Transmetatarsal Amputation Outcomes When Utilized to Address Foot Gangrene and Infection: A Retrospective Chart Review

Abstract

A transmetatarsal amputation (TMA) is a widely utilized procedure to address foot gangrene and infection. Although a common procedure, so too are the associated complications. The purpose of this review was to evaluate TMA healing and to explore if there were associated variables correlating with healed vs. failed to heal TMA sites. To do so, the Medical Department Orthopaedics Division Electronic Database, West Virginia University, College of Medicine was retrospectively searched to identify all cases of TMAs (CPT code 28805) during the period of January 2011 through June 2019, and those variables that might impact TMA healing. Then both univariate and multivariable logistic regression analyses were performed to investigate the associations between these variables and TMA healing, and sensitivity analyses were also conducted to determine if the results resisted the influence of one unmeasured confounder. There were 39 patients (41 procedures) who would undergo a TMA. The mean average patient age was 53 (range 29-73) years old. The median postoperative follow-up period was 617 (range 199-3632) days. TMA mortality data revealed 0 deaths at 30 days, 2 (5.1%) at 1 year, 8 (20.5%) at 5 years. In our study, 29 (70.7%) of the TMAs would achieve primary healing at a median of 31 (range 16-253) days. When comparing the TMA healed group to the failed to heal group the following independent variables were considered: diabetes mellitus, HgA1c >8%, neuropathy, peripheral arterial disease, chronic kidney disease, active smoking status, previous surgery, and a clean margin metatarsal bone pathology specimen positive for osteomyelitis. Of the aforementioned, only neuropathy (odds ratio [OR] = 0.056, 95% confidence interval [CI] = 0-0.501, p = .0062) and positive bone margin (OR = 0.144, 95% CI = 0.022-0.835, p = .0385) were found to be significant in univariate logistic regression analysis. In multivariable logistic regression analyses where the potential confounders age, gender, and body mass index were accounted for, of the 8 independent variables of interest, only neuropathy (OR = 0.037, 95% CI = 0-0.497, p = .0036) remained significantly associated with the healing status. Neuropathy was present in 17 (58.6%) of the healed TMAs and in 12 (100%) of the failed to heal TMAs. However, the positive bone margin failed to reach statistical significance (OR = 0.079, 95% CI = 0-1.39, p = .1331). Results from another multivariable logistic regression model where a quadratic term for age was added revealed that positive bone specimen correlated with the TMA healing status with significance (OR = 0.051, 95% CI = 0.001- 0.560, p = .0404). A positive clean margin bone specimen was found in 3 (10.3%) of the healed TMAs and in 4 (44.4%) of the failed to heal TMAs. The sensitivity analysis where current ulceration was used as an unmeasured confounder indicated that the results regarding the association between neuropathy or positive bone margin and TMA healing, though inconclusive, resisted the influence of this unmeasured confounder. Hopefully these findings will be a beneficial addition to the current TMA literature and as such, further assist with informed surgical decision making.

Amputation has been utilized for many years to address several pathologies such as infection, tumor, trauma, vascular disease, and congenital anomaly. It is estimated that 185,000 amputations occur each year in the United States and that there are approximately 1.6 million persons living with limb loss; 86% of which are lower extremity amputations. To put this into perspective, 1 in 190 Americans is currently living with the loss of a limb. It is projected that the number of people living with limb loss will more than double by the year 2050 (1-3).

Limb amputation represents one of the oldest surgical operations with evidence of such dating back to Neolithic times (4). In regards to the transmetatarsal amputation (TMA), the first published description was by Claude Bernard and Charles Huette in 1855, but it was McKittrick et al. in 1949 who would popularize this procedure as an alternative to more proximal amputations when addressing gangrene or infection (5,6). The TMA continues to be utilized widely to address foot gangrene and infection when indicated as opposed to more proximal amputations due to the benefits of maintaining the heel and a variable portion of the foot lever. By preserving limb length one is able to reduce energy expenditure during ambulation, increase the weight-bearing surface thus helping to distribute plantar pressures, and increase amputee independence by eliminating the need for prosthesis (7-10).

