Clinical and Kinematic Outcomes Based on Bone Graft Utilization for Salvage First Metatarsophalangeal Arthrodesis: A Systematic Review

Abstract

Motion-preserving options for the treatment of first metatarsophalangeal (MTP) osteoarthritis are appealing, but the conversion to arthrodesis in failed cases comes with additional challenges. Loss of first ray length may not only lower arthrodesis success rates but can also cause aberrancies in the biomechanics of the foot and ankle. Selection of the proper graft for the restoration of length is crucial in order to minimize the chance of post-operative complications. The primary objective of this systematic review was to determine the optimal graft type for MTP salvage arthrodesis in terms of clinical outcomes, kinematic outcomes, and bony integration to improve patient care. A systematic review was performed using PubMed, SPORTDiscus, Cumulated Index to Nursing and Allied Health Literature (CINAHL), MEDLINE, and Web of Science from database inception until 20 June 2023. Inclusion criteria were articles that examined clinical outcomes, examined different types of bone grafts, discussed impact of bone graft on lengthening, and articles related to first MTP arthrodesis salvage procedures. Data extraction relating to clinical metrics and kinematic metrics was performed and analyzed. Subgroup analysis was performed to compare graft types, such as (1) foot and ankle autograft, (2) non-foot and ankle autograft, and (3) allograft. A total of ten articles met eligibility criteria from 180 articles initially retrieved. Included patients (n = 164) had a frequency-weighted mean age of 55.2 ± 4.6 years with a frequency-weighted mean time from primary to salvage procedure of 36.6 ± 21.9 months and a frequency-weighted mean follow-up time of 42.7 ± 17.4 months. The non-foot and ankle autograft group had a mean length restoration of 4.4 ± 0.1 mm (n = 33, 73.3% reported) whereas the allograft group had a mean length restoration of 7.6 ± 3.5 mm (n = 49, 100% reported). The foot and ankle autograft group (n = 12 procedures) had an overall complication rate of 25.0%, the non-foot and ankle autograft group (n = 45 procedures) had an overall complication rate of 53.3%, and the allograft group (n = 49 procedures) had an overall complication rate of 10.2%. Preoperative AOFAS scores were lower but improved postoperatively, with the allograft group showing the highest postoperative scores, shorter time to union, and varying graft lengths among different autograft subgroups. The allograft group for salvage MTP arthrodesis has promise, as this group had the greatest mean length restoration and the lowest complication rate. This is the first systematic review examining different bone graft utilization for salvage MTP arthrodesis. More high-quality research is needed before solid recommendations can be made on this topic.

1. Introduction

Treatment of end-stage first metatarsophalangeal (MTP) osteoarthritis can be a challenging problem for foot and ankle surgeons. While surgical procedures such as soft-tissue arthroplasty, excisional arthroplasty, implant arthroplasty, cheilectomy, and first metatarsal osteotomy can temporarily restore motion, complete arthrodesis of the first MTP joint is the gold standard for end-stage MTP osteoarthritis [1,2,3,4,5]. Although motion preserving options are appealing to patients, the inevitable loss of first ray length is a complication that physicians must properly address when converting to arthrodesis for the first MTP joint [6,7].

The shortening of the first ray not only increases the risks of metatarsalgia, but also has upstream effects by altering hindfoot alignment [8,9]. As hindfoot alignment is governed by the weight-bearing tripod and the three-dimensional (3D) location of the MTP head, shortening of the first ray can increase risk of hindfoot injuries, development of progressive collapsing foot disease (PCFD), and alter biomechanical factors such as gait [9,10,11,12,13].

While iliac crest autograft is often used for correction of length in salvage arthrodesis, many other graft options are commercially available and can be used with surgeon ease [14]. Additionally, autograft can be expensive in areas with smaller budgets, so being able to explain possible benefits of all types of grafts could aid physicians when restoring length, regardless of budget [15]. While previous systematic reviews have investigated MTP salvage arthrodesis and various grafts, a review analyzing how different graft types impact first ray length restoration has not been conducted [16]. Within this review, comparison of graft type impact on kinematic and clinical outcomes included the use of autograft from the foot and ankle region, autograft from outside the foot and ankle, and, finally, allograft. The objective of this systematic review was to understand the impact of all graft types on salvage arthrodesis to aid physicians when deciding upon a graft for restoration of length, union rate, and patient success.

