Flexor Tendon Tenotomy Treatment of the Diabetic Foot: A Multicenter Randomized Controlled Trial

Diabetes represents one of the major medical challenges facing the world, with an estimated 537 million people affected worldwide in 2021 and a further 240 million projected to be affected by 2045 (1). Diabetes causes 1.5 million deaths annually (2), and most of these can be attributed to the complications associated with diabetes (2). One of these complications is the diabetic foot ulcer (DFU). With a lifetime risk of 19–34%, DFUs are a frequent complication in diabetes. Incurring a DFU results in a 2.4-times decrease in 5-year survival compared with no DFU. This effect on mortality is even more pronounced if the DFU leads to infection or amputation (3–6). Several factors influence the risk of developing DFUs, with foot deformities being one of the most prevalent (7–9).

A higher prevalence of foot deformities has been reported for individuals with diabetes compared with the general population, although the pathology is unclear (10). One of the most common foot deformities found in radiological and clinical studies is the hammertoe deformity (here on referred to as hammertoe) (11–14), which has been shown to be an independent risk factor for the development of DFUs (8,15).

Several definitions of hammertoe have been suggested without any consensus (16). Similarly, the pathology that leads to hammertoes is unclear for individuals with and without diabetes (17–21). Despite these shortcomings, hammertoe surgery, ranging from proximal interphalangeal joint (PIPJ) fusion to larger bony and soft tissue procedures, is performed every day worldwide (22).

Surgery of the diabetic foot has traditionally been associated with a high risk of infection and amputation (23,24). However, recent studies have rekindled the debate (25,26), and recommendations on diabetic foot surgery have made their way into the latest international guidelines on the treatment of the diabetic foot. The recommendations are still weak, with ambiguous support in the literature (27,28).

One of the recommendations supports using flexor tendon tenotomy treatment (here on referred to as tenotomy) on the basis of retrospective studies (29), including two studies from our institute (30,31). Originally described by Lountzis et al. (32) in 2007, tenotomies have gained increased use in the daily treatment of the diabetic foot. However, if surgery of the diabetic foot is to be scientifically supported, we need randomized controlled trials (RCTs). The aim of this study was to examine the ability of tenotomies to prevent and treat ulcers associated with the diabetic hammertoe.

The study was conducted as a multicenter RCT with two parallel arms of individuals with diabetes and hammertoes stratified by the presence of ulcer or impending ulcer to examine superiority of tenotomy and standard treatment compared with standard treatment alone. Furthermore, a substudy on the effects of tenotomies on standing balance was performed. The study was conducted in accordance with the Declaration of Helsinki (33). Approval from the local ethical committee (Copenhagen, Denmark) and the Danish Data Protection Agency (Valby, Denmark) and registration with ClinicalTrials.gov was obtained/done before initiation of study. The principles of Good Clinical Practice were followed throughout the study (34).

Participants were randomly assigned using the randomization module in an open source data capturing system (RedCAP; Vanderbilt University, Nashville, TN). Block randomization was performed, with each block consisting of four participants and stratified by treatment site. An allocation table was generated by a statistician with no connection to the project at Steno Diabetes Center Copenhagen (SDCC). All study personnel were blinded to the allocation table.

Between 1 December 2019 and 31 March 2021, participants were identified, screened, and treated at three sites: the outpatient clinics at the Copenhagen Wound Healing Center, Hillerød Hospital, and the foot clinic at SDCC. To facilitate recruitment, screening was extended to additional sites, including Herlev, Holbæk, Hvidovre, and Slagelse Hospitals, and collaborators in the primary sector (all authorized podiatrists and general practitioners in the Capital Region) were asked to refer candidates. All visits and treatments were performed at the three original sites, and the protocol was adjusted accordingly. All interventions were performed and visits attended by the same orthopedic surgeon (J.A.A.).

Individuals aged >18 years with diabetes and one or more hammertoes were identified. Eligible individuals had no history of above-ankle amputations, a toe blood pressure >29 mmHg, no vascular surgery planned or performed within the past 4 weeks, and no treatment with cytotoxic drugs or systemic corticoid treatment at present or during the past 8 weeks.

