Merits of publishing a negative and prematurely ended study on intrathecal methylprednisolone for severe and longstanding complex regional pain syndrome (CRPS‐I)

The Editors of EJP have accepted for publication in this issue of the journal negative results of a placebo-controlled study of a single dose intrathecal methylprednisolone 80 mg for severe pain of longstanding complex regional pain syndrome (CRPS-I) (Munts et al., 2010). The study was also discontinued prematurely. CRPS-I is a complex pain condition that in severity is often disproportionate to any inciting tissue injury. In CRPS-I differs from CRPS-II mainly in lack of evidence of permanent major peripheral nerve injury (Harden and Bruehl, 2005). The patients included by Munts et al. had longstanding (median 4.5 years) and severe pain, around 7 on a 0–10 Numeric Rating Scale (NRS). The patients had movement disorders in addition to varying numbers of the typical CRPS symptoms of sensory abnormalities (hyperalgesia/allodynia), vasomotor abnormalities (temperature/skin colour asymmetry), and oedema/hyperhidrosis. The primary outcome measure was pain intensity assessed with NRS 6 weeks after intrathecal treatment. An interim analysis could not detect any effects on pain intensity, but a worsening of myoclonus in some patients after methylprednisolone, and other significant side-effects in both groups. Therefore, the trial was stopped when only half of the planned numbers of patients were evaluated. CSF-leakage after dural puncture with relatively large bore needles caused severe headache in 38% of the 21 patients, median duration 9 days, but was persistent in three patients. Two patients had bloodpatch treatment, without effect. Fourty-three percent had backache (median duration 14 days) that was persistent in four patients. Immunological processes are activated in the CNS in response to nerve injury (Hansson, 2010); microglia (macrophages of the CNS) and astrocytes are activated and changes in the astrocyte networks and pain regulating mechanisms occur (Costigan et al., 2009; Hansson, 2010). Whereas signs of major peripheral nerve damage are not present when the diagnosis of CRPS-I is obvious often a long time after the initial injury, the inciting tissue damage may have partially damaged peripheral nerves, that later regenerated. In animal studies, CRPS-I like conditions can be generated without sectioning or constricting peripheral nerves (Vatine et al., 2001). The immunosuppressant methylprednisolone reduced the activation of spinal cord microglia and prevented development of signs of nerve injury pain in rats (Takeda et al., 2004, 2005). In a clinical study, hyperaesthesia one year after surgery was reduced by 50% when methylprednisolone 125 mg was given (i.v.) before incision for augmentation mammoplasty (Romundstad et al., 2006; Kaasa et al., 2010). Positive results on pain after four intrathecal injections of methylprednisolone in patients with post herpes zoster neuralgia inspired Muntes et al. to undertake their trial (Kotani et al., 2000). The hypothesis of Munts et al. (2010) was that while oral glucocorticoids have some effects in the early phase of CRPS (Kingery, 1997), in later stages of CRPS the low CNS-bioavailability of oral glucocorticoids would require direct subarachnoid application of the drug to obtain therapeutic concentrations and beneficial effects on a putative ongoing immunological process in the CNS. The authors propose that the completely negative results of their interrupted study of intrathecal methylprednisolone most likely were due to the very longstanding nature of CRPS in their patients, and that maladaptive changes in the CNS had become irreversible. This is not an unreasonable hypothesis. According to Milligan and Watkins (2009), glia have emerged as key contributors to pathological and chronic pain mechanisms. On activation, both astrocytes and microglia respond to and release a number of signalling molecules, which have protective and may have pathogenetic functions. Current drugs for neuropathic pain and CRPS (Oaklander, 2005; Dworkin et al., 2007) target neuronal mechanisms, and they are only partly effective, at best. Drugs that manipulate neuron–glia interactions in pain processing should be explored, as Munts et al. have done: we now know that glucocorticoids applied at late stages of CRPS-I most likely are ineffective, even after intrathecal application. The Kotani et al. study from 2000 have not been verified by others. Minor or partial nerve damages are likely to occur in any tissue-injury inciting the CRPS-I process (Oaklander and Birklein, 2005). The subsequently developing CRPS-I pain condition is often disproportionate to the inciting tissue injury (Oaklander and Birklein, 2005). Although signs of major nerve damage is present in CRPS-II, the clinical pictures of CRPS-I and -II are often similar. An initiating, partial nerve damage may have healed completely in later stages of CRPS-I, but the pathological processes in the CNS continue, which may be true of any neuro-inflammatory processes. Treede et al. (2008) suggested that CRPS-I is a non-neuropathic condition because there is no objective sign of persistent peripheral nerve damage. However, it is not possible at the later stages of CRPS-I to exclude a partial nerve damage at the time of the inciting tissue injury, a nerve injury that has subsequently healed completely. The intial nerve damage may have started the pathogenetic mechanisms in genetically or otherwise susceptible patients (as in animals – Vatine et al. (2001)). And these pathological processes may eventually lead to irreversible changes in the pain regulating networks in spinal cord and other CNS sites. In the meantime the peripheral nerve damage may have healed. Thus immunological reactions in the CNS after tissue injury (with major or minor, partial nerve damage) may be important in the early stages of CRPS, and at these stages of an incipient CRPS-condition, immunosuppressive drugs like the glucocorticoids (Rhen and Cidlowski, 2005; Kehlet et al., 2006) may have beneficial effects, preventing or reversing the processes that otherwise develop into chronic, intractable CRPS. The long-term results of microglia and astrocyte activations, in susceptible individuals, may be permanently maladaptive and cause structural changes (Hansson, 2010). However, neuronal cell death and aberrant synaptic connectivity are most likely results of multiple maladaptive mechanisms that are involved in neuropathic pain and CRPS-like pain conditions, with or without major nerve damage (Gordh and Breivik, 2010). The neuroscience group behind the Treede et al. (2008) redefinition of neuropathic pain will be unhappy with such thoughts that muddle their nosologically “clean” definitions of neuropathic pain, in which the classical definition of CRPS-I does not belong. Whatever the reason(s) for the negative effects of the Munts et al. study, it is laudable that the editors of EJP publish a negative study, even though it is a study that was prematurely stopped. There are a number of stories of publication bias towards publishing of (false) positive studies, while true negative studies are either not submitted to the journals or they are not accepted for publication (Breivik et al., 2010). This causes patients to be treated with ineffective, often expensive, and sometimes unsafe treatments. The difficult dilemmas around deciding to prematurely discontinue a well designed and well planned clinical study was recently highlighted by Kvarstein et al. (2009a,b): when a planned interim analysis of a randomised and double blind study clearly shows little or no beneficial effects and also adverse effects of active as well as of placebo or sham treatment, discontinuing the study is the ethically only correct decision (Kvarstein et al., 2009b). In the Munts et al. (2010) study no beneficial effects were observed at the half-way interim study, whereas more than 1/3 had severe post-dural puncture headache, persistent in 14%, and more than 4/10 had backache that was persistent in 19%. Clearly the correct decision was to stop this study after the interim analysis. The editors of the EJP should be complimented for accepting this paper for publication.