2. Methodology

2.1. Study Creation

This systematic review was completed in accordance with the most recent Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines [17]. A systematic review was performed on the topic of clinical outcomes for different bone graft types used during the first MTP arthrodesis salvage procedures using PubMed, SPORTDiscus, Cumulated Index to Nursing and Allied Health Literature (CINAHL), MEDLINE, and Web of Science from database inception until 20 June 2023. The search algorithm used in each of the five databases was (“bone block” OR “bone graft” OR “autograft” OR “allograft” OR “salvage” OR “primary” OR “bone”) AND (length OR lengthening OR “length correction”) AND (metatarsophalangeal OR MTP OR MTPJ OR “first ray”) AND (arthrodesis OR fusion).

2.2. Eligibility Criteria

Inclusion criteria for this systematic review were articles that examined clinical outcomes, articles that examined different types of bone grafts, articles that discussed the impact of a bone graft on lengthening, articles related to first MTP arthrodesis salvage procedures, and full-text articles in English. Exclusion criteria were articles without full-text, articles not in English, case reports, technical tips with only one patient, case series with less than five patients, articles not using bone graft for lengthening, articles with traumatic changes in length, articles without clinical outcomes, and articles without surgical procedure of first MTP arthrodesis.

2.3. Study Definitions

For the purposes of this systematic review, graft type was grouped into three categories for the pooling of data. The “foot and ankle autograft” group included any procedures that were completed using a bone autograft taken from the foot or ankle region. The “non-foot and ankle autograft” group included any procedures that were completed using a bone autograft taken from elsewhere in the body, other than the foot or ankle region, such as the iliac crest. The “allograft” group included any procedures that were completed using an allograft. Any articles that did not specify results by bone graft type were not placed into the aforementioned three groups and were eliminated from analysis.

2.4. Article Selection Process

Article screening was performed by two authors (ANB and GMT). After the search algorithm was used in the five databases, the retrieved articles were uploaded into Rayyan, a software commonly used in the literature for systematic reviews [18]. Articles were first screened for duplicates with all duplicates being removed manually. Then, articles were screened for inclusion by title and abstract and then later screened by full text for final inclusion per the aforementioned eligibility criteria.

2.5. Data Extraction

All data extraction was performed by one author (GMT). Data extracted included first author, year of publication, article type, information on bone graft, information on lengthening, number of procedures, average patient age (years), follow-up time, preoperative and postoperative American Orthopedic Foot and Ankle Society (AOFAS) score, success rate, time until union (months), graft length (mm), length restoration (mm), complications, non-union rate, time before salvage (months), preoperative and postoperative first MTP angle (degrees), preoperative and postoperative intermetatarsal angle (IMA, degrees), preoperative and postoperative hallux valgus (HV) angle (degrees), preoperative and postoperative dorsiflexion angle (DFA, degrees), preoperative and postoperative Visual Analog Scale (VAS) score, preoperative and postoperative Foot and Ankle Disability Index (FADI), as well as any relevant statistical significance values for any biomechanical outcomes in the included articles.

2.6. Article Quality Grading

All of the included biomechanical articles included in this study were graded via the Methodological Index for Non-Randomized Studies (MINORS) scale, as used previously in the literature [19]. Non-comparative studies are graded on a scale from 0–16 points whereas comparative studies are graded on a scale from 0–24 points, with each individual item worth 0–2 points [19].

2.7. Certainty Assessment

This systematic review utilized the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach for assessment of the certainty of the included evidence [20]. Overall assessment was assessed as high, moderate, low, or very low depending on the data [20].

2.8. Statistical Analysis

The statistical analysis for this systematic review used the Statistical Package for the Social Sciences (IBM Corp., Armonk, NY, USA) version 29.0. Frequency-weighted means were utilized to pool data from different studies. Due to the heterogeneity of the included articles as well as the observational nature of the included studies, meta-analysis was not performed for the included articles and a narrative approach to systematic review was undertaken for this study.

3. Results

3.1. Search Results

A total of ten articles met the eligibility criteria from the 180 articles initially retrieved from the five databases [3,21,22,23,24,25,26,27,28,29]. This systematic review utilized a PRISMA diagram to visually depict the search process from initial search to final article inclusion (Figure 1).

3.2. Study Quality Grading

All of the ten included articles were observational studies (retrospective cohort, retrospective case series) and were graded via the MINORS scale (

Table 1. Quality grading via the Methodological Index for Non-Randomized Studies (MINORS) scale for the included articles in this systematic review. Non-comparative articles are out of 16 points on the MINORS scale with each criterion worth 0–2 points.