Included participants were randomly assigned at first visit to tenotomy and standard treatment or to standard treatment alone after stratification. Intervention was performed according to random assignment at the earliest opportunity after inclusion. Delay in performing the intervention had several reasons: Participants had to be given time to prepare for the intervention, therapeutic sandals had to be prepared for participants to use after intervention, and balance analysis had to be performed before the intervention.

In the impending ulcer group, participants were seen 2 and 3 weeks and 3, 6, and 12 months after inclusion. An end point was reached if a DFU developed at one or more predilection sites of the index toes. Participants in the control group who reached an end point were subsequently offered the intervention (tenotomy). Participants who reached an end point or secondarily underwent tenotomy were still followed at planned visits at 3, 6, and 12 months.

In the ulcer group, participants were seen 2 and 3 weeks after inclusion, then every second week until week 10 and at 3, 6, and 12 months after inclusion. An end point was reached if the index ulcers healed. If the participant reached an end point, he or she was seen at 3, 6, and 12 months after visit 1. If the index ulcer of a participant in the control group had not healed by week 10 or if the ulcer was healed but recurred during observation, participants in the control group were offered intervention (tenotomy) at the earliest time possible. Participants who reached an end point or secondarily underwent tenotomy were still followed as planned at 3, 6, and 12 months (Supplementary Fig. 1).

From 1 December 2019 to 31 March 2021, participants in the tenotomy study with a DFU were eligible for inclusion in the balance study if they did not have a need for walking aids and were deemed to be able to perform the balance examination. To facilitate recruitment, screening was extended to participants in the tenotomy study with impending ulcers, and the protocol was adjusted accordingly. Balance examination was performed before the week 2 visit (intervention visit) and 3 months later. The investigators who performed the balance examination and data collection were blinded to the random assignments of the participants.

Hammertoe is defined as hyperextension in the metatarsophalangeal joint, flexion in the PIPJ or distal interphalangeal joint, or a combination of the three deformities in any of the four lateral toes. In addition, hammertoe included hyperextension in the metatarsophalangeal joint, flexion in interphalangeal joint, or a combination of the two deformities in the hallux (17).

Predilection sites for ulcers associated with hammertoes in individuals with diabetes were defined in accordance with our previous publication (31) as an ulcer on the pulpa or dorsal to the PIPJ. Impending ulcers are calluses in predilection sites for ulcers associated with hammertoes or nail changes on hammertoes (31).

DFU is a lesion of the skin on the foot of an individual with diabetes (35). An index ulcer is the ulcer associated with hammertoes chosen for examination.

Complete ulcer healing is intact skin covering the whole ulcer (36). A recurrent ulcer is an ulcer that recurs in the same anatomical location after complete ulcer healing (31). A transfer ulcer is an ulcer on one of the defined predilection sites for ulcers associated with hammertoes on a toe of the same foot where one or more toes previously have been tenotomized (31).

Transfer pressure is defined as an impending ulcer in one or more of the defined predilection sites for ulcers associated with hammertoes on a toe of the same foot where one or more toes previously have been tenotomized (31). Peripheral neuropathy is defined by a biothesiometry measurement >24 V and/or missing sensation to monofilament examination.

The procedure was performed in an outpatient setting. Participants were placed in a seated position with their feet elevated. Feet were disinfected using a colored 0.5% chlorhexidine and 96% alcohol solution. A local anesthetic solution without adrenalin was administered subcutaneously. After administration of the local anesthetic, a second disinfection of the skin was performed.

The procedure was performed using a 1.2-mm diameter, 40-mm long needle (BD Microlance 3; Becton, Dickinson and Company, Franklin Lakes, NJ), which was introduced through the plantar skin of the foot just proximal to the web level of the toes planned for surgery. Flexor digitorum longus and brevis were then severed by swiping motions of the needle (Supplementary Video 1 and Supplementary Animation 1). Gauze bandages were applied to the toes and secured with an elastic bandage. Participants were subsequently discharged with instructions to rest at home for 24 h.

