm, or simply all nodes in the PathwayAssist database) would be tested for positive association to a subnetwork of predefined disease-specific nodes would be valuable for both hypotheses generation and testing. Unfortunately, there is no such algorithm available in the PathwayAssist version used by the authors of this paper. Furthermore, the necessary data (specific connectivity information for each node in the database, i.e. which other nodes it is connected to) required for carrying out these types of tests outside the framework of the tool are not provided by PathwayAssist. The assessment of the connectivity to the CRPS network for NFκB and TP53 was performed manually inspection, an activity that is very time consuming and not defendable for a larger number of nodes. Despite the limitations regarding the assessment of the specificity of the results, a new node found by one of the network gap analysis algorithms provided by PathwayAssist, which was significantly related to the CRPS network, was considered important enough for further investigation.

One should bear in mind that results from text mining exercises as described in this paper are far from the solution to the medical problem or the complete answer to outstanding questions. The current tools in the biomedical domain are not capable of delivering clear and already assessed hypotheses. However, in this case, the results of a simple knowledge gap analysis provided a new idea that, after further manual exploration in literature, appeared very plausible and worthwhile investigating in biological experiments and epidemiological studies.

The generalizability of this method of hypothesis discovery needs to be assessed by repeating the exercise for other complex medical conditions. However, it would be an exhaustive and extremely time consuming task to repeat the exercise for a large group of complex disorders. This is a well-known problem in the evaluation of text mining tools in general [ 73 ].

Testing the new hypothesis

New biological data are needed to verify the in silico derived hypothesis concerning the involvement of NFκB in the pathogenesis of CRPS. These data could be generated through animal models, clinical studies, and epidemiological studies. A rat model for CRPS has been developed by Coderre and colleagues [ 74 ]. They named it the chronic post ischemia pain (CPIP) model, since CRPS like symptoms are provoked by ischemia and reperfusion of the hindpaw by binding it temporarily with a tourniquet. This model may prove useful to investigate the role of NFκB in the early development of CRPS, for example by measuring the transcription of NFκB in the affected tissue. Additionally, in this model the effect of administration of NFκB inhibitors could be studied. In the past, NFκB knockout and transgenic animal models have been extensively used to study the NFκB pathway [ 75 ]. An increased inflammatory response coupled with an increased susceptibility to opportunistic infections has been recorded [ 76 , 77 ], and a decrease of the electroacupuncture-induced analgesic effects has also been shown [ 78 ]. A failure to induce CRPS like symptoms in NFκB knock mice would support the hypothesis of a NFκB as a crucial factor in CRPS.

Clinical and epidemiological studies to test the hypothesis could involve the investigation of determinants of altered NFκB activity and the comparison in this view between CRPS cases and healthy controls. For example it could be tested whether other NFκB related disorders (asthma, autoimmune disorders or atherosclerosis) co-occur with CRPS. Viral infections are known to upregulate NFκB activity. Thus the time relation between CRPS and the occurrence of a viral infection may also be worthwhile investigating. Finally, one could search for features from a genetic origin. The increased prevalence of an ACE polymorphism in CRPS patients was found in a small study [ 79 ]. Since ACE is also an inhibitor of NFκB activity, further investigations regarding this polymorphism might be of interest.

Certain drugs affect (perhaps unintentionally) the NFκB pathway and could influence the development of CRPS (see also section Targeting the NFκB pathway). Currently, according to the recently developed Dutch evidence based Guideline Complex Regional Pain Syndrome type 1 [ 80 ], nine drugs have been proven beneficial in the prevention or treatment of CRPS, including ketamine [ 81 ], gabapentine [ 82 ], DMSO crème [ 83 ], N-acetylcysteine [ 83 ], corticosteroids [ 84 ], bisfosfonates [ 85 ], calcium antagonists [ 86 ], ketanserine [ 87 ], and vitamin C [ 88 ]. Interestingly, 126 small molecules are listed as NFκB inhibitors in PathwayAssist, and 5 of these overlap with the list of drugs proven beneficial for preventing or treating CRPS (ketamine [ 89 ], DMSO crème [ 90 ], N-acetylcysteine [ 91 ], corticosteroids [ 92 ], and calcium antagonists [ 93 ]).

A Dutch cohort of CRPS patients was identified in the Integrated Primary Care Database (IPCI), a database that makes electronic patient records used in routine care available to investigators [ 94 ]. This cohort has been used before in research on CRPS [ 95 ] and will be used for further testing of these hypotheses within the framework of INFOBIOMED.

