Pattern of diabetic neuropathic arthropathy ... - Semantic Scholar

Foot/Ankle

Pattern of diabetic neuropathic arthropathy associated with the peripheral bone mineral density S. A. Herbst, K. B. Jones, C. L. Saltzman From the University of Iowa Hospitals and Clinics, Iowa City, USA


S. A. Herbst, MD
K. B. Jones, MD
C. L. Saltzman, MD Department of Orthopaedic Surgery, University of Iowa Hospitals and Clinics, 200 Hawkins Drive, Iowa City, Iowa 52242, USA. Correspondence should be sent to Dr C. L. Saltzman. ©2004 British Editorial Society of Bone and Joint Surgery doi:10.1302/0301-620X.86B3. 14593 $2.00 J Bone Joint Surg [Br] 2004;86-B:378-83. Received 30 April 2003; Accepted after revision 2 September 2003 378

The relationship between the bone mineral density (BMD) and Charcot arthropathy is unclear. Prospectively, 55 consecutive diabetic patients presenting with a Charcot arthropathy of the foot or ankle were classified as having a fracture, dislocation, or a combination fracture-dislocation pattern of initial destruction. In these groups we used dual-energy x-ray absorptiometry to compare the peripheral bone of the affected and unaffected limbs. The clinical data relating to diabetes and related major comorbidities and the site of the arthropathy (ankle, hindfoot, midfoot, forefoot) were also compared. There were 23 patients with a fracture pattern, 23 with a dislocation pattern, and nine with a combination. The age-adjusted odds ratio for developing a Charcot joint with a fracture pattern as opposed to a dislocation pattern in patients with osteopenia was 9.5 (95% confidence interval 2.4 to 37.4; p = 0.0014). Groups also differed as to the site of the arthropathy. Fracture patterns predominated at the ankle and forefoot whereas dislocations did so in the midfoot. Diabetic Charcot arthropathy of the foot and ankle differs according to the pattern of the initial destruction. The fracture pattern is associated with peripheral deficiency of BMD. The dislocation pattern is associated with a normal BMD.

Diabetic neuropathic arthropathy is a destructive process of the bony components of a denervated joint. Although it was initially associated with tertiary syphilis,1-3 diabetes is now the main cause of neuroarthropathy in the developed world. Charcot arthropathy has been estimated to affect at least 1 in 680 diabetics and may occur in up to 29% of diabetics with peripheral neuropathy.4,5 Its incidence is suspected to be rising with the increasing prevalence of diabetes.6 Regional osteopenia is considered to be a classic finding of Charcot arthropathy.7 A few small studies have shown that patients have reduced bone mineral density (BMD) in their ipsilateral and contralateral peripheral skeleton.8-11 Whether peripheral deficiency of BMD is a risk factor for the development of Charcot neuroarthropathy, results from changes, or occurs with changes from a common aetiology is as yet unknown. Clinically, good results with bracing or early surgery have been reported for Charcot arthropathy in the midfoot, but not in the ankle.12-17 The cause for the discrepancy is unknown. Recognising that Charcot arthropathy at the ankle often appeared to result from fracture, whereas neuroarthropathy of the

midfoot typically had an associated dislocation, we hypothesised that Charcot changes may be subclassified by the initial pattern of injury, and further that a fracture pattern may be associated with a deficient peripheral BMD.

Patients and Methods Patients newly diagnosed as having Charcot arthropathy of the foot and ankle and having current pharmacological therapy for diabetes were eligible for enrolment in the study. Charcot arthropathy was defined as a clinically inflamed, erythematous, or swollen foot or ankle with radiological correlation of destruction of the bone and joint, fragmentation or remodelling. Patients newly presenting with already advanced Charcot changes, were not excluded. They were excluded if they had systemic inflammation, fever, an elevated ESR or level of C-reactive protein, or osteomyelitis. Informed consent was given by each patient. We recorded age, gender, the type and duration of the diabetes, the method of glycaemic control, the duration of Charcot symptoms and the presence of retinopathy, nephropathy, and neuropathy. We defined the neuropathy as loss of light touch sensation or proprioception of the involved foot. Nephropathy included a THE JOURNAL OF BONE AND JOINT SURGERY

PATTERN OF DIABETIC NEUROPATHIC ARTHROPATHY ASSOCIATED WITH THE PERIPHERAL BONE MINERAL DENSITY

Fig. 1 Radiograph showing the fracture pattern of Charcot arthropathy of the ankle.

documented history of proteinuria (not microalbuminuria), dialysis or renal transplantation. Retinopathy was determined from the patient’s notes and review of ophthalmological records.

