Systematic review of techniques to monitor remission of acute Charcot neuroarthropathy in people with diabetes

The management of acute Charcot neuroarthropathy relies on off‐loading which is costly and time‐consuming. Published studies have used monitoring techniques with unknown diagnostic precision to detect remission. We performed a systematic review of techniques for monitoring response to offloading in acute Charcot neuroarthropathy.


| INTRODUCTION
Charcot neuroarthropathy (CN) is a complication of peripheral neuropathy associated with diabetes which affects the lower limb. It may be precipitated by minor trauma or other inflammatory insult which the patient does not notice due to insensitivity to pain. When, the patient does not rest the foot, an exaggerated inflammatory response occurs. 1 The symptoms include redness, warmth and swelling in the foot and/or leg. It can cause fractures and dislocations within the foot, which may progress to deformity and ulceration.
The treatment aims to stop the inflammatory process, relieve any pain and maintain foot structure. 2 Treatment for CN is "off-loading" the application of a non-removable plaster or fibreglass cast or boot; this rests and immobilizes the foot and redistributes the weight and pressure from the foot to the leg. 3 Off-loading is continued until remission when there are no longer clinical signs of inflammation, and X-rays are stable with signs of healing. 2 Globally, evidence suggests significant variation in treatment times.
In the United Kingdom, observational studies report treatment times of 9 to 12 months before remission is achieved [4][5][6] whilst data from the United States 7-10 and other European centres report treatment times of only 4-6 months. [11][12][13][14][15][16] Several factors could contribute to global variation, include participant characteristics, different techniques for monitoring, different protocols for the same monitoring techniques, variations in approach to off-loading and study design variability. 5 The current evidence base for the treatment of CN is poor. It is principally based on small retrospective cohort and observational studies of patients attending multidisciplinary foot clinics. Evidence to support the effectiveness of techniques to monitor CN is lacking, and current practice is primarily based on expert opinion. 2 Skin temperature is used because CN involves inflammation of the soft tissue and bone. 17 Skin temperature is however, a proxy measure of inflammation measured on the dorsum of the foot over the site of injury, which may not reflect the degree of inflammation within the affected deeper tissues, bones and/or joints. X-rays show damage to the foot skeleton rather than disease activity and are a measure of foot deformity. Despite these limitations, serial temperature measurements and X-rays remain the most widely used monitoring technique in CN.

Improvements in monitoring CN could reduce treatment times.
Lack of evidence to support clinicians in the choice of the type of monitoring and decision thresholds for remission may account for variability in treatment times. To the best of our knowledge, there are no systematic reviews focused on monitoring techniques to identify remission in CN.
Therefore, this systematic review aims to identify the effective-

| Inclusion and exclusion criteria
The inclusion criteria for study design were purposefully wide, based on prior knowledge of research studies on CN. We included randomized controlled trials, preference-controlled trials, and observational studies with or without control group(s). We excluded abstracts, systematic reviews and meta-analyses, studies on surgical and pharmacological management of CN, expert opinion, observations of single case studies and laboratory studies.
We included studies on off-loading which evaluated or reported monitoring techniques in adults with diabetes with a diagnosis of acute CN managed in any setting, including hospital, primary care or community. The control condition included other techniques used to monitor CN or the same technique used differently, for example different protocols for thermographic monitoring.

| Search strategy
We completed searches in PubMed, Embase, CINAHL, the Cochrane Central Register of Controlled Trials and ClinicalTrials.gov. The searches were restricted to English language, from 1993 to June 2018 and adapted for each database. See Appendix 1 for an example search strategy for PubMed. We used search terms for diabetes, Charcot, neuroarthropathy and osteoarthropathy. We also checked the reference lists of relevant published systematic reviews.
We downloaded all papers identified into EndNote and removed duplicates. Screening was conducted independently by two reviewers (C.G. and K.G.) in all three phases: title, abstract and full-text screening. Reasons for exclusion were recorded during abstract and full text screening. Inter-rater agreement was calculated by the number of papers on which the two reviewers agreed in terms of inclusion and exclusion, divided by the total number of double screened papers.
Discrepancies were resolved by consensus (C.G. and K.G.). All records deemed eligible following this consensus process were included for full text assessment or data extraction.
We extracted information on participant characteristics including type of diabetes, duration and HbA1c. We also extracted information on sensitivity and specificity of the techniques, protocol for application of the technique, costs and feasibility, safety and participant considerations. Finally, we extracted methods of off-loading and clinical outcomes such as time to healing and relapse rates.
The first author (C.G.) extracted data from all included papers.
The completed data extraction sheets were independently validated by a second reviewer (K.G.) against the papers. Given the wide range of study designs included, data synthesis was narrative. We included 29 papers (Table 1). We used the Scottish Intercollegiate Guidelines Network criteria for assigning level of evidence.