Clinical relevance

NFκB is a transcription factor that is known to be involved in many processes, but its function is best described in inflammation [ 46 ]. In the recent past NFκB has been discovered as an important mediator in diseases due to chronic and exaggerated inflammatory responses, including sepsis, asthma, rheumatic disorders, inflammatory bowel disease, and psoriasis. However, in the current available literature NFκB has never before been mentioned in association with CRPS. Therefore, this relation is an interesting new result derived from the text mining exercise. Especially remarkable is that, in the generated network, NFκB is not only linked with the inflammatory concepts related in CRPS, but also to the non-inflammatory concepts in CRPS, such as neuropeptides and catecholamines.

Realizing the possible central role for the NFκB pathway in the mechanisms underlying CRPS (or in the stress response caused by CRPS), unanswered questions concerning the disorder can be reviewed from a new point of view. For instance, when placing NFκB in the centre of the pathogenic process, the wide variety of precipitating events could be explained by the observation that all pathogenic mechanisms in CRPS are related to each other. This relation suggests that an event that specifically triggers one of the four mechanisms induces the entire complex pathogenic process (figure 3 , figure 4 , figure 5 , and figure 6 ). For each mechanism separately the potential triggers are limited, but for the four mechanisms together the possibilities are numerous. With the addition of NFκB, the biochemical picture underlying CRPS becomes one in which several processes can initiate the disorder and in which NFκB can play a key role in its propagation.

Assuming the central unifying role of NFκB in the pathogenesis of CRPS, independent whether NFκB is the final molecule in the pathway or not, new targets for drug therapy could be provided. Current therapy in CRPS is aimed at targeting separate mechanisms. However, targeting one mechanism in CRPS is not enough, as made apparent by the limited success rate for the majority of treatments. Based on the new hypothesis, targeting the NFκB pathway may provide a way to target all the underlying mechanisms at once. Thereby, progression of the disorder may be more effective, since it will be prevented at more than one level. Independent from the inciting event or principle disturbed mechanism, NFκB pathway inhibitors might remove the 'Движок' that keeps the process running.

Targeting NFκB as a therapy in sepsis, inflammatory diseases and neuronal diseases is proposed by several authors [ 47 , 50 , 96 , 97 ]. It is suggested to be very promising for a wide variety of patients, but due to the involvement of NFκB in a wide variety of physiological processes, there are safety implications. New drugs that target one of the components of the NFκB pathway are currently under development. One of these drugs, bortezomib, was already launched in the United States in 2003 for the treatment of multiple myeloma. Other drugs are in various stages of development (table 3 ). The NFκB pathway inhibitors are designed for the use in treatment of a variety of disorders, including inflammatory diseases, autoimmune diseases, atopic disorders, arteriosclerosis and malignancies. None of the drugs currently under development is mentioned for treatment of CRPS. When these drugs have proven to be safe, the therapeutic effect of NFκB pathway inhibitors could be studied in CRPS patients.

Conclusion

Computer-assisted literature analysis to support the generation of novel and testable hypotheses has previously been proven useful [ 5 , 6 ]. This study builds upon this research, and extends it by bringing together researchers and clinicians with a wide variety of backgrounds. An experimental exercise was performed in which a text-mining/entity relationship tool has been applied to systematically synthesize the knowledge in existing literature about a complex disorder, exemplified by CRPS. Within the created literature network, a simple knowledge gap analysis was performed, using node connectivity. This approach was essential in formulating the hypothesis that NFκB might be a key player molecule in the pathogenesis of CRPS. This example of multidisciplinary research illustrates how the collaborative efforts of investigators from different fields of expertise can demonstrate new directions for future biological and epidemiological research on a complex disease.

Competing interests

The author(s) declare that they have no competing interests.

Authors' contributions

KMH participated in the design of the study, performed the pathway analyses and helped to draft the manuscript. MM provided biomedical knowledge on CRPS and drafted the manuscript. AGJB provided the biomedical concepts as input for thepathway analysesand participated in the design of the study. MW helped in performing the pathway and literature analyses and participated in the design of the study. EMM, JM and MC participated actively in the discussions and advised on the interpretation of the results. SB and JL conceived of the study, and participated in its design and coordination. SB also helped to draft the manuscript. All authors read, commented and approved the final manuscript.