Fig. 2a

Radiography. At presentation, weight-bearing (when possible) radiographs were obtained for each patient. Evolving neuroarthropathy was followed by serial radiography. Previous radiographs were also reviewed when available. The earliest available radiographs were used to group patients by the initial pattern of destruction: fracture, dislocation or combined fracture-dislocation. To be classified as a fracture, bony injury alone should be present (Fig. 1) but the joint should be intact without evidence of subluxation or dislocation. To be classified as a dislocation, separation of the joint surfaces only should be seen without evidence of bony injury (Fig. 2). The presence of small flecks of avulsion fragments was acceptable for inclusion within this group. If dislocation with more substantial associated bony injury was present the patient was placed in the combined fracture-dislocation group (Fig. 3). The Charcot process was also clarified in relation to four sites: the ankle, including the talar dome; the hindfoot, consisting of the calcaneum, the subtalar, talonavicular, and calcaneocuboid joints; the midfoot, distal to the talonavicular and calcaneocuboid joints and proximal to the proximal metadiaphysis of the metatarsals; and the forefoot which included more distal regions (Fig. 4). The Eichenholtz classification was also recorded from radiographs at presentation (Table I).18 The inter- and intra-observer reliability of each of these measurements was examined. Two authors (CLS and SAH) first reviewed 20 radiographs and reached a consensus

Fig. 2b

Radiographs showing examples of a dislocation injury of the midfoot which led to Charcot arthropathy. VOL. 86-B, No. 3, APRIL 2004

379

380

S. A. HERBST, K. B. JONES, C. L. SALTZMAN

Fig. 3 Radiograph showing a combined fracture dislocation injury which led to Charcot arthropathy. There is a talonavicular dislocation (black arrow) the fracture through the cuboid (white arrow).

Table I. Radiographic classification of Charcot arthropathy according to Eichenholtz18 Stage

Radiological features

I dissolution

Demineralisation of regional bone, peri-articular fragmentation, dislocation of joint II coalescence Absorption of osseous debris in soft tissues, organisation and early healing of fracture fragments, formation of periosteal new bone formation III resolution Smoothing of edges of large fragments of bone, sclerosis, osseous or fibrous anklyosis

Table II. Weighted kappa values (with 95% CI) for inter- and intraobserver reliability Observation

Inter-rater

Intra-rater

Classification of Charcot arthropathy according to the anatomical site in the foot or ankle. (a) ankle; (b) hindfoot; (c) midfoot; (d) forefoot.

Location Eichenholtz Pattern of initial destruction

1.0 0.77 (0.52 to 1.02) 0.81 (0.71 to 0.90)

1.0 0.94 (0.82 to 1.06) 0.97 (0.93 to 1.01)

opinion on each. Next, radiographs from 20 other patients were scored by each independently. One author then repeated the scoring on a separate day. Weighted kappa values were calculated for the inter- and intraobserver reliability of each assessment (Table II).19 There was substantial reliability for all three.20 Measurement of peripheral BMD. Dual-energy x-ray absorptiometry (DXA) was performed on all patients. The contralateral femoral neck was scanned unless the patient had bilateral Charcot changes (possibly confounding disuse osteopenia), or the radiologist thought that obesity would hinder the accuracy of DXA at the femoral neck. For these exceptions,10 the distal radius was scanned. A single machine (Hologic QDR 4500A, Hologic Inc, Bedford, Massachusetts), with interscan variability under 1%, was

used for all scans.21 The BMD (g/cm2), the t-score (SD from site- and gender-matched healthy young adult means, and the z-score (SDs from age-, site-, and gender-matched means) were recorded. Statistical analysis. The three groups were compared regarding the t-scores, age, and the duration of diabetes and of Charcot disease using one-way ANOVA followed by Tukey’s test (α = 0.05). Gender, the type of diabetes, the presence of diabetic comorbidities, and the site of neuroarthropathy were compared using Fisher’s exact test. Multiple logistic regression analysis evaluated gender, age, and the osteopenic score as predictors of the neuropathic pattern. To achieve 80% power to detect a relative risk of 8 (with α = 0.05) so that WHO criteria for osteopenia would be associated with fracture as opposed to the dislocation