The reduced mortality rate with TMAs has also been demonstrated thus adding yet another benefit to performing this procedure in lieu of a more proximal amputation when possible. The mortality rate associated with TMAs has been reported to be between 0% and 17% at 30 days (11-15), 14% to 33% at 1 year (11,16,17), 25% to 65% at 3 years (11,16-19), and 39% to 70% at 5 years (11,16,17,19). In comparison, below knee and above knee amputations have been associated with perioperative mortality rates ranging from 0.9% to 9.7% and 2.8% to 16.5% respectively (7,20-22). These ranges increase with time, as one would expect, with 5-year mortality rates ranging from 40% to 82% in those having undergone a below knee amputation and 40% to 90% in those having undergone an above knee amputation (23).

Although the benefits of a TMA compared to a more proximal amputation have been well documented, so to have the complications and variable healing rates associated with a TMA. The literature demonstrates the healing rate of a TMA ranges from 44% to 88% (12,14,18,19,24-29). Due to some of the disappointing outcomes, further studies have delved into patient factors and variables in hopes of finding predictors of healing and predictors of increased likelihood of revision surgery. These include: diabetes mellitus, neuropathy, peripheral arterial disease (PAD), chronic kidney disease (CKD), smoking, etc. (11,12,14,18,19,24-27,29-33). However, many of these studies are inadequate in that they lack details, which does not assist in guiding real world practice. For example, while it is generally understood that neuropathy is deleteriously associated with postoperative outcomes, none of the studies specifically detailed neuropathy as an independent factor alone or in combination with diabetes. Also, clean margin metatarsal bone pathology specimens have not been specifically detailed in TMA healing. Then the purpose of this study was to evaluate the healing of our patient population undergoing a TMA to address gangrene or infection and to determine if there were patient factors or variables that correlate with postoperative outcomes, and if so, how patient factors or variables correlate with postoperative outcomes. The hope is that our findings will add to the current paucity of literature to help predict postoperative healing thus assisting with tailoring both surgeon decision making and patient outcome expectations.

Patients and Methods

From January 2011 through June 2019, all patients who had undergone a TMA by the Department of Orthopaedics were identified from our medical department division electronic database, West Virginia University, College of Medicine, using the Current Procedural Terminology code 28805 (Amputation, foot; Transmetatarsal). This study period was selected, as patient electronic medical records were readily available dating back to 2011 and with an end date of June 2019 to allow for at minimum a 6-month follow-up at time of conduction of this study. Our institutional review board approved the present study. The medical records of the patients were then reviewed which revealed 47 consecutive patients (49 feet) who underwent a TMA. When reviewing surgical indication, 6 patients (6 feet) were traumatic in origin and the remaining 41 patients (43 feet) were nontraumatic in origin. The 41 patients (43 feet) having undergone a nontraumatic TMA were for either infection or gangrene. No patient in our study underwent a TMA to address malignancy or congenital anomaly. Further review excluded 2 patients (2 feet) due to failure to meet the follow-up minimum of 6 months. Both of these patients underwent a TMA in the inpatient setting and upon hospital discharge would choose to follow-up at an outside facility. Thus, our study group consisted of 39 patients (41 feet) who underwent a TMA to address either gangrene or infection. A TMA was performed by 1 of 3 podiatric surgeons, Drs. Cherie Kelly-Danhires (C.K.D.), Rusty Cain (R.C.), Kathryn Bosia (K.B.) or by a foot and ankle fellowship trained orthopaedic surgeon, Dr. Robert Santrock (R.S.). The technique used was similar in all cases as previously described in the literature. The primary outcomes reviewed included the mortality rate up to 5 years postoperative and surgical healing rate. Predictors thought to impact TMA failure were reviewed to determine statistical significance, which included: diabetes mellitus, HgA1c >8%, neuropathy, PAD, CKD, active smoking status, previous surgery, and a clean margin metatarsal bone pathology specimen with findings consistent with or suggestive of osteomyelitis. Confounders accounted for included age, gender, and BMI. In addition, an unmeasured confounder current ulceration status was used in sensitivity analyses.