Acknowledgements

The work reported in this study was sponsored by the EC funded Network of Excellence INFOBIOMED (IST-507585). We thank Frank Huygen, anesthesiologist at the Erasmus MC, The Netherlands, for critically reviewing the manuscript. We also thank Dr. Martijn Schuemie and Rob Jelier, Medical Informatics, Erasmus MC, The Netherlands, for providing expert knowledge on statistical testing.

Luscombe NM Greenbaum D Gerstein M What is bioinformatics? A proposed definition and overview of the field Methods Inf Med 2001 40 346 358 11552348

Kohane IS Bioinformatics and clinical informatics: the imperative to collaborate J Am Med Inform Assoc 2000 7 512 516 10984470 10.1136/jamia.2000.0070512 PMC79046

Maojo V de la Calle G Martin-Sanchez F Diaz C Sanz F INFOBIOMED: European Network of Excellence on Biomedical Informatics to support individualised healthcare AMIA Annu Symp Proc 2005 1041 16779328 PMC1560781

Infobiomed

Smalheiser NR Swanson DR Using ARROWSMITH: a computer-assisted approach to formulating and assessing scientific hypotheses Comput Methods Programs Biomed 1998 57 149 153 9822851 10.1016/S0169-2607(98)00033-9

Weeber M Vos R Klein H De Jong-Van Den Berg LT Aronson AR Molema G Generating hypotheses by discovering implicit associations in the literature: a case report of a search for new potential therapeutic uses for thalidomide J Am Med Inform Assoc 2003 10 252 259 12626374 10.1197/jamia.M1158 PMC342048

Veldman PH Reynen HM Arntz IE Goris RJ Signs and symptoms of reflex sympathetic dystrophy: prospective study of 829 patients Lancet 1993 342 1012 1016 8105263 10.1016/0140-6736(93)92877-V

Bruehl S Harden RN Galer BS Saltz S Backonja M Stanton-Hicks M Complex regional pain syndrome: are there distinct subtypes and sequential stages of the syndrome? Pain 2002 95 119 124 11790474 10.1016/S0304-3959(01)00387-6

Merritt WH The challenge to manage reflex sympathetic dystrophy/complex regional pain syndrome Clin Plast Surg 2005 32 575 604, vii-viii 16139630 10.1016/j.cps.2005.07.002

Huygen FJ De Bruijn AG De Bruin MT Groeneweg JG Klein J Zijistra FJ Evidence for local inflammation in complex regional pain syndrome type 1 Mediators Inflamm 2002 11 47 51 11930962 10.1080/09629350210307 PMC1781643

Huygen FJ Ramdhani N van Toorenenbergen A Klein J Zijlstra FJ Mast cells are involved in inflammatory reactions during Complex Regional Pain Syndrome type 1 Immunol Lett 2004 91 147 154 15019283 10.1016/j.imlet.2003.11.013

Alexander GM van Rijn MA van Hilten JJ Perreault MJ Schwartzman RJ Changes in cerebrospinal fluid levels of pro-inflammatory cytokines in CRPS Pain 2005 116 213 9 15964681 10.1016/j.pain.2005.04.013

Blair SJ Chinthagada M Hoppenstehdt D Kijowski R Fareed J Role of neuropeptides in pathogenesis of reflex sympathetic dystrophy Acta Orthop Belg 1998 64 448 451 9922551

Brain SD Tippins JR Morris HR MacIntyre I Williams TJ Potent vasodilator activity of calcitonin gene-related peptide in human skin J Invest Dermatol 1986 87 533 536 2428885 10.1111/1523-1747.ep12455620

Brain SD Williams TJ Tippins JR Morris HR MacIntyre I Calcitonin gene-related peptide is a potent vasodilator Nature 1985 313 54 56 3917554 10.1038/313054a0

Leis S Weber M Schmelz M Birklein F Facilitated neurogenic inflammation in unaffected limbs of patients with complex regional pain syndrome Neurosci Lett 2004 359 163 166 15050689 10.1016/j.neulet.2004.02.025

Pham T Lafforgue P Reflex sympathetic dystrophy syndrome and neuromediators Joint Bone Spine 2003 70 12 17 12639612 10.1016/S1297-319X(02)00006-4

Holzer P Neurogenic vasodilatation and plasma leakage in the skin Gen Pharmacol 1998 30 5 11 9457475 10.1016/s0306-3623(97)00078-5

Kramer HH Schmidt K Leis S Schmelz M Sommer C Birklein F Inhibition of neutral endopeptidase (NEP) facilitates neurogenic inflammation Exp Neurol 2005 195 179 184 15963503 10.1016/j.expneurol.2005.04.015