Author (Year)	Study Type	Total MINORS Score	Clearly Stated Aim	Inclusion of Consecutive Patients	Prospective Collection of Data	End Points Appropriate to Study Aim	Unbiased Assessment of Study End Point	Follow-up Period Appropriate to Study Aim	Less than 5% Lost to Follow up	Prospective Calculation of the Study Size
Burke (2021) [21]	Non-Comparative	12	2	2	0	2	2	2	2	0
Hecht (1997) [22]	Non-Comparative	11	2	2	0	2	2	2	1	0
Usuelli (2017) [23]	Non-Comparative	12	2	2	0	2	2	2	2	0
Bei (2015) [24]	Non-Comparative	12	2	2	0	2	2	2	2	0
Luk (2015) [25]	Non-Comparative	12	2	2	0	2	2	2	2	0
Malhotra (2015) [26]	Non-Comparative	12	2	2	0	2	2	2	2	0
Myerson (2000) [27]	Non-Comparative	12	2	2	0	2	2	2	2	0
Malagelada (2018) [28]	Non-Comparative	12	2	2	0	2	2	2	2	0
Gross (2013) [3]	Non-Comparative	12	2	2	0	2	2	2	2	0
Brodsky (2000) [29]	Non-Comparative	12	2	2	0	2	2	2	2	0

). The mean score for the ten included articles on the MINORS scale was 11.9 ± 0.3 (range: 11.0–12.0). All of the ten included articles were non-comparative in nature and the total score was out of 16 points. All of the outcomes of this systematic review had a certainty of “very low” due to the observational nature of the included studies and the high risk of bias with lack of comparison groups [20].

3.3. Patient Demographics

There were a total of 164 patients who underwent first MTP arthrodesis salvage procedures (n = 173 procedures) from the ten included articles in this systematic review. Included patients (n = 164) had a frequency-weighted mean age of 55.2 ± 4.6 years (n = 164, 100% of patients reported) with a frequency-weighted mean time from primary to salvage procedure of 36.6 ± 21.9 months (n = 61, 37.1% of patients reported) and a frequency-weighted mean follow-up time of 42.7 ± 17.4 months (n = 152, 92.7% of patients reported).

For subgroups based on bone graft type, the foot and ankle autograft group had 12 patients, the non-foot and ankle autograft group had 44 patients, the allograft group had 44 patients, and 64 patients were not placed in a group as the reported results were not specific to graft type. One article had patients from the foot and ankle autograft group (n = 12) with a mean age of 58.4 years, a mean time from primary to salvage procedure of 15.0 months, and a mean follow-up time of 33.4 months. The non-foot and ankle autograft group (n = 44) had a frequency-weighted mean of 54.1 ± 3.1 years (n = 44, 100% of patients reported) and a frequency-weighted mean follow-up time of 55.9 ± 10.8 months (n = 32, 72.7% of patients reported) without any data on time until salvage procedure. The allograft group (n = 44) had a frequency-weighted mean of 59.6 ± 0.7 years (n = 44, 100% of patients reported) and a frequency-weighted mean follow-up time of 33.7 ± 10.1 months (n = 44, 100% of patients reported). Additional demographic information from individual articles was recorded for this systematic review (

Table 2. Patient demographics from the included articles in this systematic review. Data recorded included first author, year of publication, number of procedures, number of patients, average age, information on bone graft, bone graft grouping, information on lengthening, time until salvage procedure from primary surgery, and follow-up time. Abbreviations: MTP, metatarsophalangeal.