Operative Technique

The patient was placed on the operating table in the supine position. A fish-mouth style incision was performed at the distal aspect of the foot, carried full-thickness down to the level of metatarsal bone. Metatarsal bone resection was performed following reflection of soft tissue allowing for adequate exposure. Metatarsal resection was performed creating a beveled plantar edge and with residual metatarsal length mimicking a normal anatomic parabola. Cultures and/or pathology were obtained following copious irrigation. Irrigation was performed with Cysto Tubing on gravity or by way of pulsed lavage. Addition of antibiotic to the irrigation or to the wound bed following irrigation was at surgeon discretion. Residual metatarsal bone stumps were rasped smooth. Closure was performed in layers and skin closure was loosely approximated to avoid excessive tension. Soft dressing was applied to the operative extremity. Splint or durable medical equipment application was at surgeon discretion.

Statistical Methods

To explore the potential associations between the predictors of interest and the postoperative outcome, a 3-phase statistical analysis was run. In phase one, univariate logistic regression analyses were conducted, where the outcome variable was the healing status (Y or N), and the independent variable was each of the 8 candidate predictors, chosen based upon extensive literature review. Permutation test was used for inference in the univariate logistic regression models due to the small sample size in the current research (34). In phase two, multivariable logistic regression analyses were run, where potential confounders age, gender, and BMI, selected based on diabetic foot infection/amputation literature, were added to each of the univariate logistic regression models in phase one. Permutation test was used for inference in this multivariable logistic regression also due to the small sample size. In phase three, sensitivity analyses were conducted to investigate the potential impact of the unmeasured confounder current ulceration status on the estimated association between neuropathy or positive bone margin and TMA healing status (35). All statistical analyses were performed using R (Version 4.0.2), the “glmperm” (Version 1.3-1) and “episensr” (Version 0.9.6) packages.

Results

39 patients underwent 41 TMAs from 2011 to 2019. 23 TMAs were performed at the right foot and 18 were performed at the left foot, 2 of which were bilateral. A TMA was performed on 25 males (64.1%) and 14 females (35.9%). 24 TMAs were performed by podiatric surgeons (58.5%; C.K.D., 17 [41.5%], R.C., 4 [9.7%], K.B., 3 [7.3%]). 17 TMAs were performed by a foot and ankle fellowship trained orthopaedic surgeon (41.5%; R.S., 17). TMAs were performed in the inpatient setting in 32 (78%) cases and in the outpatient setting in 9 (22%) cases. For those undergoing the procedure inpatient; the median hospital stay was 17 (range 1-93) days. When considering the entire patient population the mean average patient age was 53 (range 29-73) years old. The median patient BMI was 32.8 (range 18-47.3). The median postoperative follow-up period was 617 (range 199-3632) days, thus approximately 20.3 months. The overall patient population past medical history is detailed in Table 1. Median lab values for the entire population were as follows; white blood cell count <1 week prior to date of surgery was 9000 (range 4700-13,400)/mm³, erythrocyte sedimentation rate (ESR) <3 days prior to date of surgery was 70 (range 17-145)mm/hr, c-reactive protein (CRP) <3 days prior to date of surgery was 9.4 (range 0.3-250.9)mg/L, albumin <1 month from date of surgery was 2.6 (range 1.6-4.5)g/dL. The aforementioned labs were available for all patients with the exception of 3 patients lacking an ESR lab value. All patients with a past medical history significant for diabetes mellitus had a hemoglobin A1c obtained within 3 months of the date of surgery with a median value of 8.4 (range 5.4-14.9)%. A clean margin specimen for microbiological analysis, whether soft tissue or bone, was obtained in 23 (56.1%) procedures. The specimen returned positive for bacterial growth in 17 (73.9%) of these procedures and negative for bacterial growth in 6 (26.1%) of these procedures. A clean margin bone specimen from a residual metatarsal stump was obtained for pathologic examination in 38 (92.7%) procedures. The specimen returned positive for findings consistent with or suggestive of osteomyelitis in 7 (18.4%) of these procedures and negative in 31 (81.6%) of these procedures. Topical Vancomycin powder was applied to the wound bed prior to closure in 6 (14.6%) procedures. An advanced microcurrent-generating dressing was utilized in 4 (9.8%) procedures. History of previous surgeries to the foot receiving a TMA was noted to have occurred a median 1 (range 0-9) time(s). When considering previous surgeries to the foot receiving a TMA it was noted that surgeries involving resection of bone, such as a toe amputation or metatarsal resection, occurred a median 1 (range 0-6) time(s). An ulceration preceded a TMA in 27 (65.9%) of our patients. The time from ulcer presentation to TMA was found to be a median 183 (range 8-3086) days. A TMA would subsequently require a more proximal amputation in 9 (21.9%) cases, thus 32 (78.1%) cases would retain a portion of all metatarsal bases. We further reviewed patient charts to determine the incidence of contralateral lower extremity surgery to address gangrene or infection. Our results revealed 24 (58.5%) had undergone surgical intervention at the opposite foot while 17 (41.5%) had no history of surgery at the opposite foot. Interestingly, every patient with a history of contralateral lower extremity surgical intervention for gangrene or infection suffered some form of amputation at that contralateral side.