Birklein F Weber M Ernst M Riedl B Neundorfer B Handwerker HO Experimental tissue acidosis leads to increased pain in complex regional pain syndrome (CRPS) Pain 2000 87 227 234 10924816 10.1016/S0304-3959(00)00286-4

Heerschap A den Hollander JA Reynen H Goris RJ Metabolic changes in reflex sympathetic dystrophy: a 31P NMR spectroscopy study Muscle Nerve 1993 16 367 373 8455649 10.1002/mus.880160405

Kellum JA Song M Li J Lactic and hydrochloric acids induce different patterns of inflammatory response in LPS-stimulated RAW 264.7 cells Am J Physiol Regul Integr Comp Physiol 2004 286 R686 92 14695114 10.1152/ajpregu.00564.2003

Koban M Leis S Schultze-Mosgau S Birklein F Tissue hypoxia in complex regional pain syndrome Pain 2003 104 149 157 12855324 10.1016/S0304-3959(02)00484-0

van der Laan L Veldman PH Goris RJ Severe complications of reflex sympathetic dystrophy: infection, ulcers, chronic edema, dystonia, and myoclonus Arch Phys Med Rehabil 1998 79 424 429 9552109 10.1016/S0003-9993(98)90144-7

Groeneweg JG Huygen FJ Heijmans-Antonissen C Niehof S Zijlstra FJ Increased endothelin-1 and diminished nitric oxide levels in blister fluids of patients with intermediate cold type complex regional pain syndrome type 1 BMC Musculoskelet Disord 2006 7 91 17137491 10.1186/1471-2474-7-91 PMC1693561

Baba H Doubell TP Moore KA Woolf CJ Silent NMDA receptor-mediated synapses are developmentally regulated in the dorsal horn of the rat spinal cord J Neurophysiol 2000 83 955 962 10669507 10.1152/jn.2000.83.2.955

Bennett GJ An animal model of neuropathic pain: a review Muscle Nerve 1993 16 1040 1048 8413357 10.1002/mus.880161007

Woolf CJ Salter MW Neuronal plasticity: increasing the gain in pain Science 2000 288 1765 1769 10846153 10.1126/science.288.5472.1765

Maihofner C Handwerker HO Neundorfer B Birklein F Patterns of cortical reorganization in complex regional pain syndrome Neurology 2003 61 1707 1715 14694034 10.1212/01.wnl.0000098939.02752.8e

Dommerholt J Complex regional pain syndrome-1: history, diagnostic criteria and etiology. Journal of Bodywork and Movement Therapies 2004 8 167 177 10.1016/S1360-8592(03)00100-1

Birklein F Complex regional pain syndrome J Neurol 2005 252 131 138 15729516 10.1007/s00415-005-0737-8

Gonzales R Sherbourne CD Goldyne ME Levine JD Noradrenaline-induced prostaglandin production by sympathetic postganglionic neurons is mediated by alpha 2-adrenergic receptors J Neurochem 1991 57 1145 1150 1654387 10.1111/j.1471-4159.1991.tb08272.x

Janig W Baron R Complex regional pain syndrome: mystery explained? Lancet Neurol 2003 2 687 697 14572737 10.1016/S1474-4422(03)00557-X

Baron R Levine JD Fields HL Causalgia and reflex sympathetic dystrophy: does the sympathetic nervous system contribute to the generation of pain? Muscle Nerve 1999 22 678 695 10366221 10.1002/(SICI)1097-4598(199906)22:6<678::AID-MUS4>3.0.CO;2-P

Watkins LR Maier SF Beyond neurons: evidence that immune and glial cells contribute to pathological pain states Physiol Rev 2002 82 981 1011 12270950 10.1152/physrev.00011.2002

Guo TZ Offley SC Boyd EA Jacobs CR Kingery WS Substance P signaling contributes to the vascular and nociceptive abnormalities observed in a tibial fracture rat model of complex regional pain syndrome type I Pain 2004 108 95 107 15109512 10.1016/j.pain.2003.12.010

Janig W Baron R Experimental approach to CRPS Pain 2004 108 3 7 15109501 10.1016/j.pain.2004.01.005

Khalil Z Andrews PV Helme RD VIP modulates substance P-induced plasma extravasation in vivo Eur J Pharmacol 1988 151 281 287 2458946 10.1016/0014-2999(88)90809-6