Records
Three papers were case control and one a cohort study, that is, level 2 studies. The remaining 25 were level 3, non-analytic case series. Ten studies were prospective and the remaining 19 retrospective reviews of medical records. All included studies were of low or very low quality.

| Study and participant characteristics
Eight studies were conducted in the United States, four studies in Germany, and two in Denmark, Switzerland, Italy, and Brazil (  Table 2.

Evidence grading
Armstrong et al. 10 Level 3 Chantelau. 12 Level 2 Chantelau and Richter. 26 Level 3 Christensen et al. 13 Level 3 de Souza. 8 Level 3 Dixon et al. 27 Level 3 Fabrin et al. 14 Level 3 Holmes and Hill. 28 Level 3 O'Loughlin et al. 29 Level 3 Osterhoff et al. 30 Level 2 Pakarinen et al. 31 Level 3 Parisi et al. 32 Level 3 Renner et al. 15 Level 2 Ruotolo et al. 16 Level 3 Pinzur et al. 9 Level 3 Saltzman et al. 33 Level 3 Sinacore. 7 Level 3 Stark et al. 6 Level 3 Thewjitcharoen et al. 34 Level 3 Verity et al. 35 Level 3 Visan et al. 36 Level 3 Wukich et al. 37 Level 2   21 The third study compared bone marrow oedema on MRI at baseline and after 4 months, and correlated this to symptoms of CN in 13 participants. There was a statistically significant decrease in bone oedema over 4 months, with a statistically significant correlation between pain and soft tissue oedema and the bone marrow oedema over the same timescale. 24 Two studies evaluated infrared thermometry to identify disease remission. 23,38 The first study described in detail the protocol for measuring temperature using the Exergen Model DT 1001. They controlled for ambient room temperature, allowed a 15 minute acclimatization period, and measured seven sites on the foot, compared with the contralateral limb as the physiologic control, at monthly intervals. 38 Casting was discontinued based on reduction or absence of clinical signs of inflammation, radiologic signs of healing and when the temperature difference between feet had stabilized with a cut-off point of less than 4 F (2.2 C) difference. The authors report that the choice of the cut-off figure was based on clinical experience The second study referenced the protocol described by Armstrong and Lavery 38 for measuring temperature but used the Minitemp, Raytec 23 to monitor temperature.
Casting was discontinued when the temperature difference between feet was recorded as less than 2 C.
One study evaluated Doppler spectrum analysis as a novel diagnostic tool for planning treatment. 25 The study compared the Doppler spectra of the first metatarsal arteries in both feet using a 10 MHz linear ultrasound probe (ATL HDI3000 or HDI5000; ATL, Bothel, Washington). The Doppler spectra in the unaffected limb were triphasic, compared to the affected limb which showed monophasic forward flow. The Doppler spectra analysis was repeated every 2 weeks in the affected limb until it returned to normal. At this point, participants either started weight-bearing or underwent surgical reconstruction of the ankle joint. The authors concluded that Doppler spectra analysis of the foot may be used as a guide to begin weight bearing.
They reported a discrepancy between the two monitoring techniques: only four out of 15 patients had X-rays which showed healing when the foot was healed according to the Doppler Spectra analysis.
In the final study, a subset of eight participants from a larger study received three monthly three-phase quantitative bone scans of both feet for a maximum of 12 months. They compared the ratio of isotope uptake between feet, between the affected foot and the tibia and compared isotope uptakes to the clinical indicators of inflammation. There was strong correlation between temperature difference and the ratio of isotope uptake in the affected vs unaffected foot, the perfusion of the affected foot in the dynamic phase and the isotope uptake in the delayed phase of the bone scans. 22 The study also reported on the change in temperature difference between the affected and unaffected foot from baseline 3.3 C, at 6 months 1.3 C, and at 12 months 0.8 C noting a progressive decrease over time. 22  In the remaining 22 studies, the primary aim was to evaluate the outcomes of CN but they described the monitoring techniques used (