Table 1.

Patient past medical history (TMA procedures, N = 41)

Patient Past Medical History	n (%)
Diabetes mellitus	29 (70.7%)
Neuropathy	29 (70.7%)
PAD	22 (53.7%)
History of lower extremity vascular intervention	14 (34.15%)
CAD	21 (51.2%)
CKD	17 (41.5%)
Active smoking status	14 (34.15%)
Former smoking status	14 (34.15%)
Never smoker	13 (31.7%)

Of the 41 TMAs, 29 (70.7%) would heal the primary amputation at a median of 31 (range 16-253) days whereas 12 (29.3%) would fail to achieve a healed intact skin envelope following the procedure. Our overall mortality data revealed 0 deaths at 30 days, 2 (5.1%) at 1 year, 8 (20.5%) at 5 years. In the healed TMA population the 1y mortality was 2 (7.4%), 5y mortality was 4 (14.8%). In the failed to heal TMA population there were no deaths at 1y. The 5y mortality was 4 (33.3%).

The mean average patient age in the healed population was 53.7 (range 29-73) years old and in the failed to heal population was 51.5 (range 32-72) years old. The median BMI in the healed population was 31.5 (range 18-47.3) and in the failed to heal population was 32.2 (range 22-40.6). The median follow-up period from date of TMA was 388 (range 199-3632) days in the healed population and 637 (range 225-2777) days in the failed to heal population. History of previous surgeries to the foot undergoing a TMA was a median 0 (range 0-9) time(s) in the healed population and 3 (range 0-5) time(s) in the failed to heal population. History of previous surgeries involving the removal of bone at the foot undergoing a TMA was a median 0 (range 0-6) time(s) in the healed population and 1 (range 0-5) time(s) in the failed to heal population. History of previous contralateral foot surgery to address gangrene or infection was noted in 18 (62.1%) of the healed population and in 6 (50%) of the failed to heal population. For those undergoing a TMA inpatient, those that healed had a median hospital stay of 17 (range 1-93) days compared to a median hospital stay of 12 (range 5-37) days in those that failed to heal. The median lab values for those that healed and those who failed to heal a TMA are shown in Table 2.

Table 2.

Patient median lab values (TMA procedures, N = 41)

	WBC	ESR	CRP	HgA1c	Albumin
Healed n = 29	9300/mm3 (4700-13,400/mm3)	70 mm/hr (17-145 mm/hr)	9.0 mg/L (0.3-250.9 mg/L)	8.4% (7%-13.4%)	2.6 g/dL (1.6-4.5 g/dL)
Failed to heal n = 12	7600/mm3 (5400-10,100/mm3)	69.5 mm/hr (22-135 mm/hr)	9.8 mg/L (0.6-228.5 mg/L)	9.3% (5.4%-14.9%)	2.7 g/dL (2.0-3.5 g/dL)

Of the 29 healed TMAs, 10 (34.5%) would subsequently develop an ulceration at a median of 50 (range 42-708) days from the date of noting a healed TMA site. Of these 10, 5 (17.2%) would subsequently heal with local wound care and the remaining 5 (17.2%) would require return to the operating room. Four (13.8%) of these patients would undergo further debridement, but would not require a more proximal amputation thus retaining all metatarsal bases. Only 1 (3.4%) patient in this group would undergo a more proximal amputation in the form of a below knee amputation. The median time of return to the operating room in the 5 patients requiring this was 113 (range 21-1726) days. Of note, the patient that subsequently required a below knee amputation had this procedure performed 4.75 years status post TMA. Overall 5 (17.2%) TMAs in the healed population would subsequently require return to the operating room and 1 (3.4%) would require a more proximal amputation.