Kingery WS Agashe GS Guo TZ Davies MF Clark JD Maze M Capsaicin sensitive afferents mediate the development of heat hyperalgesia and hindpaw edema after sciatic section in rats Neurosci Lett 2002 318 39 43 11786220 10.1016/S0304-3940(01)02464-8

Yonan AL Palmer AA Smith KC Feldman I Lee HK Yonan JM Fischer SG Pavlidis P Gilliam TC Bioinformatic analysis of autism positional candidate genes using biological databases and computational gene network prediction Genes Brain Behav 2003 2 303 320 14606695 10.1034/j.1601-183X.2003.00041.x

Serafica MD Goto T Trounson AO Transcripts from a human primordial follicle cDNA library Hum Reprod 2005 20 2074 2091 15845590 10.1093/humrep/dei030

Nikitin A Egorov S Daraselia N Mazo I Pathway studio--the analysis and navigation of molecular networks Bioinformatics 2003 19 2155 2157 14594725 10.1093/bioinformatics/btg290

Novichkova S Egorov S Daraselia N MedScan, a natural language processing engine for MEDLINE abstracts Bioinformatics 2003 19 1699 1706 12967967 10.1093/bioinformatics/btg207

Daraselia N et al Extracting human protein interactions from MEDLINE using a full-sentence parser. Bioinformatics 2004 20 604 611 15033866 10.1093/bioinformatics/btg452

Hattori Y Hattori S Sato N Kasai K High-glucose-induced nuclear factor kappaB activation in vascular smooth muscle cells Cardiovasc Res 2000 46 188 197 10727667 10.1016/S0008-6363(99)00425-3

Chen F Castranova V Shi X Demers LM New insights into the role of nuclear factor-kappaB, a ubiquitous transcription factor in the initiation of diseases Clin Chem 1999 45 7 17 9895331

Barnes PJ Karin M Nuclear factor-kappaB: a pivotal transcription factor in chronic inflammatory diseases N Engl J Med 1997 336 1066 1071 9091804 10.1056/NEJM199704103361506

Jongeneel CV Regulation of the TNF alpha gene Prog Clin Biol Res 1994 388 367 381 7831369

Sehgal PB Regulation of IL6 gene expression Res Immunol 1992 143 724 734 1439145 10.1016/0923-2494(92)80011-9

Lieb K Fiebich BL Berger M Bauer J Schulze-Osthoff K The neuropeptide substance P activates transcription factor NF-kappa B and kappa B-dependent gene expression in human astrocytoma cells J Immunol 1997 159 4952 4958 9366421

Xie P Browning DD Hay N Mackman N Ye RD Activation of NF-kappa B by bradykinin through a Galpha(q)- and Gbeta gamma-dependent pathway that involves phosphoinositide 3-kinase and Akt J Biol Chem 2000 275 24907 24914 10801799 10.1074/jbc.M001051200

Delgado M Ganea D Inhibition of endotoxin-induced macrophage chemokine production by VIP and PACAP in vitro and in vivo Arch Physiol Biochem 2001 109 377 382 11935377 10.1076/apab.109.4.377.4237

Eisenberg E Erlich T Zinder O Lichinsky S Diamond E Pud D Davar G Plasma endothelin-1 levels in patients with complex regional pain syndrome Eur J Pain 2004 8 533 538 15531221 10.1016/j.ejpain.2003.12.003

Ball HJ Shine J Herzog H Multiple promoters regulate tissue-specific expression of the human NPY-Y1 receptor gene J Biol Chem 1995 270 27272 27276 7592987 10.1074/jbc.270.45.27272

Schmeisser A Soehnlein O Illmer T Lorenz HM Eskafi S Roerick O Gabler C Strasser R Daniel WG Garlichs CD ACE inhibition lowers angiotensin II-induced chemokine expression by reduction of NF-kappaB activity and AT1 receptor expression Biochem Biophys Res Commun 2004 325 532 540 15530425 10.1016/j.bbrc.2004.10.059

Lozano FS Barros MB Garcia-Criado FJ Gomez-Alonso A Exogenous nitric oxide can control SIRS and downregulate NFkappaB J Surg Res 2005 124 52 58 15734479 10.1016/j.jss.2004.08.029

Matthews JR Botting CH Panico M Morris HR Hay RT Inhibition of NF-kappaB DNA binding by nitric oxide Nucleic Acids Res 1996 24 2236 2242 8710491 10.1093/nar/24.12.2236 PMC145924

