Sixty people with type 2 DM and PN were recruited for this longitudinal, randomized, controlled, clinical trial (Clinicaltrials.gov, NCT02616263) (Fig. 1). Participants were recruited from 3/1/2016 to 5/10/2018 via the Recruitment Enhancement Core of the Institute of Clinical and Translational Sciences at the Washington University School of Medicine in St. Louis and patient databases of the Hastings and Mueller lab groups. Recruitment and retention information is in Fig. 1. Data collections were performed at the Washington University School of Medicine in St. Louis Program in Physical Therapy and Center for Clinical Imaging and Research. A priori power analysis was conducted to power regression analysis for predictors of 2nd MTPJ extension progression and was based on Hsieh et al’s19 sample size calculation for logistic and linear regression. Sixty-two participants were needed to detect a significant correlation of at least 0.44 between predictor and dependent variables after adjusting for covariates; however, study enrollment ended after recruiting 60 participants to allow completion of the 3-year follow-up.

Fig. 1.

Consort diagram.

Type 2 DM was diagnosed by the participants’ physicians. PN was defined as: (1) inability to sense the 5.07 monofilament on ≥ 1 of six plantar foot locations, (2) plantar toe vibration perception threshold >25 V (Biothesiometer, Biomedical Instrument Co, Newbury, OH, USA), or (3) Michigan Neuropathy Screening Instrument physical assessment score ≥ 2.

Exclusion criteria were inability to complete the study testing requirement, age > 75 years, pregnancy, dialysis, severe arterial disease (ankle-brachial index > 1.3 or < 0.9), rigid metatarsophalangeal deformity, presence of ulceration or lower extremity amputation greater than a single toe, weight greater than 180 kg, metal implants and/or pacemaker, or PN from causes other than DM (e.g., chemo toxic, alcoholic, lumbar radiculopathy). All participants provided informed consent. The protocol was approved by the Washington University School of Medicine in St. Louis Institutional Review Board.

Participants were randomly assigned to a foot- (experimental group) or shoulder-specific (control group) intervention. Groups were stratified by MTPJ angle on baseline computed tomography (CT) scans such that an even number of individuals with an MTPJ angle >50° were in each intervention group. The randomization code was prepared by the biostatistician shared via password-protected Excel sheet, and group assignment was provided by a clinical coordinator not involved in participant testing. Outcome assessors were blinded to group assignment and were not provided the password to the randomization code.

The shoulder intervention provided similar therapeutic interactions between participant and physical therapist while intervening on a joint remote to the foot. The participants and physical therapists could not be blinded to the intervention assignment. The intervention was delivered through 8 in-person visits completed over 6 months with a physical therapist and unmonitored home exercise program to be completed 5 times per week. Participants completed three visits during the first month, two visits in the second month, and a single visit per month for months three through six. The home program progression and intensity were individualized to participant level and tolerance.

The foot intervention (Supplemental Material 1) consisted of foot care (moisturizing cream and soft tissue massage to the foot by the participant), ankle dorsiflexion and toe flexion stretch in standing, progressive strengthening of the intrinsic and extrinsic foot muscles, and practice of functional activities while preventing toe extension during the activity (sit-to/from-stand, walking, and active dorsiflexion/plantarflexion). The shoulder intervention (Supplemental Material 2) included shoulder flexion stretch progressing to weighted shoulder flexion, shoulder circles through a large arc of motion, shoulder flexion return in abduction range of motion progressing to weighted exercise, triceps strengthening, and the functional activity of reaching items on a shelf. Exercise program progression was individualized based on visual assessment of the participant’s exercise performance by a physical therapist trained in the study’s intervention protocol. Once able to successfully complete the full number of sets/repetitions while maintaining proper technique, participants were progressed to the next level of the protocol as long as they were able to perform the higher intensity exercise with proper technique.

Home exercise program compliance was assessed by daily self-report in an exercise compliance log and reported as percent of prescribed exercises completed per week. The number of in-person visits for intervention was recorded.

Primary and secondary outcome measures were collected at baseline, 6 months, 1.5 years, and 3 years. Lower extremity measures were completed on the participant’s foot at baseline with (1) consistent toe extension pattern with dorsiflexion, (2) fewer foot complications (i.e., history of surgery or traumatic injury), and (3) the greatest MTPJ angle from visual inspection.

Blood markers included (1) Hemoglobin A1c, an indicator of 3-month diabetes control, values >5.7 % indicate abnormal blood glucose control20 and (2) high-sensitivity C-reactive protein (Hs-CRP) a clinical marker of chronic inflammation, >1 is associated with abnormal inflammation.21 Blood markers were assessed at the baseline, 1.5, and 3 year visits, as we did not believe there would be sufficient time for these measures to markedly change by the 6-month timepoint.

