Two authors retrieved the electronic databases, including PubMed, Embase and the Cochrane Central Register of Controlled Trials (CENTRAL) from the inception dates to Nov. 2022, independently. The keywords used for study searching include “diabetic foot ulcer”, “diabetic foot”, “transverse tibial bone transport”, and so on. We combined the subject terms and free terms together to ensure full coverage of the potential eligible studies. The list of the searching strategies of three databases is available in Appendix 2. The related studies in the references list of each included study were also hand-searched and included for analysis.

After exporting the literature records from the databases, two authors screened all of them one by one to identify eligible studies, according to the inclusion criteria as follows: (1) patients were diagnosed with DFU (type-I or -II diabetes)DFU; (2) patients were operated with TTT on the DFU affected leg; (3) studies observed the treatment outcomes of the TTT surgery, such as healing time, healing rate, ABI, skin temperature, VAS pain scale, complications, and so on; (4) clinical studies designed as randomized controlled trials (RCTs), cohort studies, case-control studies, or case series. The publication language was restricted in English.

Studies would be excluded when meeting the following criteria: (1) duplicated studies; (2) patients operated for foot ulcer derived from non-diabetic diseases (such as occlusive vascular disease); (3) studies designed as case report, literature review, systematic review/meta analysis and letter to editors.

The initially retrieved records were imported into EndNote version 20.2.1 (Clarivate Analytics, Philadelphia, USA), and the duplicated studies were merged together. After then, we screened the title and abstract of each record to assess the eligibility and excluded the obviously non-related studies. The full-text of the remained studies were finally reviewed to identify the final eligible studies. The whole process of study screening was conducted by two authors independently, according to the inclusion and exclusion criteria.

According to the PICOS principle, we perused all of the included studies, and extracted the items as follows: (1) Participants (P): patients number, drop-out patients, age, sex, body mass index (BMI), type of diabetes, lengths of diabetes history and DFU history, ulceration grade, ulceration area, glycosylated hemoglobin (HbA1C), and ankle-brachial index (ABI); (2) Interventions (I): detailed treatment protocol, perioperative management, anesthesia method, site of TTT, bone window size, fixations of external fixator and bone block, and detailed bone transportation protocol; (3) Comparisons (C): when a control group was set, the information about the treatment process was extracted; (4) Outcomes (O): healing rate, healing time, limb salvage rate, ABI, skin temperature, visual analogue scale (VAS) pain scale, and complications such as fracture at transportation site, pin-site infection, DFU recurrence; (5) Study (S): lead author’s name, publication year, country, study period, study design, and follow-up period. The process of data extraction was conducted according to the checklist of data collection which was proposed by the Cochrane Collaboration. The data were extracted by two individual reviewers independently, and cross-checked.

The quality of the RCTs, case-control studies, case series was assessed using the Cochrane Collaboration tool for assessing risk of bias, Newcastle-Ottawa scale (NOS), and JBI Meta-Analysis of Statistics, Assessment, and Review Instrument (JBI-MAStARI) scale. This process was performed by two authors independently, and the disagreement was solved by the third author.

(1) For outcomes such as healing rate, limb salvage rate, fracture incidence at transport site, and pin-site infection risk in the TTT group, single proportional meta-analyses were performed to calculate the pooled proportions, with the “PRAW” model; (2) for outcomes such as healing time of TTT group, meta-analyses of single-group continuous data were performed to calculate the pooled mean values, with the “MRAW” model; (3) for comparisons between pre-operative and post-operative ABI, skin temperature, and VAS, meta-analyses of continuous data were conducted, with effect size of mean difference (MD); (4) for comparisons of healing rate and limb salvage rate between TTT and control groups, meta-analyses of binary data were performed with effect size of odds ratio (OR); (5) for comparison of the DFU recurrence risk between TTT and control group, meta-analyses of binary data were performed with effect size of risk ratio (RR).

The heterogeneity was tested with I², and random- or fixed-effect model would be employed, when presenting with or without significant heterogeneity (I²>50%). Z test was used to test the statistical significance of the pooled results. Funnel plot and Egger’s/Begg’s tests (p<0.1 and p<0.05 indicate significant publication bias for Egger’s and Begg’s tests, respectively) were used to detect the risk of publication bias when five or more studies were included in a meta-analysis. If significant publication bias was detected, non-parameter trim-and-fill method was used to adjust the bias. Sensitivity analyses were performed when significant heterogeneity was evident in meta-analyses with five or more studies. The statistical significance was defined as a two-side P value of less than 0.05. The statistical procedures were completed using R 4.1.3 for Windows (R Foundation for Statistical Computing, Vienna, Austria).

The flowchart of the study searching and selecting is presented in the Figure 1 . In total, 99 articles were retrieved through databases and manual searching. The titles/abstracts of 72 articles were reviewed after removing 27 duplicates. A total of 35 studies not related to this topic were excluded after reviewing the titles and abstracts. Then, the full-text of 37 studies were screening for final eligibility. A total of 7 studies (3, 15–20) were finally included for analysis.