Taylor BS de Vera ME Ganster RW Wang Q Shapiro RA Morris SM Jr. Billiar TR Geller DA Multiple NF-kappaB enhancer elements regulate cytokine induction of the human inducible nitric oxide synthase gene J Biol Chem 1998 273 15148 15156 9614127 10.1074/jbc.273.24.15148

Kanda N Watanabe S Substance P enhances the production of interferon-induced protein of 10 kDa by human keratinocytes in synergy with interferon-gamma J Invest Dermatol 2002 119 1290 1297 12485430 10.1046/j.1523-1747.2002.19626.x

Bhattacharyya J Biswas S Datta AG Mode of action of endotoxin: role of free radicals and antioxidants Curr Med Chem 2004 11 359 368 14965237 10.2174/0929867043456098

Takeshita K Suzuki Y Nishio K Takeuchi O Toda K Kudo H Miyao N Ishii M Sato N Naoki K Aoki T Suzuki K Hiraoka R Yamaguchi K Hypercapnic acidosis attenuates endotoxin-induced nuclear factor-[kappa]B activation Am J Respir Cell Mol Biol 2003 29 124 132 12600832 10.1165/rcmb.2002-0126OC

Xu L Fidler IJ Acidic pH-induced elevation in interleukin 8 expression by human ovarian carcinoma cells Cancer Res 2000 60 4610 4616 10969814

Chiechio S Copani A De Petris L Morales ME Nicoletti F Gereau RW Transcriptional regulation of metabotropic glutamate receptor 2/3 expression by the NF-kappaB pathway in primary dorsal root ganglia neurons: a possible mechanism for the analgesic effect of L-acetylcarnitine Mol Pain 2006 2 20 16764720 10.1186/1744-8069-2-20 PMC1550235

Tegeder I Niederberger E Schmidt R Kunz S Guhring H Ritzeler O Michaelis M Geisslinger G Specific Inhibition of IkappaB kinase reduces hyperalgesia in inflammatory and neuropathic pain models in rats J Neurosci 2004 24 1637 1645 14973242 10.1523/JNEUROSCI.3118-03.2004 PMC6730471

Kemler MA van de Vusse AC van den Berg-Loonen EM Barendse GA van Kleef M Weber WE HLA-DQ1 associated with reflex sympathetic dystrophy Neurology 1999 53 1350 1351 10522900 10.1212/wnl.53.6.1350

Ledeboer A Gamanos M Lai W Martin D Maier SF Watkins LR Quan N Involvement of spinal cord nuclear factor kappaB activation in rat models of proinflammatory cytokine-mediated pain facilitation Eur J Neurosci 2005 22 1977 1986 16262636 10.1111/j.1460-9568.2005.04379.x

Inoue G Ochiai N Ohtori S Nakagawa K Gemba T Doya H Ito T Koshi T Moriya H Takahashi K Injection of nuclear factor-kappa B decoy into the sciatic nerve suppresses mechanical allodynia and thermal hyperalgesia in a rat inflammatory pain model Spine 2006 31 2904 2908 17139220 10.1097/01.brs.0000248424.46652.67

Ballard-Croft C Maass DL Sikes P White J Horton J Activation of stress-responsive pathways by the sympathetic nervous system in burn trauma Shock 2002 18 38 45 12095132 10.1097/00024382-200207000-00008

Le Tulzo Y Shenkar R Kaneko D Moine P Fantuzzi G Dinarello CA Abraham E Hemorrhage increases cytokine expression in lung mononuclear cells in mice: involvement of catecholamines in nuclear factor-kappaB regulation and cytokine expression J Clin Invest 1997 99 1516 1524 9119995 10.1172/JCI119314 PMC507971

Shahani R Klein LV Marshall JG Nicholson S Rubin BB Walker PM Lindsay TF Hemorrhage-induced alpha-adrenergic signaling results in myocardial TNF-alpha expression and contractile dysfunction Am J Physiol Heart Circ Physiol 2001 281 H84 92 11406472 10.1152/ajpheart.2001.281.1.H84

Bierhaus A Wolf J Andrassy M Rohleder N Humpert PM Petrov D Ferstl R von Eynatten M Wendt T Rudofsky G Joswig M Morcos M Schwaninger M McEwen B Kirschbaum C Nawroth PP A mechanism converting psychosocial stress into mononuclear cell activation Proc Natl Acad Sci U S A 2003 100 1920 1925 12578963 10.1073/pnas.0438019100 PMC149934