Author (Year)	Article Type	Patients (n)	Procedures (n)	Average Age	Bone Graft Information	Graft Group	Lengthening Information	Time before Salvage	Follow-Up (Months)
Burke (2021) [21]	Case Series	36	38	59.3 (range: 33–80)	Patellar wedge bone allograft. Crushed cancellous used to fill holes as needed	Allograft	9.5 mm (5–15). The preoperative minimum length of the proximal phalanx necessary to obtain adequate stability using 3 distal locking screws with the interposition graft was 17 mm	-	38.4 (range: 12–120)
Hecht (1997) [22]	Case Series	14	16	54 (range: 36–74)	33% cancellous bone from metatarsal head, 67% used tricortical iliac crest autograft	Unclassified	Described a length ratio (first ray with respect to second toe)	38 (range: 9–84)	55 (range: 36–94)
Usuelli (2017) [23]	Retrospective	12	12	58.4 (range: 25–77)	Autologous calcaneus bone graft (ipsilateral bone block fusion)	Foot and ankle autograft	Average size of block was 14.3 (range: 11–19)	15 (range: 12–22)	33.4 (range: 12–64)
Bei (2015) [24]	Retrospective	8	11	61 ± 10.5	Interpositional allograft	Allograft	Compared to contralateral foot. Subtracted 2 mm from MTP joint space. Mean of 1.1 mm of lengthening	-	12.6 (range: 6–23)
Luk (2015) [25]	Retrospective	15	15	57 (range: 31–68)	Interposition allograft bone block, used autograft from tibia and iliac crest to pack in the medullary cavity prior to placing the allograft bone	Unclassified	Lengthening was custom for each patient. Metatarsal head was reamed according to the degree of bone loss	-	11.5 (range: 4.75–24.25)
Malhotra (2015) [26]	Case Series	24	25	52.1 (range: 34–69)	Tricortical autograft	Non-foot and ankle autograft	Shortening of the first ray was determined via contralateral comparison. Measured the gap of the hallux during surgery by gently retracting.	-	62 (range: 11–117)
Myerson (2000) [27]	Retrospective	24	24	46.4 (range: 28–66)	Used structural autograft in 66% of patients and allograft from the tibial head was used in 33%. Used cancellous graft in cases with large holes to fill	Unclassified	Lengthening was specific to the patient. Looked for proper alignment and a graft that did not prevent wound closure. As long as the hallux was positioned at 15 degrees of dorsiflexion and 5–10 degrees of valgus with respect to the MT1, then it was considered good	26.6 (range: 3–64)	63 (range: 26–108)
Malagelada (2018) [28]	Retrospective	8	8	60.5 (range: 45–80)	Bi-cortical autograft from iliac crest	Non-foot and ankle autograft	Measured based upon size of the graft required appropriate first ray. Additional 2 cm were harvested in order to leave room for structuring	-	37.4 (range: 12–102)
Gross (2013) [3]	Retrospective	11	12	56.9 ± 10.2	Allograft or autograft. 5/12 were iliac, 1/12 were interpositional allograft from iliac, 1/12 were bone grafting with calcaneal autograft and bone putty. 4/12 were bone grafting with morselized metatarsal heads, and 1/12 had no graft	Unclassified	Defect size dependent	80 (range: 8–252)	33 (range: 4–144)
Brodsky (2000) [29]	Retrospective	12	12	54 (range: 35–72)	Tri-cortical iliac autograft and cancellous graft for all cases	Non-foot and ankle autograft	Defect size dependent	-	-

).

3.4. Clinical and Radiographic Outcomes

The entire cohort (n = 173 procedures) had a frequency-weighted mean preoperative AOFAS score of 44.7 ± 4.1 points (n = 97, 56.1% reported), a frequency-weighted mean postoperative AOFAS score of 77.2 ± 4.4 points (n = 109, 63.0% reported), a frequency-weighted mean time until union of 15.1 ± 4.8 weeks (n = 165, 95.4% reported), and a frequency-weighted mean graft length of 13.8 ± 4.8 mm (mm) (n = 126, 72.8% reported). For subgroup analysis, the foot and ankle autograft group (n = 12, one article) had a mean preoperative AOFAS score of 50.7 points, a mean postoperative AOFAS score of 73.8, a time until union of 11.0 weeks, and a graft length of 14.3 mm. The non-foot and ankle autograft group had no information on preoperative AOFAS score, had a mean postoperative AOFAS score of 70 points (n = 12, one article), a frequency-weighted mean time until union of 14.7 ± 1.9 weeks (n = 37, 82.2% reported), and a frequency-weighted mean graft length of 10.0 ± 1.0 mm (n = 37, 82.2% reported). The allograft group had a preoperative AOFAS score of 43.7 ± 0.4 points (n = 49, 100% reported), a postoperative AOFAS score of 79.6 ± 4.4 points (n = 49, 100% reported), a frequency-weighted mean time until union of 14.8 ± 2.2 weeks (n = 49, 100% reported), and a mean graft length of 10.5 (n = 38, one article). The allograft group had the highest postoperative AOFAS score (79.6 points versus 70.0 points and 73.8 points).