Summary of the characteristics of the eligible studies is shown in Table 1 . All of the studies were published after 2019, in China. The study designs include single-arm RCT (n=1) (3), case-control study (n=4) (16, 18–20), and case-series study (n=2) (15, 17). A total of 818 participants were included, with 661 participants treated with TTT operation. The mean age was ranged between 40.0 ± 11.0 to 70.4 ± 6.0 years, and the male percentage was ranged between 52.6% to 83.3%. The type of diabetes was reported in 4 studies (3, 16, 17, 19), with an overall percentage of type II diabetes of 98.91% (type I: 8; type II: 727). Data about Wagner and TEXAS ulceration grades were reported in 4 (3, 15, 18, 20) and 3 (16, 17, 19) studies, respectively. In TTT and non-TTT groups, 3 and 2 patients were lost to follow-up respectively. The quality assessment result of the studies is presented in Supplementary Table S1 .

The detailed operation process is summarized in Table 2 . A routine debridement on the ulceration site was performed at the same time in most of the studies (15–17, 19, 20). Vacuum sealing drainage (VSD) and antibiotic bone cement (ABC) were applied at ulceration site in 2 (17, 20) and 1 (16) studies, respectively. In the peri-operative period, the managements mainly include the following four aspect: (1) antibiotics treatment according to drug sensitivity test; (2) debridement and drainage; (3) blood sugar controlling; and (4) dressing changing and disinfection. The operation was performed under general anesthesia (15, 20), nerve block anesthesia (3, 16, 17, 19), or lumbar anesthesia (15, 19, 20). The transportation site mainly located at the anteromedial area of the tibia, but the heights of the bone window were divergent among these studies. In Fan et al. (15), bone window was located at 10-20 cm below knee as they stated. In Ding et al. (16), Yuan et al. (17), and Chen et al. (19), bone window was located at the proximal tibia. In Ou et al. (3) and Zeng et al. (18), transportation site was located at middle or distal tibia. The size of bone window was reported in three studies (3, 18, 19), with different sizes. Generally, two individual pins were applied at the bone block to transport it and the tibia shaft to fixing the external fixator, respectively. All of the studies initiated the transportation at the time of 3-5 days post-operatively. Two different protocols to transport the bone block were reported: (1) 1mm per day for 14 days (3, 15, 16, 20); (2) 0.25 mm per 6h for 14 days (17, 19). Three different protocols to reset the bone block were reported: (1) 1mm per day for 14 days (15, 20); (2) 0.25 mm per 6h for 14 days (19); (3) 2 mm per day for 7 days (3, 16, 17).

Figure 2 shows the treatment outcomes of the TTT operation at the final follow-up. The pooled healing rate was 0.96 (95% confidence interval [95%CI]: 0.93~0.98; see Figure 2A ), using a fixed-effect model. The pooled limb salvage rate was as high as 0.98 (95%CI: 0.95~1.00; see Figure 2B ) after treatment with TTT. The pooled mean healing time was 15.03 (95%CI: 9.05~21.00; see Figure 2C ) months.

When compared with the pre-operative baseline values, the ABI (random-effect model; MD: 0.23; 95%CI: 0.03~0.44; p<0.001; see Figure 3A ), skin temperature (random-effect model; MD: 1.56; 95%CI: 0.30~2.81; p<0.001; see Figure 3B ), and VAS-pain scale (random-effect model; MD: 3.70; 95%CI: 1.97~5.44; p<0.001; see Figure 3C ) were all significantly improved at the final follow-up.

When compared with non-TTT group, the TTT group was associated with higher healing rate (OR: 10.43; 95%CI: 3.96~27.43; p<0.001; see Figure 4A ) and limb salvage rate (OR: 9.65; 95%CI: 3.30~28.20; p<0.001; see Figure 4B ) as shown in Figure 4 .

Concerning the complications of the TTT process, (1) the pooled risk of fracture at the transportation site was 0.02 (95%CI: 0.00~0.04; see Figure 5A ); (2) the pooled pin-site infection incidence was 0.08 (95%CI: 0.00~0.22; see Figure 5B ); (3) the DFU recurrence rate in TTT group was significantly lowered comparing to that of the non-TTT group (RR: 0.18; 95%CI: 0.06~0.49; p=0.001; see Figure 5C ).

The forest plot of sensitivity analysis for healing rate in TTT group is presented in Supplementary Figure S1 . One study (Yuan et al. (17)) was found to cause instability on the pooling result, thus it was omitted from the final pooling (see the final forest plot in Figure 2 ).

The forest plot of sensitivity analysis for limb salvage rate in TTT group is presented in Supplementary Figure S2 . There was no study was found to cause instability on the pooling result. Significant publication bias was detected according to Egger’s (p = 0.019) and Begg’s test (p = 0.327). Thus, non-parameter trim-and-fill method was performed to adjust the bias (see Supplementary Figure S3 ), in which three studies were filled. The adjusted effect size was 0.99 (95%CI: 0.96~1.00).