Sensory PN was assessed with biothesiometry and the Michigan Neuropathy Screening Instrument Score. A biothesiometer was placed on the plantar aspect of the great toe with a vibration perception threshold >25 V indicating impaired sensation.22 The Michigan Neuropathy Screening Instrument23 physical assessment, incorporating sensory and observational assessments, was administered with higher scores indicating worse neuropathy (maximum score: 10, ≥2 was considered neuropathy).

Alignment measures were completed by a single rater (PKC) blinded to group assignment. The rater completed a computer-based training session and demonstrated reliability compared to a foot and ankle fellowship trained orthopedic surgeon using sample radiographs and previously determined precision criteria18 prior to performing measurements.

The primary outcome variable was 2nd MTPJ extension angle measured from CT scan (Siemens Biograph 40 CT, Siemens Medical Systems, Inc., Iselin, NJ, USA). Scan parameters were as follows: table speed=24mm/sec/rotation, pitch=1, 24 mm collimation, 0.6 mm slice thickness, 220 mAs, and 120 kVp. A phantom (Mindways, Austin, TX, USA; Calibration value: L11G 11F1 1171 611 L) was in each scan to calibrate the Hounsfield units (HU). Ankle position was standardized with a wooden board holding the ankle in 30° plantarflexion. The MTPJ angle was defined as 180 degrees minus the angle measured between the bisection of the proximal phalanx and the extension of the metatarsal bisector. A smaller angle indicates less forefoot deformity (Supplementary Material 3).24

Intrinsic foot compartment muscle and fat volume, BMD, foot kinematics during sit-to/from-stand and walking tasks, and AGEs in the skin were assessed. See supplemental materials for detailed methods (Supplemental Material 4). Briefly, intrinsic foot compartment muscle and fat volume25–28 and tarsal/metatarsal BMD29–31 were quantified with CT scan. Second metatarsal BMD and average BMD across all tarsal/metatarsal bones were used in analysis. Foot kinematics during sit-to/from-stand and walking were assessed using motion capture (Vicon MX, Los Angeles, CA, USA; Visual 3D, C-Motion Inc., Germantown, MD, USA) with a 4-segment foot model.32,33 Three trials of each task were averaged for analysis. Extension excursion was the difference between the peak MTPJ extension angle during the sit-to-stand task and the average MTPJ angle during the 3-second-standing component of the sit-to-stand task. For walking, peak MTPJ extension angle was the variable of interest. Forearm skin AGE accumulation was assessed using skin intrinsic fluorescence measured with a SCOUT DS skin fluorescence spectrometer (VeraLight, Albuquerque, NM, USA).34

All analyses were done as intention-to-treat based on group assignment. Because of the presence of missing data, longitudinal analysis to assess the effect of intervention on 2nd MTPJ extension was tested using a linear mixed-effects model repeated measures analysis of variance (RM-ANOVA). Unlike traditional repeated measures ANOVA, the mixed model permits missing data. The RM-ANOVA included a Satterthwaite adjustment for unequal variance. The focus of the RM-ANOVA was on the significance of the interaction between group and visit. Interactions test the null hypothesis that change over visits is not different across groups. For the primary outcome (collected at the four time points), a compound symmetry covariance structure was used and additional statistical contrasts within the RM-ANOVA tested pre-specified hypotheses regarding the equivalence of change between baseline and each post-baseline visit across groups. Previous work has identified cutpoints for change by calculating least significant change values (changes considered real and not the result of chance or error) for radiological measures of the foot. Our original data analysis plan included an a priori cutpoint of progression in 2nd MTPJ extension angle of at least +4°, a rounded up value from our published cutpoint value of 3.94°18 In our analysis, we returned to the published cutpoint of 3.94° and individuals with an increase in 2nd MTPJ extension angle of 3.94° over 1.5 years and 7.88° over 3 years were classified as having 2nd MTPJ extension progression.35 Cutpoints were calculated from published standard error of the mean (SEM) converted to a least significant change (LSC) using the formula LSC=1.96*2*SEM where SEM=1.42° for a 1.5 year interval. The proportion of participants with progression at 1.5 and 3 years was compared across groups by chi-square test.

Group comparison analysis found no detectible effect of the intervention on MTPJ deformity. Therefore, the two intervention groups were combined for remaining analyses to improve our ability to characterize progression of MTPJ deformity in a larger group of participants and detect smaller magnitude changes over time. Change over time in primary and ancillary MTPJ deformity measures was performed using the combined group of all participants with RM-ANOVA where the focus of the analysis was on the main effect of visit. Analysis of the primary outcome included contrasts to assess change between baseline and each post-baseline visit. Cutpoints for deformity progression were as described above.