The entire cohort (n = 173 procedures) had a frequency-weighted mean preoperative IMA of 8.5 ± 2.0 degrees (n = 55, 31.8% reported), a frequency-weighted mean postoperative IMA of 8.3 ± 1.3 degrees (n = 55, 31.8% reported), a frequency-weighted mean preoperative HV angle of 15.0 ± 4.8 degrees (n = 47, 27.2% reported), a frequency-weighted mean HV angle of 13.3 ± 1.0 degrees (n = 47, 27.2% reported), a frequency-weighted mean preoperative DFA of 29.7 ± 5.3 degrees (n = 55, 31.8% reported), and a frequency-weighted mean postoperative DFA of 20.7 ± 4.5 degrees (n = 43, 24.9% reported). For the foot and ankle autograft group, the mean preoperative IMA was 9.5 degrees (n = 12, one article), the mean postoperative IMA was 8.4 degrees (n = 12, one article), the mean preoperative HV angle was 19.7 (n = 12, one article), and the mean postoperative HV angle was 14.3 degrees (n = 12, one article) with no information on the DFA. For the non-foot and ankle group, the mean preoperative IMA was 9.1 degrees (n = 12, one article), the mean postoperative IMA was 7.2 degrees (n = 12, one article), the frequency-weighted mean preoperative HV angle was 14.3 ± 6.0 degrees (n = 20, 44.4% reported), the frequency-weighted mean postoperative HV angle was 13.1 ± 1.3 degrees (n = 20, 44.4% reported), the mean preoperative DFA was 35.0 degrees (n = 12, one article), and the mean postoperative DFA was 22.0 degrees (n = 12, one article). There was no information on preoperative and postoperative IMA, HV, or DFA for the allograft group. The success rate ranged from 41.7% to 94.70% for included procedures, with the allograft group demonstrating higher success rates overall. The entire cohort (n = 173 procedures) had a frequency-weighted mean length restoration of 7.8 ± 3.9 mm (n = 106, 61.3% reported). No graft length information was available for the foot and ankle autograft group. The non-foot and ankle autograft group had a mean length restoration of 4.4 ± 0.1 mm (n = 33, 73.3% reported), whereas the allograft group had a mean length restoration of 7.6 ± 3.5 mm (n = 49, 100% reported). Overall, the allograft group had the greatest mean length restoration of all subgroups with included data. The information from the individual articles on clinical outcomes were recorded for this systematic review (

Table 3. Clinical and radiographic outcomes for the individual articles in this systematic review. Data recorded includes first author, year of publication, number of procedures, preoperative and postoperative American Orthopedic Foot and Ankle Society (AOFAS) score, and success rate.

Author (Year)	Procedures (n)	AOFAS Pre	AOFAS Post (Max 95)	Success Rate
Burke (2021) [21]	38	43.5	77.2	94.70%
Hecht (1997) [22]	16	-	-	80% for tricortical graft
Usuelli (2017) [23]	12	50.7 (range: 38–75)	73.8 (range: 60–90)	75%
Bei (2015) [24]	11	44.50 (range: 32–60)	87.75 (range: 76–96)	90.90%
Luk (2015) [25]	15	-	-	87%
Malhotra (2015) [26]	25	-	-	88%
Myerson (2000) [27]	24	40 (range: 22–60)	79 (range: 66–90)	79%
Malagelada (2018) [28]	8	-	-	87.50%
Gross (2013) [3]	12	51.7 ± 13.1	74.9 ± 8.6	41.70%
Brodsky (2000) [29]	12	-	70 (range: 52–82)	91.70%

). The information from the individual articles on radiographic outcomes were recorded for this systematic review (

Table 4. Radiographic outcomes for the individual articles in this systematic review. Data recorded includes first author, year of publication, number of procedures, length restoration, preoperative and postoperative intermetatarsal angle (IMA), preoperative and postoperative hallux valgus (HV) angle, and preoperative and postoperative dorsiflexion angle (DFA).

