Quantification of joint mobility limitation in adult type 1 diabetes

Patients and Clinical Setting

The study was conducted at the Diabetes unit, KEM Hospital Research Centre, a tertiary care specialized unit in Pune, India. We screened consecutive adult patients (>18 years of age) with type 1 diabetes attending the Type 1 diabetes clinic, from March 2021 to December 2022. Type 1 diabetes was initially classified clinically, viz diagnosed <30 years, ketosis at presentation, and clinical dependence on insulin treatment; later confirmed with C-peptide measurements and the presence of islet autoantibodies. We excluded pregnant patients, those who needed hospital admission, and those with previous hand trauma or concurrent inflammatory arthritis involving the hand joints. Demographic information including age, gender, residence, education, and occupation were recorded. Those who worked in agriculture, manual labor or operating heavy machinery for more than 4 hours a day were classified as manual work. Those who worked on computer keyboards for more than six hours a day were classified as keyboard workers. Smoking (current, previous, never) and alcohol habits were recorded.

Data concerning the course of diabetes included date of diagnosis, insulin compliance, and the use of alternative medications. Evidence of micro- and macro-vascular complications was extracted from patient files: including evidence of retinopathy in fundus photographs, nephropathy from the presence of proteinuria and/or end stage renal disease, neuropathy based on biothesiometry and clinical composite score Michigan Neuropathy Screening Instrument, and coronary artery disease. Common comorbidities, such as hypertension and hypothyroidism, were recorded; we also asked for conditions that could possibly contribute to profibrotic manifestations including epilepsy, systemic sclerosis, skin disease involving the palms or hands, and malignancy. Similarly, we asked for a history of treatments that could possibly be fibrogenic, including methotrexate, amiodarone, aspirin, statins, and anti-epileptics.

Anthropometric measurements (height and weight) and blood pressure were recorded using standard procedures. We also noted the presence of keloid scars and lipo-hypertrophy or atrophy at insulin injection sites. We used skin autofluorescence as an indirect measurement of advanced glycation end-product (AGE) measurements using the AGE reader, that has been previously linked to complication burden in type 1 diabetes.[18] The reader provides a quantitative estimate, as well as a risk category calculated considering age and duration of diabetes. These measurements were performed on the forearm of the non-dominant hand, in standardized light conditions.

Quantitative measures of hand manifestations

We screened for hand involvement using a structured history, clinical examination, and measurements by trained research staff. The musculoskeletal history included presence of pain on the palmar surface of the hand, symptoms of compressive neuropathy (sensory and motor), grip difficulty, finger triggering, and perceived stiffness and tightness of palmar skin. Hand function was measured using the Duruoz Hand Index (DHI), a self-reported questionnaire that assesses activity limitation in 18 daily activities each measured on a visual analogue scale from 0-5 with possible scores ranging from zero to 90, that has been validated for use in diabetes related hand conditions.[19]

Hand examination included the prayer sign, defined as a visible gap between the two palms with an inability to approximate them fully. We measured the distance between the two 5th metacarpophalangeal (MCP) joints and 5th proximal interphalangeal (PIP) joints, viewed from the ulnar aspect.[Figure 1B] As a measure of limited joint mobility, we recorded maximum possible passive extension at the MCP joint (2nd to 5th of both hands) until restriction or pain, with the palm approximated on a flat surface, using a protractor.[Figure 1A] Mean passive MCP extension was calculated for each hand. The presence of flexor tendon thickening, nodularity, triggering, and crepitus was noted. For the presence of carpal tunnel syndrome, we examined for the Tinel sign in both hands, the Phalen sign and sensation in the median nerve distribution, recorded as normal, reduced, or absent. Hand grip strength was measured for both hands, using a Jamar hand dynamometer (Patterson Medical, Warrenville, IL) Three readings were taken for each hand and a mean was calculated.

• Download figure
• Open in new tab

Figure 1:

Measurement methodology, Clinical and MRI. A: measurement of maximum passive extension at MCP joint of left second finger, here showing 50 degrees extension. B: Measurements of distance between fifth proximal interphalangeal joints and fifth MCP joints as a quantification of the prayer sign. C: Measurement of tenosynovial thickness at the level of carpal tunnel; D. Measurement of median nerve area at the level of the carpal tunnel E. Measurement of skin and subcutaneous fat and F: Measurement of palmar fascia overlying the fourth flexor tendon on PDFS images. MRI: magnetic resonance imaging; PDFS: Proton density fat saturation

A physician (rheumatologist or diabetologist) examined all patients independently without access to the above measurements, and opined if one or more hand manifestation was present (LJM, flexor tenosynovitis/trigger finger, CTS, DD) was present, along with a physician-perceived severity (mild, moderate, severe). Inter-rater reliability between two physicians (SD, SP), seen in 30 patients, was 0.88.