Fig. 4

THE JOURNAL OF BONE AND JOINT SURGERY

PATTERN OF DIABETIC NEUROPATHIC ARTHROPATHY ASSOCIATED WITH THE PERIPHERAL BONE MINERAL DENSITY

381

Table III. Details of the 55 patients and statistical differences by number and percentage where applicable

Number Men Women Mean (± SD) age in years Mean (± SD) duration of diabetes in years Type-I diabetes Type-II diabetes Use of insulin Neuropathy Nephropathy Retinopathy Mean (±SD) duration of Charcot in months Eichenholtz stage I stage II stage III

Fracture pattern

Dislocation pattern

23 6 (26) 17 (74) 49.1 ± 15.6 20.0 ± 12.0 10 (43) 13 (57) 20 (87) 19 (83) 12 (52) 15 (65) 20 ± 30

23 9 (39) 14 (61) 54.8 ± 9.1 21.0 ± 12.0 4 (17) 19 (83) 18 (78) 23 (100) 11 (48) 14 (61) 17 ± 22

13 (57) 7 (30) 3 (13)

16 (70) 6 (26) 1 (4)

20 18

Fracture Fracture/dislocation Dislocation

16 14 12 10 8 6 4 2 0

Ankle

Hindfoot

Midfoot

Forefoot

Fig. 5 Frequencies of each pattern of arthropathy at different sites in the foot and ankle.

pattern, 20 patients were needed for each of these two groups. Secondary analysis to determine if t-score variance was associated with other variables used the Pearson correlation coefficient for continuous variables and the t-test for categorical variables. Further analyses, stratified for the type of diabetes, were performed for all variables with p < 0.2 on the initial secondary analysis, to test for differing trends between the two types of diabetes.

Results Between June 1999 and March 2001, 55 patients with 61 involved feet or ankles (six bilateral) consented to be included in the study. Twenty-three had the fracture patVOL. 86-B, No. 3, APRIL 2004

p value (fract v dis)

0.53 0.14 0.68 0.11 0.70 0.11 1.0 1.0 0.65

0.43

Combined fracturedislocation 9 5 (56) 4 (44) 51.4 ± 10.9 12.0 ± 8.5 2 (22) 7 (78) 8 (89) 9 (100) 3 (33) 4 (44) 24 ± 36 2 (22) 6 (67) 1 (11)

p value (3-way)

0.28 0.31 0.17 0.17 0.88 0.06 0.67 0.62 0.78

0.13

tern, 23 the dislocation pattern, and nine the combined fracture-dislocation pattern. The groups did not differ with regard to gender, the duration of diabetes, the use of insulin, the presence of retinopathy or nephropathy, the duration of Charcot symptoms, or the Eichenholtz stage (Table III). There was a trend towards significance for differences between the fracture and dislocation groups for the type of diabetes, the presence of neuropathy and age. Site of Charcot arthropathy. The site of neuroarthropathy differed significantly between the fracture and dislocation groups. Overall, Charcot changes were found in 12 ankles (19%), 17 hindfeet (28%), 30 midfeet (50%), and two forefeet (3%). All patients with changes in the ankle and forefoot were in the fracture group (Fig. 5). The midfoot primarily contributed to dislocations. The hindfoot was affected in all three groups. Peripheral BMD. The fracture group had significantly lower t-scores than the dislocation group (p = 0.0011; Fig. 6). Sixteen fracture-pattern patients (74%) had t-scores of less than -1.0, the threshold for osteopenia according to the WHO criteria.22 Nine of these (39% of the entire fracturepattern group) had t-scores below -2.5, the WHO threshold for osteoporosis. Only four patients with dislocation had tscores below the threshold for osteopenia. The age-adjusted odds ratio of a patient with osteopenia according to the WHO criteria having a fracture rather than a dislocation was 9.5 (95% confidence interval (CI) 2.4 to 37.4; p = 0.0014). The fracture-pattern group also had a significantly lower mean z-score than the dislocation group (p = 0.0009). Secondary analysis. Among all the groups together, the t-score was not significantly related to gender (p = 0.51), the presence of nephropathy (p = 0.27) or neuropathy (p = 0.76), the duration of symptoms of Charcot arthropathy (p = 0.55), or the Eichenholtz stage (p = 0.95). Insignificant trends towards correlation were noted between the t-score and the duration of diabetes (p = 0.08), the use of