Author (Year)	Procedures (n)	Length Restoration (mm)	Preoperative IMA	Postoperative IMA	Preoperative HV Angle	Postoperative HV Angle	Preoperative DFA	Postoperative DFA
Burke (2021) [21]	38	9.5 (range: 5–15)	-	-	-	-	-	-
Hecht (1997) [22]	16	length ratio shortened from 1.17° (1.02–1.30) pre to 1.14° (1.06–1.21)	5.5 (−2–10)	7.3 (−1–11)	-	-	22.1 (8.5–64)	15.2 (5–40)
Usuelli (2017) [23]	12	length of first ray average decreased from 10.4 cm to 10.3 cm	9.5 (6.4–14.2)	8.4 (4.8–13.1)	19.7 (3.6–45.8)	14.3 (7.3–29.7)	-	-
Bei (2015) [24]	11	1.1 ± 4.5	-	-	-	-	-	-
Luk (2015) [25]	15	increased from mean of 10 cm to 10.6 cm	10.5 (4–17)	10.2 (4–15)	12.2 (−28–30)	12.7 (4–22)	33.9 (15–71)	25.6 (16–40)
Malhotra (2015) [26]	25	4.4 (range: 0–8)	-	-	-	-	-	-
Myerson (2000) [27]	24	13 (range: 0–29)	-	-	-	-	-	-
Malagelada (2018) [28]	8	4.6 (range: 2.7)	-	-	21.4 (2.6)	11.6 (3.2)	-	-
Gross (2013) [3]	12	-	-	-	-	-	29.1 (14.3)	-
Brodsky (2000) [29]	12	-	9.1 (5–12)	7.2 (5–10)	9.5 (−8–30)	14.1 (5–23)	35 (20–60)	22 (13–38)

).

3.5. Complications

There were a total of 50 reported complications out of 173 surgical procedures, indicating a 28.9% total complication rate. The most common complication was non-union (n = 18 cases), representing 36.0% of complications with a total complication rate of 10.4%. Other reported complications for first MTP arthrodesis salvage include neuroma (n = 3, 1.73% rate), superficial wound infection (n = 3, 1.73% rate), painless osteoarthritis (n = 1, 0.58% rate), stress fracture (n = 1, 0.58% rate), hardware removal (n = 4, 2.3% rate), deep infection (n = 1, 0.58% rate), numbness (n = 1, 0.58% rate), prominent plate (n = 2, 1.2% rate), graft site irritation (n = 1, 0.58% rate), and primary complaints of pain and/or discomfort (n = 15, 8.7% rate). Non-union rates varied from 5% to 25% for the included articles. Individual articles were assessed for complications and recorded for this systematic review for further examination (

Table 5. Complications for the individual included articles in this systematic review. Data recorded included first author, year of publication, number of procedures, graft group, non-union rate, and complications with number of cases and percentage of total procedures.

Author (Year)	Procedures (n)	Graft Group	Nonunion Rate	Complications
Burke (2021) [21]	38	Allograft	5%	Pain (n = 1, 2.6%), nonunion (n = 2, 5.3%),
Hecht (1997) [22]	16	Unclassified	-	Neuroma (n =1, 6.3%)
Usuelli (2017) [23]	12	Foot and ankle autograft	25%	Nonunion (n = 3, 25%),
Bei (2015) [24]	11	Allograft	9.09%	Superficial wound (n = 1, 9.09%), painful nonunion (n = 1, 9.09%)
Luk (2015) [25]	15	Unclassified	6.67%	Nonunion (n = 1, 6.6%), painless arthritis, (n = 1, 6.6%), stress fracture (n = 1, 6.6%)
Malhotra (2015) [26]	25	Non-foot and ankle autograft	-	Nonunion (n = 3, 12.0%), painful neuroma (n = 2, 8%), pain at graft site (n = 1, 4%), hardware removal (n = 3, 12%), discomfort (n = 6, 33%)
Myerson (2000) [27]	24	Unclassified	21%	Nonunion (n = 5, 21%), deep infection (n = 1, 4.2%), superficial wound infection (n = 2. 8.3%), pain syndrome (n = 1, 4.2%)
Malagelada (2018) [28]	8	Non-foot and ankle autograft	-	Removal due to discomfort (n = 1, 13%), pain (n = 1, 13%), numbness (n = 1, 13%)
Gross (2013) [3]	12	Unclassified	-	Painful nonunion (n = 2, 16.7%), prominent plate (n = 2, 16.7%), graft site irritation (n = 1, 8.35%)
Brodsky (2000) [29]	12	Non-foot and ankle autograft	8.33%	Discomfort (n = 1, 8.25%), sesamoid discomfort (n = 4, 33%), painless nonunion (n = 1, 8.25%)

).