Selection of patients for MRI scans

Regardless of the cause of limited joint restriction, viz., LHM, DD or FT, or the number of fingers affected, we expected mean passive MCP extension to be compromised. Thus, non-dominant hand mean passive MCP extension was used as a metric of joint mobility restriction to segregate patients into 20-degree bins (0-20 degrees, most severe joint mobility limitation; 20-40 degrees, 40-60 degrees and more than 60 degrees).

Patients were approached for an MRI scan of the hand, all performed at the Star Imaging Research Centre using a 3T MRI Superconducting system with eight channel extremity coil (Ingenia Release 5, Philips Healthcare, Amsterdam, The Netherlands). Patients who agreed were invited to the Diabetes Unit, where earlier clinical findings were confirmed. Contraindications to MRI were excluded (metallic bone, cardiac, cochlear, or dental implants, metallic intrauterine contraceptive devices, pregnancy, and claustrophobia). Two patients were on insulin pumps, both the pump and sensor-transmitters were removed for the scan. Random plasma glucose measurements were performed, and the diabetologist (SD) managed insulin dosages accordingly, to prevent hypoglycemia whilst inside the MRI scanner. MRI of the non-dominant hand was performed, unless hand involvement was unilateral, in which case, the involved hand was imaged.

MR Sequences and measurements

The following MR sequences were performed: Axial and coronal T2 weighted images, pre-contrast fat saturated T1-weighted axial coronal with fat-saturation. All MRI scans were read by the same musculoskeletal radiologist (PM) who provided a qualitative report (altered signals, thickening, edema) on the status of bones, tendons, joints, median nerve, and other salient findings. In addition, quantitative measurements were performed by one of two trained researchers (SC, SJ) on axial images, in addition to the radiologist (PM). The readers had good internal consistency (Cronbach alpha 0.95) and agreement with each other (correlation coefficient 0.78). Measurements included tenosynovial thickness for the four flexor digitorum longus tendons and the flexor hallucis longus tendon, at four levels: at the metacarpophalangeal joint, the midpoint of the proximal phalanx, midpoint of the metacarpal bone, and the midpoint of the carpal tunnel.[Figure 1C] Mean tenosynovial thickness was calculated as average of these 20 data points. In addition, the thickness at a point of visually perceived maximum thickness for each tendon sheath was recorded. Skin and subcutaneous tissue thickness was measured at 4 points at the level of MCP joint, and four points at the level of the carpal tunnel, and an average was calculated.[Figure 1E] Palmar fascia was visible only when thickened; it was measured in those who had nodular thickenings (DC) at the site of thickening and was considered zero in others.[Figure 1F] The median nerve cross sectional area was measured at the level of the carpal tunnel outlet and median nerve signal abnormalities were noted.[Figure 1D] An instinctive ‘total hand fibrosis’ score was calculated by adding mean tenosynovial thickness, palmar fascia thickness and palmar skin thickness. A similar score was also calculated using Z-scores of mean tenosynovial thickness, palmar fascia thickness, and palmar skin thickness (total hand fibrosis Z-score). Patients with least joint flexibility had highest skin thickness on hand MRI.[Figure 2]

• Download figure
• Open in new tab

Figure 2:

MRI findings in two patients with MCP extension restriction. A1: A patient with severely restricted MCP extension – mean 10 degrees. A2: MRI demonstrates considerable palmar skin thickening alone on T1 weighted axial images, seen at mid-metacarpal level; there is no tenosynovial or palmar fascia thickening. B1: A patient with bilateral Dupuytren’s contracture on the fourth finger (black arrow); PDFS scan shows Ill-defined PDFS hyperintense soft tissue thickening in relation to flexor thickening also involving A1 pulley and the palmar fascia.

Statistical analysis

Data are presented as mean (standard deviation) and median (interquartile range) as appropriate for the variable. Patients who were judged to have diabetic cheiroarthropathy were compared with those without, using t-test. Patients with different levels of MCP extension limitation in 20-degree bins (0-20 degrees, 20-40 degrees, 40-60 degrees, and >60 degrees) were compared with each other with respect to demographics, diabetes characteristics and MRI characteristics using ANOVA; we report p value for trend, with <0.05 considered significant. MRI quantitative descriptors are described as mean (standard deviation) for each group. Multiple linear regression was used for structural determinants of hand stiffness: we used average MCP angle of the imaged hand as the dependent variable, while average skin thickness, average subcutaneous thickness, tenosynovial thickness, palmar fascia thickness, and median nerve cross-sectional area were used as predictor variables. We first standardized the independent and the dependent variables of the dataset into their corresponding Z-scores. All statistical analysis were performed using SPSS (IBM corporation, Armonk, NY) and R (R Foundation, Vienna, Austria).

Cohort characteristics [Table 1]

We examined 237 adults with type 1 diabetes (90 males, median age 26.8 years). Most were college students, 15 (6%) operated heavy machinery, and 3 worked on keyboards for more than 6 hours. All had received their diagnosis of diabetes in childhood and had median duration of diabetes of 13.7 years. All took injectable insulin, and 57 also received metformin. Two patients used insulin pump. One-fifth of the cohort (20 patients, 8.5%) had retinopathy, 28 (11.8%) had nephropathy, and 45 (19%) had neuropathy on clinical examination.