At all visits, standard treatment in the form of an examination of the feet, trimming of calluses, nail treatments, ulcer debridement, and bandaging were performed by educated staff.

Rocker bottom sandals (New Feet A/S, Randers, Denmark) with individual custom-made insoles were offered to all participants as part of the standard treatment. In addition, the footwear used by participants was checked for fit at the screening visit. If deemed necessary, participants were referred to a shoemaker for custom-made footwear.

At each visit, offloading devices were checked and corrections performed by trained staff as deemed necessary by the investigator. If corrections were needed beyond the capabilities of the treatment facility, participants were referred to relevant professionals. Participants were instructed to not use the specific offloading device before applied corrections had been controlled by the investigator.

The standing balance analysis was performed using a force plate (model OR6-7; Advanced Mechanical Technology, Inc, Watertown, MA). Participants were asked to stand on a biomechanical force plate with feet parallel and 6 cm apart (measured by a standard block), instructed to look straight ahead at a target, and remain as still as possible for 30 s with eyes open. The procedure was repeated three times, with 60 s of light walking between trials. Data collection was performed using Nexus 2.8.1 (Vicon Motion Systems Ltd., Oxford, U.K.) and processed using MATLAB (Natick, MA) software. Measurements were collected at 1,000 Hz, but downsampled to 100 Hz.

Standing balance was assessed by measurements of the center of pressure (COP) and quantified as length of sway path and COP area. Sway path measures the length of the path traversed by COP during the sway examination (in mm). This translates to how much the participant swayed from their starting point throughout the examination; hence, a shorter sway path was interpreted as a better standing balance. The COP area is the area (in mm²) containing 95% of the measured COP points; hence, a smaller COP area was interpreted as a better standing balance.

Vibration sensation of the feet was examined using a biothesiometer (Bio-Medical Instrument Co., Newbury, OH) and measured in volts. Monofilament examination was performed using a 10g Semmes Weinstein monofilament. Sensation was absent if the participant did not register the monofilament on two or more of the three tested points on the foot.

Categorical and ordinal variables are presented as percentages of the analyzed group. Continuous variables are presented as mean ± SD. If continuous variables were not normally distributed, they are presented as median (25–75% quartile) and log-transformed before further analysis.

Categorical and ordinal variables were compared across groups using χ² test (Pearson test). The ulcer group was analyzed using Fisher exact test because of the low number of expected frequencies. For analysis of differences in continuous variables between groups, t test was performed. In the impending ulcer group, a Kaplan-Meier analysis of ulcer-free days was performed, using the log-rank test to determine statistical significance. In the ulcer group, a Kaplan-Meier analysis of days to ulcer healing was performed, using the log-rank test to determine statistical significance. In the impending ulcer group, linear regression was performed to assess the effect of relevant clinical variables on the primary end points in the three studies.

In the analysis comparing COP area and sway path between intervention and control groups (t test), variables were not log-transformed because of the Δ for COP area and sway path having both negative and positive values.

All analyses were performed according to the intention-to-treat principle. The intention-to-treat population analyzed included all participants who completed the baseline/intervention visit.

All statistical analyses were performed in the SAS Studio 3.7 software (SAS Institute, Cary, NC). Statistical significance was defined as P < 0.05.

A total of 61 individuals with ulcers were screened at the three study sites, of whom 21 were randomly assigned. Of the 21 randomly assigned participants, 4 were excluded because of their ulcers healing before the first visit (intervention), and 1 was lost to follow-up after attending the screening visit, leaving 16 who completed the study and were included in the analysis (Supplementary Fig. 2).

The mean age of participants with ulcers was 67.0 ± 11.3 years, 11 (68.8%) were male, 5 (31.3%) had type 1 diabetes, and the median diabetes duration was 17.5 years (25–75% quartile 9–24 years). All participants had peripheral neuropathy, and 13 (81.3%) had palpable foot pulses. No differences in baseline characteristics were found when comparing the intervention and control group (Table 1). One participant in the intervention group had two toes with ulcers associated with hammertoes. Both ulcers achieved healing at the same visit and are treated as one index ulcer in the analysis.