For subgroup analysis, the foot and ankle autograft group (n = 12 procedures) had an overall complication rate of 25.0% (n = 3 complications) with three cases of non-union (25.0% complication rate). The non-foot and ankle autograft group (n = 45 procedures) had an overall complication rate of 53.3% (n = 24 complications) with four cases of non-union (8.9% rate), two cases of neuroma (4.4% rate), one case of graft site pain (2.2% rate), four cases of hardware removal (8.9% rate), 12 cases of primary pain and/or discomfort (26.6% rate), and one case of numbness (2.2% rate). The allograft group (n = 49 procedures) had an overall complication rate of 10.2% (n = 5 complications) with one case of primary pain and/or discomfort (2.0%), three cases of non-union (6.1%), and one case of superficial wound infection (2.0%). Overall, the allograft group had the lowest overall complication rate (10.2% versus 25.0% and 53.3%). All other complications (n = 9 complications) were from articles that did not distinguish complications by graft type.

4. Discussion

In this systematic review, the clinical and kinematic outcomes of salvage MTP arthrodesis for restoration of length were investigated to improve patient care and surgeon decision making. Due to the limited number of high-evidence studies, most analyses in this study were centered around describing the clinical outcomes of three different graft groups (foot and ankle autograft, non-foot and ankle autograft, and allograft) qualitatively.

While previous systematic reviews have summarized the current literature on salvage MTP arthrodesis, their study was limited to salvage procedures for failed arthroplasties and did not separate outcomes based on graft type [16]. As arthroplasty is not the only conservative treatment to treat MTP arthritis, our study included cases with failed bunion osteotomy, hemiarthroplasty, total joint arthroplasty, infected MTP joint, and even diseases such as gout. Additionally, while our review focused on clinical outcomes, it also had a strong focus on the kinematics of the MTP joint. Length restoration, IMA, HV angle, and DFA are all important metrics for not only the success of the arthrodesis, but the likelihood of preventing foot and ankle biomechanical issues post-operation [6]. As described by Linzt et al., the three-dimensional location of the plantar aspect of the MTP head is an important measurement within foot and ankle offset—a surrogate measurement for hindfoot alignment [9]. As length, IMA, DFA, and HV angle all directly alter the location of the MTP head, it is crucial that proper salvage arthrodesis is performed with caution to ensure proper kinematic alignment. Additionally, a shortening of the first ray can have negative effects on forefoot loading that can lead to decreased first ray plantar pressure, potentially increasing the risk of developing lesser toe metatarsalgia [8,13]. As stated by Geng et al., a shortening of greater than 6mm is outside the safe zone. Within this review, all studies reported a mean shortening of greater than 6 mm pre-operation, further demonstrating the importance of length restoration within salvage MTP arthrodesis [13].

When comparing the success of graft groups that reported kinematic results, the foot and ankle autograft group had the larger improvement in HV angle (−5.4° (n = 12)) when compared to the non-foot and ankle autograft group (−1.2° ± 1.3 (n = 20)). This change in HV angle is reflective of the difference in HV before and after surgery. As HV angle is a direct predictor of hallux valgus severity, the foot and ankle autograft had a greater impact on improving MTP alignment than non-foot and ankle autografts [30]. Ideally, this change in angulation would be related to the deviation from the planned outcome, as some salvage procedures may optimally require a smaller decrease in HV angle to be deemed a successful correction. However, as the articles that differentiate by graft type did not report planned outcomes, improvements in angulation had to be compared qualitatively based upon pre-operative alignment. For IMA angle, both autograft groups had similar improvements in alignment, with non-foot and ankle autografts having a slightly larger improvement (−1.9° (n = 12)) than autografts taken from within the foot and ankle (−1.1° (n = 12)). This similarity in IMA measurements is expected due to the nature of the IMA measurement. IMA was measured from the base of the first and second metatarsal; any change distally should not have much effect. While there was not enough high-level evidence to perform proper statistical analysis on these two kinematic angles, qualitatively on an absolute scale, foot and ankle autograft performed better at realigning the hallux towards more normal HV angulation compared to the non-foot and ankle autograft group, while also having similar results for IMA. As foot and ankle autograft can be taken without the need for a separate surgical site, it could potentially be a cheaper and more optimal way of correcting kinematic alignment in salvage MTP arthrodesis. Previous reviews did not comment on these kinematic measurements, and additional investigation into the effect of allograft on kinematic alignment should be explored in future studies.