Musculoskeletal history and hand examination

Only nine (3.8%) complained of hand pain at the time of clinical assessment; 26 had a history of trigger finger. Sixteen (6.8%) complained of paresthesia and numbness in the hands. As a group, mean passive extension at the MCP joint was 50.8 degrees (SD) on the right hand and 54.2 degrees on the left. Prayer sign was seen in 73 (30%). Fifty-three had carpal tunnel syndrome on clinical examination. Seventy-nine (33.8%) had a cheiroarthropathy manifestation on physical examination; 30 had flexor tenosynovitis, 46 had LJM, 24 had CTS, and three had DC. Twenty-four had more than one condition. Patients with a physician-diagnosed cheiroarthropathy had a significantly different average MCP extension angle (39 degrees vs 61 degrees, p<0.01) and a significantly higher DHI (1.67 vs 0.21, p<0.01).

Patient characteristics as divided by MCP extension limitation

Average MCP extension (2nd to 5th finger) of the non-dominant hand had a range of zero to 88 degrees; patients were grouped into four bins of 20 degrees each.[Table 2] Most patients (103, 43%) had mean extension in the 40-60 degrees range, while 64 (27%) had extension limited to less than 40 degrees. The group with the most severe limitation (<20 degrees) was predominantly male (11/15, 73%), while all the other groups had higher numbers of female participants. Groups with restriction (0-20, 20-40) tended to be older than the groups without and had diabetes for nearly a decade longer. One-fifth of the <20 degrees group had participated in regular manual labor; the group also had a much higher prevalence of smoking and/or alcohol use (nearly 30%) as compared to the other groups (nearly 10%). The group with the most severe restriction also had the highest prevalence of retinopathy (20%) and nephropathy (33%), but not neuropathy. Body composition did not vary across groups of hand stiffness; subjects with MCP restriction had a small but statistically significant difference in systolic blood pressure but not diastolic. They also had the highest prevalence of lipohypertrophy and keloid scar formation. Autofluorescence on AGE reader did not differ in the four groups.

Hand joint restriction was rarely painful: Only 7% complained of hand pain even in the most restricted group; one fourth had a history of trigger finger. Similarly, though more often seen in the restricted groups, only a small fraction had symptoms (14%) of neuropathy. Both observations translated into a small but statistically significant higher DHI in the restricted groups (mean 1.9 compared to 0.2). Only six percent in the <20 degrees group had a DHI of more than 10, that we considered clinically relevant hand function restriction. Most patients (80%) were observed to have a prayer sign in the <20 degrees group, and they consequently also had the maximum distance between the two MCP joints (mean 3 cm) and PIP joints (mean 5 cm). The examining physician in an independent opinion felt all patients in the <20 degrees group and three-fourth of the 20-40 degrees group had at least one described cheiroarthropathy phenotype. In univariate linear regression analyses adjusted for gender and duration of diabetes, height, systolic blood pressure, urine albumin creatinine ratio/clinically determined nephropathy, and the presence of frozen shoulder was significantly associated with average non-dominant hand MCP extension angle.[Table 3]

MRI findings

MRI scans of the hand were performed in 61 patients (12 in <20 degrees mean MCP extension group, 14 in the 20-40 degrees group, 20 in the 40-60 degrees group, and 15 in the >60 degrees group). The radiologist’s findings included flexor tendon thickening or edema in four patients, while one patient has tenosynovial inflammation in the abductor pollicis longus tendon suggestive of De-Quervain tenosynovitis. One patient had an enlarged median nerve suggestive of median nerve neuritis, while one other was found to have a bifid median nerve. Eight patients were found to have ganglion cysts, most commonly in the dorsal scapholunate ligament (n=5), and one had a subcortical cyst. One patient had early degenerative changes in metacarpal joints. All other MRI scans were reported as normal.

When analyzed by group, there was no significant difference within the groups in both mean tendon sheath thickness and maximum tendon sheath thickness.[Table 2] The group with the most stiffness had the highest skin thickness and reduced progressively; but subcutaneous thickness did not differ across groups. Similarly, median nerve area was not different across groups. Both the total hand fibrosis score and the total hand fibrosis Z-score were able to differentiate between the four groups.[Table 2]

Mean palmar skin thickness correlated significantly with palmar fascia thickness (correlation coefficient 0.298, p=0.02) but not with mean tenosynovial thickness (p=0.25). Similarly, mean tenosynovial thickness also correlated with palmar fascia thickness (correlation coefficient 0.28, p=0.03). Univariate linear regression showed that only palmar skin thickness and palmar fascia thickness correlated with MCP angle restriction.[Table 4] Only mean palmar skin thickness remained significantly associated with MCP angle in multiple linear regression.[Table 4]