Drummond PD Involvement of the sympathetic nervous system in complex regional pain syndrome Int J Low Extrem Wounds 2004 3 35 42 15866786 10.1177/1534734604263365

Shatkay H Feldman R Mining the biomedical literature in the genomic era: an overview J Comput Biol 2003 10 821 855 14980013 10.1089/106652703322756104

Coderre TJ Xanthos DN Francis L Bennett GJ Chronic post-ischemia pain (CPIP): a novel animal model of complex regional pain syndrome-type I (CRPS-I; reflex sympathetic dystrophy) produced by prolonged hindpaw ischemia and reperfusion in the rat Pain 2004 112 94 105 15494189 10.1016/j.pain.2004.08.001

Gerondakis S Grumont R Gugasyan R Wong L Isomura I Ho W Banerjee A Unravelling the complexities of the NF-kappaB signalling pathway using mouse knockout and transgenic models Oncogene 2006 25 6781 6799 17072328 10.1038/sj.onc.1209944

Ishikawa H Claudio E Dambach D Raventos-Suarez C Ryan C Bravo R Chronic inflammation and susceptibility to bacterial infections in mice lacking the polypeptide (p)105 precursor (NF-kappaB1) but expressing p50 J Exp Med 1998 187 985 996 9529315 10.1084/jem.187.7.985 PMC2212206

Lawrence T Bebien M Liu GY Nizet V Karin M IKKalpha limits macrophage NF-kappaB activation and contributes to the resolution of inflammation Nature 2005 434 1138 1143 15858576 10.1038/nature03491

Park HJ Lee HS Lee HJ Yoo YM Lee HJ Kim SA Leem K Kim HC Seo JC Kim EH Lim S Chung JH Decrease of the electroacupuncture-induced analgesic effects in nuclear factor-kappa B1 knockout mice Neurosci Lett 2002 319 141 144 11834313 10.1016/S0304-3940(01)02582-4

Kimura T KT Hosada R, Nishiwaki K, Shimada Y. Angiotensin-converting enzyme gene polymorphism in patients with neuropathic pain Proceedings of the 9th World Congress in Pain Seattle (WA): IASP press 2000 471 476

Geertzen JHB Perez RSGM Dijkstra PU Kemler MA Rosenbrand CJGM Richtlijn Complex Regionaal Pijn Syndroom type I Van Zuiden Communications BV 2006 chapters 2 and 5

Correll GE Maleki J Gracely EJ Muir JJ Harbut RE Subanesthetic ketamine infusion therapy: a retrospective analysis of a novel therapeutic approach to complex regional pain syndrome Pain Med 2004 5 263 275 15367304 10.1111/j.1526-4637.2004.04043.x

van de Vusse AC Stomp-van den Berg SG Kessels AH Weber WE Randomised controlled trial of gabapentin in Complex Regional Pain Syndrome type 1 [ISRCTN84121379] BMC Neurol 2004 4 13 15453912 10.1186/1471-2377-4-13 PMC523854

Perez RS Zuurmond WW Bezemer PD Kuik DJ van Loenen AC de Lange JJ Zuidhof AJ The treatment of complex regional pain syndrome type I with free radical scavengers: a randomized controlled study Pain 2003 102 297 307 12670672 10.1016/S0304-3959(02)00414-1

Grundberg AB Reflex sympathetic dystrophy: treatment with long-acting intramuscular corticosteroids J Hand Surg [Am] 1996 21 667 670 8842963 10.1016/S0363-5023(96)80023-2

Manicourt DH Brasseur JP Boutsen Y Depreseux G Devogelaer JP Role of alendronate in therapy for posttraumatic complex regional pain syndrome type I of the lower extremity Arthritis Rheum 2004 50 3690 3697 15529370 10.1002/art.20591

Muizelaar JP Kleyer M Hertogs IA DeLange DC Complex regional pain syndrome (reflex sympathetic dystrophy and causalgia): management with the calcium channel blocker nifedipine and/or the alpha-sympathetic blocker phenoxybenzamine in 59 patients Clin Neurol Neurosurg 1997 99 26 30 9107464 10.1016/S0303-8467(96)00594-X

Kingery WS A critical review of controlled clinical trials for peripheral neuropathic pain and complex regional pain syndromes Pain 1997 73 123 139 9415498 10.1016/S0304-3959(97)00049-3

Zollinger PE Tuinebreijer WE Kreis RW Breederveld RS Effect of vitamin C on frequency of reflex sympathetic dystrophy in wrist fractures: a randomised trial Lancet 1999 354 2025 2028 10636366 10.1016/S0140-6736(99)03059-7