382

S. A. HERBST, K. B. JONES, C. L. SALTZMAN

1.5 1 0.5

t-score 0.06

-0 n=23

-0.5

-0.35 n=9

-1 -1.5 -2 -2.5

*Fracture significantly lower than dislocation (p < 0.005)

-1.48 n=23

-3 -3.5

Dislocation* Fracture/dislocation

Fracture*

Fig. 6 Variation of the DXA-derived BMD t-score between the fracture, the dislocation, and the combined fracture-dislocation patterns of Charcot arthropathy.

insulin (p = 0.09), and age (p = 0.09). The trend towards correlation with age was positive (BMD increasing with age). The t-scores significantly correlated with the presence of retinopathy (p = 0.0016) and type-I diabetes (p < 0.0001). When stratified for type of diabetes the use of insulin and the duration of diabetes did not correlate with tscore (p = 0.28 and 0.74, respectively). Retinopathy when stratified for type of diabetes showed a trend towards correlation with t-score (p = 0.06). Age did not correlate with the t-score in type-II diabetes (p = 0.5), but did in type-I (p = 0.03) with probably spuriously improved BMD with age.

Discussion Reduced peripheral, but not axial BMD, has been demonstrated in patients with Charcot arthropathy.10,11 A causal relationship between neuroarthropathy and regional osteopenia has been considered. Our study simply refined the question, by dividing feet and ankles into three subgroups. One type, the fracture pattern, was associated with pronounced deficiency in systemic (not site-specific) peripheral BMD. The other two groups had a normal BMD. Previous studies may have found a reduced BMD, but not frank osteopenia because they grouped all three patterns of Charcot arthropathy together. The modified case-control design primarily examined the effect of osteopenia defined on WHO criteria on the development of fracture- versus dislocation-patterns. Osteopenia was highly predictive of the fracture-pattern destruction. Osteoporosis was found only in patients with the fracture pattern of Charcot arthropathy. It is likely that the osteopenia in patients with a fracture pattern of Charcot arthropathy does not result from

regional arthropathy. The BMD was measured in the femoral neck on an unaffected limb, or in the distal radius. Also, in secondary analysis, there was no relationship between the BMD and the duration or Eichenholtz stage of Charcot arthropathy. A number of known risk factors for osteopenia were highlighted, including type-I diabetes, microangiopathy, hypothyroidism, abuse of alcohol, anorexia nervosa, and end-stage renal disease. Osteopenia from these and other causes, as measured by peripheral DXA, is a predictor of the risk for fracture, which is measurable after fracture has occurred. The evidence for this association is strongest in women.23-25 These data suggest that it also predicts the Charcot pattern in patients who will eventually develop neuroarthropathy. This is important because, as distinct from a possible but yet unproven regional osteopenia from neuroarthropathy, our study identifies a possibly modifiable systemic risk factor for developing the fracture pattern and general arthropathy around the ankle. The strength of the study includes the simplicity of the classification of the pattern of arthropathy, the relatively large number of patients, and the inclusion of patients with type-I and type-II diabetes. Weaknesses include the relative paucity of patients in the combination fracturedislocation group, the necessity of using distal radius DXA scans in ten of the patients, and the inclusion of patients presenting at different stages of disease because of the tertiary nature of the clinic’s referral base. Regarding the ten patients for whom the distal radius was used instead of the contralateral femoral neck (four because of obesity, six because of bilateral neuroarthropathy), a subanalysis showed that the results of the study were not changed by inclusion or exclusion of these patients, suggesting that scans of either site represent the peripheral BMD systemically, rather than the BMD associated with the specific region. Neuropathic ankle fractures are notoriously difficult to treat. High rates of complications are found with both nonoperative and operative treatment.12-15 An underlying BMD deficiency may contribute to these difficulties. Conversely, the recent successful results16,17 of early arthrodesis of the midfoot and hindfoot Charcot deformities may be related to a normal BMD in dislocation-pattern injuries prevalent at these anatomical sites. In conclusion, an identifiable subgroup of diabetic patients with Charcot arthropathy have an initial fracture pattern of destruction. They most commonly present with difficulties at the ankle and have marked peripheral BMD deficiency, which may adversely affect treatment. Further study to determine the treatment of osteopenia as a risk factor for this pattern of Charcot arthropathy in diabetes may advance preventative measures for these often devastating problems. This study was supported by NIH grant award 5 R01 NR007721-04. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. THE JOURNAL OF BONE AND JOINT SURGERY