| Financial implications to healthcare providers and clinical feasibility of different techniques
No studies reported the cost of the monitoring used in terms of capital cost to purchase equipment.

| Safety considerations, and participant acceptability of different techniques
Ten out of 29 studies used MRI as a monitoring tool for identifying remission of acute CN. Of these, four reported using contrast during the MRI in all or some images at the radiologist's discretion. 11,16,24,26 A further four studies used advanced methods of radiological imaging which require the use of contrast media. 12,13,16,22 Only one of these 14 studies which used contrast specifically reported on the incidence of adverse events from the administration of the contrast, reporting no adverse events. Another study reported using bone biopsy as a diagnostic aid to confirm CN, but this was not used in monitoring. 30 They did not report any safety considerations that may be relevant to this technique. X-rays are associated with exposure to ionizing radiation, but their potential risk was not discussed in any  The key finding is the lack of a consistent approach to monitoring in CN. Common techniques included X-ray, temperature monitoring and MRI. Techniques were poorly described, and where the information was reported there was variability in the devices used and how the technique was applied. It is not clear whether the devices used were validated for the temperature ranges commonly found in feet.
Some studies still rely on subjective measures of temperature difference between feet to monitor CN. 7,8,15,21,26,35 The first paper included in this review which used temperature measurement for monitoring in CN was published in 1997. 38 The authors report that the cut-off point of 4 F (2.2 C) for healing was not evidence-based but it appears to have been adopted as the standard for clinical decision making in subsequent studies and guidelines. 2,39 This protocol has not been validated and other studies have not specified sites, repeated measures, or acclimatization times making evaluation of studies using this technique difficult.
We found a lack of evidence on the sensitivity, specificity, costeffectiveness, safety and patient acceptability for all monitoring techniques. There is continued uncertainty about the relationship between monitoring techniques and treatment outcomes.
In the absence of reliable evidence, we are unable to recommend any changes to current national 39 and international guidance 2 which are predominantly based on level IV evidence, that is, expert opinion.
The strengths of our systematic review include the broad range of inclusion and exclusion criteria for study type, which allowed us to describe the variability in the current approach to monitoring CN in research as well as clinical practice. Screening and data extraction were completed by two researchers who are experienced podiatrists.
Our review also had some limitations: we did not search the grey literature. We limited searches to English language, we acknowledge that this may mean we missed some relevant studies and potentially introduced bias into the review. However, we feel that the impact of this would be relatively small.
In the 1990s, it was acknowledged that using subtle changes in skin temperature to inform clinical decisions may not be an accurate way to monitor CN 20 but this is still widely used in clinical practice.
Further high quality research is needed to identify the optimal method of monitoring CN. We recommend that researchers accurately describe the population at baseline, standardize definitions for diagnosis and outcome measures, and provide detailed protocols for monitoring techniques in future research.
MRI as a monitoring tool for CN is increasingly acknowledged as a potentially more accurate method for monitoring and this is supported by the studies we included. 11,21,24 This warrants further investigation. An ongoing randomized feasibility study aims to explore the feasibility of using serial MRI without contrast in the monitoring of CN to decide whether a large-scale trial is warranted https://doi. org/10.1186/ISRCTN74101606. 40

| CONCLUSION
Multiple techniques have been used to evaluate remission in acute CN, but the quality of published studies to support any one technique is low or very low. Uncertainty therefore remains about the effectiveness of the different monitoring techniques, and whether the different monitoring techniques influence time to remission and recurrence rates. Therefore, we are unable to make recommendations for clinical practice.
There is an urgent need for high-quality studies to identify the most accurate, safe and cost-effective monitoring techniques in CN.