The forest plot of sensitivity analysis for mean healing time in TTT group is presented in Supplementary Figure S4 . There was no study was found to cause instability on the pooling result. No significant publication bias was detected according to Egger’s (p = 0.125) and Begg’s test (p = 0.624).

In patients with diabetic foot, the peripheral neuropathy and vascular disease are frequently encountered, which would develop to ulceration and even amputation. The DFU, as a terminal complication of the diabetes, is of quite complex pathogenesis, being derived from a combined action of ischemia, mechanical injury, infection, and so on. Usually, the ischemia and hypoxia status is caused by damage of the small blood vessels which could not be rescued by vascular surgery (21). TTT, as a novel developed technique which was based on the Ilizarov tension-stress law, has been recently used for treatment for DFU patients with primarily satisfied success rate (18–20). These studies demonstrated that repeated mechanical stretching of the tibia bone block could stimulate the regeneration of blood vessel and accelerate the ulceration healing.

Yang et al. (22) explored the biological mechanism of the TTT procedure in rat model, and demonstrated that TTT was associated with higher blood flow in the wound area according to laser speckle imaging, and enhanced neovascularization according to double immune-labelling of CD31 and α-Smooth Muscle Actin (α-SMA). Previous studies have shown that bone distraction could enhance neovascularization through a pathway involving chemokine stromal cell-derived factor-1 (SDF-1), which is a key factor responsible for homing and migration of endothelial progenitor cells (23). In a case-control study by Chen et al. (19), they compared the treatment outcome of severe and recalcitrant DFUs with TTT and solitary standard operation, showing that tibial transverse distraction group had higher healing rate, limb salvage rate, density of small vessels, blood flow and blood volume, compared with the control group. In our results, the healing rate and limb salvage rate were demonstrated to be as high as 96% and 98% at final follow-up, and significant improvements on ABI (MD = 0.23), skin temperature (MD = 1.56) and VAS (MD = 3.70) were identified. When compared with control group, the healing rate (OR = 10.43) and limb salvage rate (OR = 9.65) were both obviously increased. These findings all confirmed the acceleration effect on neovascularization. The ABI and skin temperature is directly related with the microcirculatory perfusion of foot soft tissue. With an improved blood perfusion, sufficient oxygen and nutrition supplies can be guaranteed for ulceration healing.

However, though TTT was proven to be effective in promote ulceration healing, DFU is a multi-disciplinary condition which is difficult to be completely solved by sole TTT operation (24, 25). Many assistance procedures were applied at the same time, including debridement of the ulceration lesion (15–17, 19, 20), antibiotic bone cement filling (16), and vacuum sealing drainage (17, 20), which have all been proven to be valuable in promoting healing of ulcer wounds. At the peri-operative period, blood sugar controlling is the basic requirement to guarantee a hypoglycaemic condition for tissue repairing. Antibiotics (intravenous or per oral) according to drug sensitivity testing is also essential to control the infection and ensure the healing process. It had been reported that adequate foot care can prevent 80% of DFUs in diabetes patients (26) and effectively prevent amputation caused by ulcerations (27). Thus, it is of importance to continue standard wound care (dressing changing and disinfection) and off-loading casts in peri-operative period.

The procedure of TTT, however, is related with some potential complications, especially the fracture at the tibial bone window (15, 18, 19), infection of the pin site (15, 19) and skin necrosis at surgical site (10). Our results showed a total of 5 tibia fracture among 178 patients (pooled proportion: 2%), and 8 pin-site infections among 166 patients (pooled proportion: 8%). In Fan et al. (15), the authors suggested to avoid the fracture risk by establishing standard tibial osteotomy criteria and performing post-operative education on falling prevention. They also recommended to narrow the bone window for those patients with short stature. We firmly in favour of their proposal. Additionally, those with severe osteoporosis especially among the postmenopausal older women should be referred to standard osteoporosis treatments to prevent risk of fracture at osteotomy site. Moreover, it is of great importance to let the patients return for regular follow-up after operation. To avoid the risk of infection of pin channel, peri-operative antibiotics and daily wound care (especially pin site disinfection) are mostly important. The necrosis of local soft tissue is another major concerning during TTT, which is mainly caused by long-term continuous pressure on the skin overlying the anterior tibia (28, 29). The surgeons should try their best to preserve the blood supply of the skin flap and avoid excessive interference to the soft tissue. Post-operatively, close attention should be applied on the skin status, and termination of transportation is indicated if signs of ischemia or necrosis are evident.

In 2020, with the concerted efforts of experts in various disciplines, the “Expert Consensus on the Treatment of Diabetic Foot Ulcers Using Tibial Transverse Transport” (30) was published in China. It has emphasized the importance of further simplification of the external fixator aiming to reduce the risk of complications. During the operation period, tourniquet should not be applied, to protect the blood supply of lower limb. It is of vital significance to narrow the incision and bone window sizes and preserve the periosteum, as far as possible. Through these strategies, incidence of adverse events could be significantly reduced without addition on the difficulty of surgery process.