PATTERN OF DIABETIC NEUROPATHIC ARTHROPATHY ASSOCIATED WITH THE PERIPHERAL BONE MINERAL DENSITY

References 1. Charcot JM. Sur quelques arthropathies qui paraissent deponder d’une lesion du cerveau ou de la moele epiniere. Arch Physiol Norm Pathol 1868;1:161-71. 2. Jordan W. Neuritic manifestations in diabetes mellitus. Arch Intern Med 1936;57: 307-66. 3. Kelly M. Musgrave’s de arthritide symptomatica. Bull Hist Med 1965;37:372-6. 4. Cofield RH, Morrison MJ, Beabout JW. Diabetic neuroarthropathy in the foot: patient characteristics and patterns of radiographic change. Foot Ankle 1983;4: 15-22. 5. Sinha S, Munichoodappa CS, Kozak GP. Neuro-arthropathy (Charcot joints) in diabetes mellitis (clinical study of 101 cases). Medicine (Baltimore) 1972;51: 191-210. 6. National Diabetes Data Group. Diabetes in America. 2nd ed. NIH Publication, 1995. 7. Purewal T. Charcot’s diabetic neuroarthropathy: pathogenesis, diagnosis and management. Practical Diabetes Int 1996;13:88-91. 8. Childs M, Armstrong DG, Edelson GW. Is Charcot arthropathy a late sequela of osteoporosis in patients with diabetes mellitus? J Foot Ankle Surg 1998;37:437-9. 9. Cundy TF, Edmonds ME, Watkins PJ. Osteopenia and metatarsal fractures in diabetic neuropathy. Diabet Med 1985;2:461-4. 10. Jirkovska AP, Kasalicky P, Boucek P, Hosova J, Skibova J. Calcaneal ultrasonomentry in patients with Charcot osteoarthropathy and its relationship with densitometry in the lumbar spine and femoral neck and with markers of bone turnover. Diabet Med 2001;18:495-500. 11. Young MJ, Marshall A, Adams JE, Selby PL, Boulton AJ. Osteopenia, neurological dysfunction, and the development of Charcot neuroarthropathy. Diabetes Care 1995;18:34-8. 12. Bibbo C, Lin SS, Beam HA, Behrens FF. Complications of ankle fractures in diabetic patients. Orthop Clin North Am 2001;32:113-33.

VOL. 86-B, No. 3, APRIL 2004

383

13. Chi TD, McWilliam J, Gould JS. Lateral plate-washer technique for revision tibiocalcaneal fusion. Am J Orthop 2001;30:588-90. 14. Marks RM. Complications of foot and ankle surgery in patients with diabetes. Clin Orthop 2001;391:153-61. 15. Pinzur MS, Kelikian A. Charcot ankle fusion with a retrograde locked intramedullary nail. Foot Ankle Int 1997;18:699-704. 16. Simon SR, Tejwani SG, Wilson DL, Santner TJ, Denniston NL. Arthrodesis as an early alternative to nonoperative management of Charcot arthropathy of the diabetic foot. J Bone Joint Surg [Am] 2000;82-A:939-50. 17. Stone NC, Daniels TR. Midfoot and hindfoot arthrodeses in diabetic Charcot arthropathy. Can J Surg 2000;43:449-55. 18. Eichenholtz S. ed. Charcot Joints. 1966, CC Thomas: Springfield, IL. 19. Fleiss J, Cohen J. The equivalence of weighted kappa and the intraclass correlation coefficient as a measure of reliability. Education and Psychological Measurement 1973;33:613-9. 20. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. 21. Hologic 4500A Product Specification Sheet. 22. Kanis JA. Assessment of fracture risk and its application to screening for postmenopausal osteoporosis: synopsis of a WHO report. WHO Study Group. Osteoporosis Int 1994;4:368-81. 23. Black DM, Cummings SR, Genant HK, et al. Axial and appendicular bone density predict fractures in older women. J Bone Miner Res 1992;7:633-8. 24. Cummings SR, Black DM, Nevitt MVC, et al. Bone density at various sites for prediction of hip fractures: the study of Osteoporotic Fractures Research Group. Lancet 1993;341:72-5. 25. Melton LJ, Atkinson EJ 3rd, O’Fallon WM, Wahner HW, Riggs BL. Long-term fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res 1993;8:1227-33.