In those who failed to heal the primary TMA, 11 (92%) would return to the operating room. The median time of return to the operating room was 42 (range 12-113) days. Of the 12 patients who failed to heal the TMA, 8 (66.7%) would require a more proximal amputation. Of the 8 patients requiring a more proximal amputation the levels of amputation were as follows: 1 Lisfranc, 6 below the knee amputations, 1 above the knee amputations.

When considering past medical history factors, the following data were ascertained for both the healed and failed to heal populations as represented in Graph 1: history of diabetes mellitus (healed 20 (69%), failed to heal 9 (75%)), neuropathy (healed 17 (58.6%), failed to heal 12 (100%)), PAD (healed 16 (55.2%), failed to heal 6 (50%)), history of lower extremity vascular intervention (healed 10 (34.5%), failed to heal 4 (33.3%)), coronary artery disease (CAD) (healed 15 (51.7%), failed to heal 6 (50%)), CKD (healed 11 (37.9%), failed to heal 6 (50%)) with stage of disease (healed median 3 (range 1-4), failed to heal median 3 (range 2-5)), smoking status in the healed population (active 9 (31%), former 11 (37.9%), never 9 (31%)), smoking status in the failed to heal population (active 5 (41.7%), former 3 (25%), never 4 (33.3%)). A clean margin specimen for microbiologic analysis, whether soft tissue or bone, was only obtained in 13 (44.8%) of the healed population. Of the 13 specimens collected, 7 (53.8%) were positive for bacterial growth. A clean margin specimen for microbiologic analysis, whether soft tissue or bone, was obtained in 10 (83.3%) of the failed to heal population. Of the 10 specimens collected, 8 (80%) were positive for bacterial growth. A clean margin bone specimen from a residual metatarsal stump was obtained from all patients in the healed population for pathologic analysis. The results revealed 3 (10.3%) were positive for findings consistent with or suggestive of osteomyelitis while 26 (89.7%) were negative for evidence of osteomyelitis. A clean margin bone specimen from a residual metatarsal stump was obtained from 9 (75%) of the TMAs that failed to heal for pathologic analysis. Of this population, 4 (44.4%) were positive for findings consistent with or suggestive of osteomyelitis while 5 (55.6%) were negative for evidence of osteomyelitis. Topical Vancomycin powder use was noted in 5 (17.2%) of the healed population TMAs and in 1 (8.3%) of the failed to heal population TMAs. An advanced microcurrent-generating dressing was noted to having been utilized in 3 (10.3%) of the healed population TMAs and in 1 (8.3%) of the failed to heal population TMAs. The supplemental use of topical Vancomycin powder with a TMA began in 2018 and the addition of microcurrent-generating dressing began in 2019.

Graph 1.

TMA healed vs. failed to heal patient past medical history (%) (TMA procedures, N = 41).

Univariate logistic regression analyses revealed that only neuropathy (OR = 0.056, 95% CI = 0-0.501, p = .0062) and a positive pathology finding of osteomyelitis of the residual clean margin metatarsal bone (OR = 0.144, 95% CI = 0.022-0.835, p = .0385) were significantly associated with the healing status as seen in Table 3. Multivariable logistic regression analyses revealed that after controlling for potential confounders age, gender, and BMI, only neuropathy (OR = 0.037, 95% CI = 0.001-0.497, p = .0036) was still significantly associated with the healing status as seen in Table 4. To further investigate why positive bone margin ceased to be significant, another multivariable logistic regression analysis was run where a quadratic term for age was added, and positive bone specimen correlated with the TMA healing status with significance (OR = 0.051, 95% CI = 0.001-0.560, p = .0404). The results revealed that the age variable complicated the association between positive bone margin and the healing status, which deserves further investigation in the future. Although current ulceration has been suggested to be an important confounder of the association between neuropathy or positive bone margin and the TMA healing status, current ulceration measures were not available. To determine if the results resisted the influence of this unmeasured confounder, sensitivity analyses were conducted. Sensitivity analyses revealed that for neuropathy, with bias parameters listed in Table 5, which were from an educated guess, the results (OR = 0.054, 95% CI = −3.884-2.565) were similar to those from the corresponding univariate logistic regression analysis, except that the confidence interval contained 1; for positive bone margin, with bias parameters listed in Table 6, also from an educated guess, the results (OR = 0.162, 95% CI = −12.047-14.370) were similar to those from the corresponding univariate logistic regression analysis as well, except that the confidence interval contained 1.