During follow-up, all eight ulcers in the intervention group healed, and three (37.5%) healed in the control group (P = 0.026) (Table 2). On comparison of time to healing between the intervention and control groups using a Kaplan-Meier analysis, a significant difference was found in favor of tenotomy (P = 0.04) (Fig. 1). Of the healed ulcers, none of the eight ulcers in the intervention group recurred during follow-up, while two (66.7%) recurred in the control group (P = 0.055) (Table 2).

In the control group, seven participants were offered secondary intervention (five who did not attain healing and two with recurrence) according to the protocol. All seven participants accepted secondary intervention, and their ulcers consequently healed, with no recurrence during observation.

During follow-up, four (50%) participants incurred complications associated with the intervention; two experienced pain at the plantar aspect of the foot, and two developed a hematoma of the operated foot. All the registered complications were alleviated at the visit 2 weeks after the intervention, and none of the complications had long-term implications for the participants. No participants incurred amputations or infections associated with the intervention. An infection in the index toe was registered in one participant in the control group (Table 2).

A total of 163 individuals with impending ulcers were screened of whom 80 were randomly assigned. One withdrew consent after the screening visit, and one died during follow-up, leaving 78 who completed the study (Supplementary Fig. 2). In addition to the 78 participants who completed the study, the participant who died during follow-up was included in the analysis performed (data were censored at time of death).

The mean age of the participants with impending ulcers was 67.7 ± 9.5 years, 45 (57.0%) were male, and 23 (29.1%) had type 1 diabetes. The median diabetes duration was 21.0 years (25–75% quartile 14.0–27.0 years), 71 (89.9.%) had peripheral neuropathy, and 68 (86.1.%) had palpable foot pulses. No differences were found between the intervention and control groups in baseline characteristics (Table 1).

During the follow-up, one participant (2.6%) incurred progression from impending to active ulcer in the intervention group, while seven participants (17.5%) incurred progression from impending to active ulcer in the control group (P = 0.028) (Table 2). The logistic regression model, with adjustment for age, sex, and diabetes type, did not show any association between tested clinical variables and incidence of progression. Comparing DFU-free days between the intervention and control groups by Kaplan-Meier analysis, a significant difference was found favoring the intervention group (P = 0.043) (Fig. 1).

All seven participants in the intervention group who incurred an ulcer were offered secondary intervention according to the protocol. All seven accepted secondary intervention, and their ulcers consequently healed, with no recurrence during observation.

No infection in the index toes was observed in the intervention group, while one infection was observed in the control group (2.5%) (Table 2). During follow-up, 26 participants (66.7%) incurred a complication associated with tenotomies. Pain at the plantar aspect of the foot (n = 21) and hematoma (n = 17) were the most common complications. In all but one participant, these complications were alleviated 2 weeks after tenotomy; in one participant, pain persisted for the duration of the study. There were no amputations or infections correlated with the procedures reported. Five participants incurred transfer complications (two transfer ulcers and three transfer pressure). One participant reported a change in balance when walking in the form of changed gait, which was described as advantageous when traversing inclinations and disadvantageous when walking on level ground. In the intervention group, one participant died during follow-up (2.5 months after intervention) as a result of complications related to heart disease diagnosed before inclusion and without any connection to the intervention shown or suspected.

Of the participants in the tenotomy study, 48 were included in the balance study, and 45 completed the study (1 died before the follow-up visit, and 2 did not have a follow-up visit because of COVID-19 restrictions on nonvital hospital visits). The median difference in sway path between the first and second sway analysis was −2.1 mm (25–75% quartile −88.2 to 64.9 mm) in the intervention group and 15.7 mm (−67.3 to 42.5 mm) in the control group (P = 0.29), correlating to a median 0.23% decrease for the intervention group and a median 3.5% increase in the control group (P = 0.76). The median difference in sway area between first and second sway analysis was −12.3 mm² (−85.9 to 93.0 mm²) in the intervention group and 3.2 mm² (−64.3 to 165.7 mm²⁾ in the control group (P = 0.92), correlating with a median 7.3% decrease in the intervention group and a 0.8% increase in the control group (P = 0.30) (Table 3).