As for first ray lengthening, foot and ankle autograft had a negligible effect on length restoration (average decrease of 0.1 cm, p = 0.42). As only one group used strictly foot and ankle autograft (no supplementing with crushed allograft), more investigation into this technique is needed to further understand its effect on length restoration. The non-foot and ankle autograft group had a mean length restoration of 4.4 ± 0.1 mm (n = 33, 73.3% reported), whereas the allograft group had a mean length restoration of 7.6 ± 3.5 mm (n = 49, 100% reported). Interestingly, both subgroups reported similar graft lengths used for length restoration (allograft = 10.5 (n = 38, one article)): non-foot and ankle autograft (10.0 ± 1.0 (n = 37, 82.2% reported)). As allograft had a greater length restoration, it can be inferred that allograft likely has greater bony integration. This could be due to allograft often being supplemented with additional orthobiologics as seen in the study by Bei et al., and the increased bony integration capabilities were also reflected union rates [24]. The two studies that used allograft for length restoration had the highest success rates at 94.70% and 90.90% [21,24]. However, as the studies included in our review were low-level evidence, a proper comparison study between allograft and non-foot and ankle autograft is needed to fully understand which graft is best for length restoration.

To update previously reported clinical outcomes, our review had similar results for successful union and the AOFAS post-operative score [16]. Mao et al. reported the AOFAS post-operative score average of 77.13, which is nearly identical to our updated mean score of 77.20, after salvage first MTP arthrodesis [16]. This indicates that roughly no change in alignment, pain, and patient satisfaction has occurred between 2016 and 2023 via the completion of additional studies. Additionally, our rate of successful union ranged from 41.70% to 94.70%, compared to the previously reported range of 67% to 97% [16]. These conditions can be attributed to two reasons: the first being how our review defined successful union and the second being our exclusion criteria. We defined successful union as union without the need for reoperation to achieve adequate fusion across the arthrodesis site. In the study by Gross et al., successful union occurred in 10/12 feet as reported by Mao et al. [3,16]. However, reoperation occurred in 7/12 feet, which is reflected in our 41.70% success rate. As for our exclusion criteria, we did not include any studies that did not use graft for restoration of length. As salvage procedures not requiring bone graft for length restoration are less severe cases, the higher previously reported successful union rate was expected due to their meta-analysis including such studies. Differentiating by subgroups, the foot and ankle autograft subgroup had a success rate of 75% (1 study; n = 12), autograft from a place other than the foot and ankle ranged from 87.50–91.70%, and allograft procedures demonstrated success rates ranging from 90.90–94.70%. While there were not enough studies to perform proper statistical analysis, allograft had the highest success rates qualitatively.

Future trends in MTP arthrodesis should begin implementing allograft into more salvage procedures based on the conclusions from this review. This increase in allograft use would improve the quality of evidence surrounding this graft type for restoration of length, possibly strengthening our review’s results. Additionally, with advancements in bioengineering, allograft is the only graft subgroup studied that can benefit from improved engineering due to being synthetic. Improvements in the engineering and manufacturing of allograft may show great promise in the coming years within MTP salvage arthrodesis.

It is important to acknowledge the limitations of this review. As the certainty of the evidence was listed as “very low” due to the observational aspect of the study, we suggest future randomized control trials to properly determine the relationship between graft type and successful salvage procedure. Additionally, future studies analyzing the cost-effectiveness of various graft types would further help physicians when deciding upon a graft type, as this systematic review was unable to comment on this factor. This systematic review also included a relatively small sample size, especially when the patients were placed into subgroups by graft type. Indeed, some graft types, such as foot and ankle autograft, only had one study, thus limiting the analysis that could be performed on this topic. As there was limited comparison of graft types within studies, a meta-analysis was not performed and so the true relationship between graft types and outcomes remains to be elucidated. However, this study performs the important task of being a foundation for this topic, as well as highlighting areas of future research. More studies are needed that compare the different bone graft types for salvage first MTP arthrodesis in order to allow for meta-analysis in the future.

5. Conclusions

From this systematic review, it appears that the use of allograft may hold promise as an effective treatment option for MTP salvage arthrodesis, with results demonstrating the greatest mean length restoration, mean post-operation AOFAS score, and success rate, in addition to the lowest complication rate as compared to other graft types. Improving the size of the international bone bank could allow for more widespread use of allograft for salvage procedures, regardless of budget. However, this suggestion is based on low-quality evidence with a relatively small sample size, thus, caution is needed with these results. Future high-quality clinical studies investigating allograft MTP salvage procedures would help validate this conclusion. Furthermore, additional comparison studies are needed to determine the superiority or non-inferiority of the different graft types for salvage first MTP arthrodesis.