Table 3.

Univariate logistic regression analysis of TMA healing association with predictors (TMA procedures, N = 41)

Variable	Odds Ratio (95% CI)	p Value
Diabetes	0.741 (0.140, 3.210)	.6059
HgA1c >8%	0.907 (0.593, 1.390)	.6643
Neuropathy	0.056 (0, 0.501)	.0062
PAD	1.231 (0.314, 4.842)	.8221
CKD	0.611 (0.153, 2.407)	.3690
Active smoker	1.111 (0.245, 4.576)	.99
Previous Ipsilateral foot surgery	0.708 (0.438, 1.086)	.1494
Positive bone margin	0.144 (0.022, 0.835)	.0385

Note: Permutation of regressor residuals tests (Potter, 2005) were conducted to derive p values.

Table 4.

Multivariable logistic regression analysis of TMA healing association with predictors with age, gender, and BMI being accounted for (TMA procedures, N = 41)

Variable	Odds Ratio (95% CI)	p Value
Diabetes	0.659 (0.091, 3.887)	.6763
HgA1c >8%	1.001 (0.582, 1.710)	.9976
Neuropathy	0.037 (0.001, 0.497)	.0036
PAD	0.504 (0.057, 3.714)	.5352
CKD	0.631 (0.134, 2.908)	.5875
Active smoker	0.602 (0.096, 3.211)	.5878
Previous Ipsilateral foot surgery	0.733 (0.454, 1.147)	.2080
Positive bone margin	0.079 (0.004, 0.686)	.1331

Note: Permutation of regressor residuals tests (Potter, 2005) were conducted to derive p-values.

Table 5.

Bias parameter distributions for a probabilistic bias analysis of the relationship between neuropathy and TMA healing (N = 41) stratified by current ulceration

Bias Parameter	Description	Min	Max	Estimate	95% CI
p1 (%)	Prevalence of current ulceration among those with neuropathy	85	99
p2 (%)	Prevalence of current ulceration among those without neuropathy	0	20
OR	Odds ratio between ulceration and TMA healing			0.04	0.0 0.79

Note: Monte Carlo sampling techniques (Lash et al., 2011) were used.

Table 6.

Bias parameter distributions for a probabilistic bias analysis of the relationship between positive bone margin and TMA healing (N = 41) stratified by current ulceration

Bias Parameter	Description	Min	Max	Estimate	95% CI
p1 (%)	Prevalence of current ulceration among those with positive bone margin	70	75
p2 (%)	Prevalence of current ulceration among those without positive bone margin	35	50
OR	Odds ratio between ulceration and TMA			0.04	0.0 0.79

Note: Monte Carlo sampling techniques (Lash et al., 2011) were used.

Discussion

Our results demonstrated a healing rate of 70.7% (29 TMAs) in the sample. Of the 12 TMAs that failed to heal, 8 (66.7%) would require a more proximal amputation. We further compared the 2 populations, healed vs. failed to heal, to determine if there were variables that would help to predict outcome. We took into consideration diabetes mellitus, HgA1c >8%, PAD, CKD, active smoking status, previous surgery, neuropathy and the presence of a residual metatarsal bone stump positive for residual osteomyelitis on pathology review.

The incidence of diabetes mellitus, similar to amputation rates, is on the rise. It is estimated that approximately 34 million Americans, 10.5% of the U.S. population, have diabetes. Another 88 million have prediabetes, a condition that if not treated often leads to type 2 diabetes within 5 years (36). This is an important consideration as diabetes is the most common underlying cause of lower extremity amputations in the United States, as diabetics comprise approximately 82% of those undergoing lower extremity amputations (37,38). The incidence of lower extremity amputation in the diabetic population ranges from 78 to 704 per 100,000 person-years with the relative risks between diabetic and nondiabetic patients varying between 7.4 and 41.3 (39-44). Individuals with diabetes have a 15- to 46-fold greater risk of lower extremity amputation than those without diabetes (45,46). Due to this alarming figure, we reviewed our outcomes to determine if a history of diabetes mellitus correlated with a failure to heal a TMA. A history of diabetes mellitus was noted in 20 (69%) of the healed population TMAs and in 9 (75%) of the failed to heal population TMAs. The median hemoglobin A1c in the healed population was 8.4 (range 7-13.4)% and in the failed to heal population was 9.3 (range 5.4-14.9)%. Although there are several studies in the foot and ankle literature that demonstrate a correlation between abnormally elevated hemoglobin A1c and postoperative complications (47-49), the literature in regards to TMA healing has been mixed. Some studies have demonstrated a correlation between poor glycemic control and postoperative TMA healing complications (12,25,26,31,32) with patients with noninsulin-dependent diabetes being at a 5.4 times greater likelihood of requiring a more proximal amputation (25). While other studies, similar to ours, have not found a correlation between diabetes and postoperative TMA healing complications (11,14,17,18).