We found that tenotomy performed by needle significantly increases the proportion of ulcers healed and reduces the time to ulcer healing in individuals with DFUs associated with hammertoes. The tenotomy procedure significantly reduced the risk of progression from impending to active DFU and increased the number of ulcer-free days for individuals with impending DFUs associated with hammertoe. In addition, the complications associated with tenotomies were few, mild, and transient.

Surgery undertaken to treat the DFU always comes with the risk of leaving the treated individual in a worse situation than they presented. This has led to a wait-and-see approach to the diabetic foot, often resulting in surgery being performed only when an ulcer has already occurred and even then, with a cautious approach. In the latest published international guidelines, a total of five surgical procedures have received recommendations (27,28). The results from this study suggest that tenotomy treatment should be considered whenever a hammertoe shows beginning signs of an impending DFU and is effective in lowering healing times if a DFU has already formed.

It has been questioned whether tenotomies can have a positive effect on the diabetic hammertoe if the deformity is rigid, and arthroplasty or arthrodesis have classically been recommended for the rigid hammertoe (26). There are, however, no studies that have examined the effect of tenotomy treatment of rigid diabetic hammertoes. This study did not exclude participants on the basis of hammertoe rigidity. Our theory is that tenotomizing both toe flexors, while not necessarily correcting the deformity, will release some of the force that pulls the toe into the underlying surface, thereby reducing the pressure at the plantar aspect of the toe. This study was not powered or designed to evaluate the difference between tenotomy treatment of rigid and flexible diabetic hammertoes. On the basis of our clinical experience, we suggest that tenotomy can be undertaken for individuals with diabetes and rigid hammertoes, and more elaborate surgical interventions (i.e., arthrodesis, arthroplasty) can be performed if tenotomy is not sufficient to offload the treated toes.

One of the main complications after a healed DFU is recurrence (37). In this study, ulcer recurrence rates tended to favor tenotomies, although there was no statistically significant difference. However, the lack of significance could be attributed to the lack of power in the study.

A possible solution to solve the challenge of ulcer recurrence is prevention of the first DFU. This study indicated that tenotomy treatment of hammertoes with signs of DFUs could prevent the development of DFUs. In addition to eliminating the risk of recurrence, prevention of a DFU eliminates the risk of infections and amputations. To our knowledge, this is the first prospective study to describe that tenotomies effectively decrease the risk of incurring a DFU in the case of a hammertoe.

The classic mantra of do no further harm is as true when treating the diabetic foot as for any other field. This study found no serious complications (i.e., infections or amputations associated with tenotomy). In the intervention group, one participant died as a result of complications of heart disease, which was diagnosed before and had no suspected association with the study intervention. Pain and hematoma were common complications of tenotomies; however, both were found to be mild and self-limiting, with no negative long-term effects. On the other hand, transfer ulcer and pressure are potentially serious because they can lead to ulcers, infections, and amputations. Transfer lesions occur when pressure is relieved from one toe and form on another point of the foot. The problem of transfer lesions is, however, easily remedied by tenotomizing the toes that show signs of transfer pressure before an ulcer forms or, alternatively, tenotomizing all toes of the treated foot, which we have shown is a safe alternative in a previous study (31).

A concern could be an altered balance after losing the function of the toe flexors. No altered balance was found in this study. It is well established that individuals with diabetes and neuropathy have an impaired balance (38) and individuals who have tenotomies performed will often experience neuropathy, as was also the case in this study. This may help to explain why tenotomy treatment was not found to influence standing balance in this study. The proprioceptive information from the foot to the brain could likely already be cut, delayed, or at least altered before tenotomy is performed. This could result in late responses from the toe flexors that are already compromised, rendering their contribution to maintaining balance negligible; thereby, tenotomizing them would have no effect. Only one participant complained of negative effects on balance consisting of an altered gait, which was not clinically discernible or had any consequences in the form of falls or need for walking aids in the short or long term. In addition, sway analysis showed no decrease in balance 3 months after tenotomy in the intervention group, and no difference in balance was observed when comparing the changes before intervention/start of study and 3 months later. This study is the first on the impact of tenotomy on balance; therefore, the power of the conclusion is hard to evaluate. However, neither the sway analysis nor the clinical findings gives cause for concern that balance may be negatively affected after tenotomy.