Relationships between 2nd MTPJ extension angle and predictor variables (AGEs, foot intrinsic muscle quality measures, movement pattern during sit-to/from-stand and walking, BMD) were examined with bivariate Spearman correlations. These analyses excluded participants with missing data for the component variables. Predictors with p-values <0.10 in bivariate correlations were included in stepwise multivariable logistic regression to predict 2nd MTPJ deformity progression (progressed vs. did not progress over the duration of the study). The stepwise multivariable model used a p-value=0.10 for entry of a predictor and for remaining in the model. Note that this analysis was exploratory in nature, and p-values should be interpreted with caution due to the number of analyses conducted. Additional details regarding data analysis are in the supplemental materials.

Results indicated no effect of a foot-compared to shoulder-specific intervention on 2nd MTPJ extension angle (Fig. 2). The RM-ANOVA showed no statistically significant main effect of group (p = 0.56), visit (p = 0.29), or group by visit interaction (p = 0.79). Specific statistical comparisons of the changes from baseline to each follow-up between groups were not significant (baseline to 6-month, p = 0.45; baseline to 1.5-year, p = 0.84; baseline to 3-year, p = 0.70).

Fig. 2.

Change in 2nd metatarsophalangeal joint extension deformity in degrees from baseline in combined group (all participants), foot intervention (experimental) group, and shoulder intervention (control) group. Note that there were no significant between group differences in change in 2nd metatarsophalangeal joint angle from baseline to month 6 (p = 0.45), baseline to 1.5-year (p = 0.84) or baseline to 3-year follow-up (p = 0.70).

Out of concern that missing data at the 3-year follow-up could influence results, a sensitivity analysis was performed with data from the 45/56 participants with non-missing data at the 3-year follow-up with time between baseline and 3-year follow-up included as a covariate to adjust for the variable length of follow-up across participants. In this model, the covariate (p = 0.19), interaction of group and visit (p = 0.74), and baseline to final follow-up contrast (p = 0.80) were not statistically significant.

Participants in the foot intervention group completed a mean (SD) of 67.6 (34.4) % of the prescribed home exercise program in week 1, 66.1 (31.9) % in week 2, 69.1 (29.4) % in weeks 3–4, 58.3 (33.4) % in weeks 5–6, 59.7 (33.5) % in weeks 7–9, 58.7 (34.8) % in weeks 10–13, 50.2 (35.3) % in weeks 14–17, and 17.1 (25.2) % in weeks 18–24. Throughout the intervention, participants in the foot intervention group completed 7.7 (0.7) sessions of a possible 8 sessions.

Progression of 2nd MTPJ extension deformity. There were no significant between group differences in proportion of participants meeting the criteria for forefoot deformity progression at 1.5 or 3-year follow-up. In the foot intervention group, 7 of 25 (28 %) participants met the cutpoint of progression of 2nd MTPJ angle within the first 1.5 years of the study and 6/25 (24 %) in the shoulder group (p = 0.75). At the 3-year time point, 2/22 (9 %) of the foot intervention group and 2/23 (9 %) of the shoulder intervention group met the cutpoint for progression (p = 0.96).

In the combined group analysis, there was no significant change in 2nd MTPJ angle over time with a mean (SD) of 53.19 (13.12)° at baseline, 54.79 (14.30)° at 6-month, 52.76 (13.85)° at 1.5-years, and 54.15 (13.75)° at 3-year follow-up. Change in 2nd MTPJ from baseline was a mean (SD) of 1.60 (8.64)° at 6-month, −0.12 (7.98)° at 1.5-year, and −0.38 (7.18)° at 3-year follow-up. Main effect of visit was not statistically significant, p = 0.29; specific contrasts to assess change between baseline to 6-month (p = 0.16), baseline to 1.5-year (p = 0.90), and baseline to 3-year follow-up (p = 0.69) were not statistically significant (Fig. 2).

Predictors of 2nd MTPJ extension deformity. BMD change from baseline was the only predictor of 2nd MTPJ extension deformity progression, accounting for 22 % of the variability in 2nd MTPJ extension angle progression (R2 = 0.22, p = 0.01); with 78 % of variability in 2nd MTPJ extension angle progression remaining unexplained. For each unit increase in BMD from baseline, the odds of 2nd MTPJ extension deformity progression was reduced by 7 % (odds ratio, 0.93; 95 % confidence bounds, 0.87 to 0.98). After BMD change was included in the model, no other potential predictor satisfied the inclusion significance level. Results of bivariate analysis for relationships between predictor variables and 2nd MTPJ extension angle progression are in Table 3.