Commonly associated with diabetes mellitus, neuropathy affects approximately 20 million Americans (50). Neuropathy predisposes patients to trauma as well as infection thus affecting the healing of a TMA. In our study, 29 (70.7%) patients undergoing a TMA suffered with neuropathy. Neuropathy was determined on clinical exam utilizing both a 5.07/10-g Semmes-Weinstein monofilament and a 128 Hz Tuning Fork. We compared the incidence of neuropathy in the healed and failed to heal populations. Neuropathy was present in 17 (58.6%) of the healed TMAs and was present in 12 (100%) of the failed to heal TMAs. Although the incidence of diabetes mellitus was not found to significantly increase the likelihood of TMA healing failure, the presence of neuropathy did significantly correlate with failure to heal, which, though inconclusive in the sensitivity analysis, was evident in the univariate and multivariable logistic regression analyses. This correlates with previous studies that have demonstrated the deleterious effects neuropathy has on healing (48,51,52). Malay et al. (52) assessed the risks of failure to heal a lower extremity amputation to address a diabetic neuropathic foot ulceration. Their findings argue in support of performing foot amputations at or proximal to the tarsometatarsal joint as opposed to a more distal site when addressing diabetic neuropathic foot ulcerations. Wukich and colleagues (48) demonstrated that the presence of neuropathy increases the risk of surgical site infection even in patients without diabetes. Most neuropathies affect the 3 types of nerve fibers (motor, sensory, autonomic) to a varying degree. The damage to sensory fibers can result in the decline of a patient’s ability to detect light touch, temperature and pain thus predisposing patients to injury and potential delay in care. The damage to motor fibers is linked to diminished tendon reflex and the development of hammertoe deformity due to weakness of the lumbricals and interosseous muscles. Progression of motor neuropathy can lead to further atrophy of intrinsic muscles of the foot as well as gait instability. Autonomic dysfunction can lead to alterations in microcirculation resulting in a reduced vasodilatory response thus affecting perfusion. In addition, autonomic neuropathy can affect sweat glands resulting in dry or cracked skin predisposing patients to ulceration (51,53). A combination of these factors likely explains why those with neuropathy were more likely to suffer failure to heal a TMA in our study.

Another variable evaluated in our study was a history of PAD. It has been estimated that approximately 8.5 million people in the United States have PAD and that the risk of PAD increases substantially with age (54,55). The prevalence of amputation in PAD patients has been shown to be approximately 3% to 4% (56). Patients in our study with poor pedal perfusion on clinical exam, determined by weakly or non-palpable pedal pulses, would be diagnosed with PAD and all would subsequently undergo noninvasive vascular studies to further assess severity. When indicated, patients would undergo vascular surgical intervention to address the underlying PAD. Our overall patient population consisted of 22 (53.6%) suffering with PAD and with 14 (34.2%) of those having undergone vascular intervention prior to TMA. When reviewing the healed TMA population, the incidence of PAD was noted in 16 (55.2%) patients of whom 10 (34.5%) would undergo vascular intervention prior to TMA. When reviewing the failed to heal TMA population, the incidence of PAD was noted in 6 (50%) patients of whom 4 (33.3%) would undergo vascular intervention prior to TMA. Existing studies comparing TMA healing rates and PAD have had mixed results. Some studies have demonstrated an increased failure rate of healing a TMA with a history of PAD (11,14), while others have not demonstrated such an association (17,18,24,26,27,32). In our study we did not find PAD to significantly affect the postoperative healing of a TMA. We also did not find CKD to significantly affect the postoperative healing of a TMA. It has been estimated that approximately 37 million adults in the United States suffer with CKD, thus approximately 1 in 7 adults (57). CKD has previously been shown to be associated with an increased amputation incidence and mortality rate (58-60). Armstrong et al. (2010) demonstrated that the 10-year mortality among patients on dialysis who have an amputation is 3 times greater than those who require amputation without CKD (60). Our overall patient population consisted of 17 (41.5%) suffering with CKD. When reviewing the healed TMA population the incidence of CKD was noted in 11 (37.9%) patients with the stage of disease being a median 3 (range 1-4). When reviewing the failed to heal TMA population the incidence of CKD was noted in 6 (50%) patients with the stage of disease being a median 3 (range 2-5). Some studies, similar to ours, have not demonstrated an increased failure to heal a TMA with a history of CKD (17,26,32); however, there are existing studies that have revealed an association between CKD and an increased failure to heal a TMA (11,14,29,20).