Sun J Wang XD Liu H Xu JG Ketamine suppresses intestinal NF-kappa B activation and proinflammatory cytokine in endotoxic rats World J Gastroenterol 2004 10 1028 1031 15052687 10.3748/wjg.v10.i7.1028 PMC4717093

Chang CK Albarillo MV Schumer W Therapeutic effect of dimethyl sulfoxide on ICAM-1 gene expression and activation of NF-kappaB and AP-1 in septic rats J Surg Res 2001 95 181 187 11162043 10.1006/jsre.2000.6033

Pyo H Joe E Jung S Lee SH Jou I Gangliosides activate cultured rat brain microglia J Biol Chem 1999 274 34584 34589 10574921 10.1074/jbc.274.49.34584

Donjerkovic D Mueller CM Scott DW Steroid- and retinoid-mediated growth arrest and apoptosis in WEHI-231 cells: role of NF-kappaB, c-Myc and CKI p27(Kip1) Eur J Immunol 2000 30 1154 1161 10760805 10.1002/(SICI)1521-4141(200004)30:4<1154::AID-IMMU1154>3.0.CO;2-B

Amrani Y Lazaar AL Hoffman R Amin K Ousmer S Panettieri RA Jr. Activation of p55 tumor necrosis factor-alpha receptor-1 coupled to tumor necrosis factor receptor-associated factor 2 stimulates intercellular adhesion molecule-1 expression by modulating a thapsigargin-sensitive pathway in human tracheal smooth muscle cells Mol Pharmacol 2000 58 237 245 10860946 10.1124/mol.58.1.237

van der Lei J Duisterhout JS Westerhof HP van der Does E Cromme PV Boon WM van Bemmel JH The introduction of computer-based patient records in The Netherlands Ann Intern Med 1993 119 1036 1041 8214981 10.7326/0003-4819-119-10-199311150-00011

de Mos M de Bruijn AG Huygen FJ Dieleman JP Stricker BH Sturkenboom MC The incidence of complex regional pain syndrome: A population-based study Pain 2006 17084977 10.1016/j.pain.2006.09.008

Howard PW Stewart HD Hind RE Burke FD External fixation or plaster for severely displaced comminuted Colles' fractures? A prospective study of anatomical and functional results J Bone Joint Surg Br 1989 71 68 73 2915010 10.1302/0301-620X.71B1.2915010

Abraham E Nuclear factor-kappaB and its role in sepsis-associated organ failure J Infect Dis 2003 187 Suppl 2 S364 9 12792853 10.1086/374750

Leis S Weber M Isselmann A Schmelz M Birklein F Substance-P-induced protein extravasation is bilaterally increased in complex regional pain syndrome Exp Neurol 2003 183 197 204 12957502 10.1016/S0014-4886(03)00163-8

Chard MD Ghatei MA Bloom S Crisp AJ Vasoactive intestinal polypeptide in algodystrophy Br J Rheumatol 1990 29 489 490 2257465 10.1093/rheumatology/29.6.489-a

Victor VM Rocha M De la Fuente M Immune cells: free radicals and antioxidants in sepsis Int Immunopharmacol 2004 4 327 347 15037211 10.1016/j.intimp.2004.01.020

Bai Y Hart RP Cultured sympathetic neurons express functional interleukin-1 receptors J Neuroimmunol 1998 91 43 54 9846818 10.1016/S0165-5728(98)00149-0

Hartrick CT Increased production of nitric oxide stimulated by interferon-gamma from peripheral blood monocytes in patients with complex regional pain syndrome Neurosci Lett 2002 323 75 77 11911993 10.1016/S0304-3940(02)00112-X

Azzolina A Bongiovanni A Lampiasi N Substance P induces TNF-alpha and IL-6 production through NF kappa B in peritoneal mast cells Biochim Biophys Acta 2003 1643 75 83 14654230 10.1016/j.bbamcr.2003.09.003

Millet I Phillips RJ Sherwin RS Ghosh S Voll RE Flavell RA Vignery A Rincon M Inhibition of NF-kappaB activity and enhancement of apoptosis by the neuropeptide calcitonin gene-related peptide J Biol Chem 2000 275 15114 15121 10809748 10.1074/jbc.275.20.15114

Ludwig J Baron R Complex Regional Pain Syndrome: an inflammatory pain condition? Drug Discovery Today 2004 1 449 455

Investigational Drugs DataBase