There are limitations to this study. The recruitment for the study was challenging partly as a consequence of the COVID-19 pandemic (39), where recruitment was diminished because of restrictions on hospital visits, resulting in fewer potential participants being screened. Despite this, the findings supporting tenotomy treatment were significant. The tenotomy study was unblinded. Blinding in studies of surgical intervention is challenging in general, and this study was no exception. Blinding of the participant is not feasible, unless a sham operation is performed on the control group. However, individuals would likely be aware of the retained ability to flex their toes after surgery. The surgeon cannot be blinded, but the investigators who perform the follow-up could be. The required setup would entail that the participant’s toes were placed back in a hammertoe position, and they did not try to flex them during examination or treatment. It would likely require an external assessor providing examination of the feet, which was not available in this study, but future studies should aspire to implement this.

The strengths of the study are the prospective RCT design and the well-defined cohort with close follow-up. All surgeries were performed by the same surgeon, and the surgeon was present for all visits, which could be both a strength and a potential risk of bias.

In conclusion, this RCT shows that needle tenotomy is a safe, effective, and simple tool to treat and prevent ulcers of the diabetic hammertoe without effects on the balance of individuals who undergo the procedure. Tenotomies can be performed anywhere in the world where there is access to a needle and skin disinfectant. The surgical technique is simple, and the individual who receives the procedure is, in most cases, able to return to normal life at best the following day and at worst a couple weeks after the procedure, with no apparent short- or long-term complications. In our opinion, flexor tendon tenotomies should be considered in all cases of diabetic hammertoes with an impending or active ulcer.

Acknowledgments. The authors thank several colleagues who contributed to the completion of the study, including the podiatrists and nurses who assisted in the identification and recruitment of participants in addition to their treatment, the administrative staff at the inclusion sites for assistance in planning and booking participant visits, and the scientific staff who performed the balance examination at the Human Movement Analysis Laboratory, Hvidovre Hospital. The authors also thank statistician Lars Jorge Diaz (Clinical Epidemiology, SDCC) for assistance with producing an allocation table and in the statistical analysis and project manager Maria Athena Campbell (SDCC) for assistance in proofreading the manuscript. The authors especially thank the study participants, without whom there would have been no study.

Funding. This study was supported by Jascha Fonden and Aase and Ejnar Danielsens Foundation. In addition, the research was funded from internal funds from SDCC.

Duality of Interest. P.R. has received consultancy and/or speaking fees (to SDCC) from AbbVie, Astellas Pharma, AstraZeneca, Bayer, Boehringer Ingelheim, Gilead Sciences, Eli Lilly, MSD, Novo Nordisk, and Sanofi and research grants from Novo Nordisk and AstraZeneca outside the submitted work. M.F.-M. has received speaking fees from Boehringer Ingelheim, Novartis, Baxter, and Sanofi outside the submitted work. K.K.-M. has received consultancy and/or speaking fees from Coloplast, Mölnlycke, SoftOx A/S, and Bayer Pharmaceuticals AG outside the submitted work. S.E. is employed by Novo Nordisk A/S. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.A.A. designed the study; recruited and treated the participants; collected, analyzed, and interpreted the data; wrote and edited the manuscript; and approved the final version of the manuscript. A.R., S.E., M.F.-M., K.K.-M., and P.R. contributed to the conception and design of the study, reviewed the manuscript, and approved the final version of the manuscript. J.B. collected data, reviewed the manuscript, and approved the final version of the manuscript. J.A.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented during the Diabetic Foot Study Group Meeting, Stockholm, Sweden, 16–18 September 2022, and in abstract form at the European Association for the Study of Diabetes 58th Annual Meeting, Stockholm, Sweden, 19–23 September 2022.