The literature is robust as far as citing the deleterious effects smoking has on one’s health as well as its negative impact on surgical site healing. (61-66). Despite great efforts to educate patients on smoking-related diseases, approximately 13.7% of the U.S. population continues to smoke equating to an estimated 34.2 million adults. Cigarette smoking continues to be the leading cause of preventable disease and death in the United States, accounting for approximately 1 in 5 deaths (66). It has been shown that digital blood flow velocity can reduce by 42% for up to 1 hour after smoking just one cigarette (61). Evaluation of postoperative healing across surgical specialties have demonstrated complications occur significantly more often in smokers than nonsmokers. Sorensen et al. found necrosis to be 4 times more frequent in smokers than nonsmokers and found surgical site infection, dehiscence, delayed healing to be 2 times more frequent in smokers than nonsmokers (64). In forefoot surgery, Bettin et al. found active smokers suffered a notably higher complication rate of 36.4% when compared to the 8.5% in nonsmokers (65). In regards to the available literature on the relationship of tobacco use and TMA healing the results have been mixed. Some studies have shown increased complication rates with current tobacco use (32) while others have not demonstrated a correlation (17,18,27,28). In our study, 9 (31%) of those with healed TMA’s were active smokers in comparison to 5 (41.7%) of those who failed to heal a TMA, which failed to reach statistical significance.

In conclusion, we reviewed the pathology report of each case in which a clean margin metatarsal bone specimen was obtained and sent for pathologic evaluation. When positive, each patient would complete a standard 6 week course of parenteral antibiotic therapy. We found a positive clean margin in 10.3% of the healed TMA population vs. 44.4% in the failed to heal TMA population. Similarly in a study by Atway et al., findings of a positive bone margin following amputation resulted in worse outcomes. They found 81.8% of patients with a positive bone margin had poor outcomes, whereas 25% of patients with a negative bone margin had poor outcomes (67). Although in our study statistical significance was found in the univariate logistic regression analysis, this was not found when performing multivariable logistic regression analysis where age, gender, and BMI were accounted for. However, with a quadratic term of age further accounted for, the positive bone margin was significantly associated with TMA healing status, which, in combination with the corresponding inconclusive sensitivity analysis, warrants future studies.

Weaknesses of our study include the small TMA population size, 39 patients (41 cases). Another weakness with this study is the short follow-up period. Future studies will improve the quality by allowing for a longer postoperative assessment as well as a larger patient population. Lastly, being that this is a retrospective chart review, one must accept the inherit bias associated with this type of study. In conclusion, we found 70.7% of our patient population would primarily heal a TMA at a median of 31 (range 16-253) days. Our overall mortality data revealed 0 deaths at 30 days, 2 (5.1%) at 1 year, and 8 (20.5%) at 5 years. In regards to predictors associated with TMA healing, neuropathy was found to be significantly associated with TMA healing. Patients’ age complicated the association between the positive bone margin and TMA healing. Although it seems that the unmeasured ulceration status did not confound the association between either neuropathy or the positive bone margin and TMA healing, more studies are required to confirm this finding. Hopefully these findings will be a beneficial addition to the current literature on the management of foot gangrene and infection. As well as further support the deleterious association neuropathy has on postoperative outcomes.