Association of Magnetic Resonance Imaging–Guided Management With Reamputation Rates in Diabetic Foot Osteomyelitis

Abstract

Background

Diabetic foot infections constitute 20% of hospital admissions and are the underlying cause of 80% of amputations annually. Staging magnetic resonance imaging (MRI) may decrease reamputation rates in diabetic foot osteomyelitis.

Methods

A retrospective chart review was conducted to analyze the risk of reamputation in patients with diabetic foot osteomyelitis hospitalized between 2005 and 2022. MRI-guided therapy and proximal surgical margin pathologic evidence of osteomyelitis were compared. The primary outcome was the rate of reamputation up to 365 days.

Results

Enrollment consisted of 386 veterans with a diabetic foot infection complicated by osteomyelitis that required initial amputation, of which 110 patients required reamputation. MRI-guided therapy occurred in 89 of these patients. Preoperative MRI and subsequent MRI-guided therapy were associated with a significant decrease in the chance of reamputation for up to a year after the initial amputation as compared with non–MRI-guided therapy (14/89 [15.7%] vs 96/297 [32.3%], respectively; P = .0024, χ² test). A Cox proportional hazard model demonstrated that MRI-guided therapy had a significant association with decreasing the relative risk of reamputation (hazard ratio, 0.47; 95% CI, .26–.83; P = .0098). Initial proximal margin pathologic findings consistent with osteomyelitis were not associated with the risk of reamputation (hazard ratio, 1.25; 95% CI, .81–1.93; P = .31).

Conclusions

These findings support that incorporating preamputation MRI and MRI-guided therapy into the diagnostic and treatment approach for diabetic foot osteomyelitis may reduce the risk for subsequent amputations over 1 year.

Diabetic foot infections (DFIs) constitute 20% of hospital admissions and are the underlying cause of 80% of amputations annually. In the United States, there are an estimated 15 million persons with diabetes, and this number is growing by 6% each year [1]. Diabetes is more prevalent among US veterans when compared with the general population due to age, exposure to environmental factors during service, and mental health conditions, which can contribute to unhealthy behaviors and increased risk of diabetes [1, 2]. DFI can affect patients’ independence, morbidity, and way of life [3]. Early detection of osteomyelitis secondary to DFI is crucial for prompt and appropriate treatment to maximize the chances of cure and minimize complications [4]. In persons with diabetes, foot ulcers are a primary cause of nontraumatic amputations, which are not only costly but potentially deadly [5]. A major amputation in the United States is estimated to cost $150 000, and approximately 50% of patients who undergo a lower extremity amputation will die within 5 years [6, 7].

DFI complicated by osteomyelitis can have significant, life-altering consequences if not promptly diagnosed and appropriately treated. One study examined patients with DFI and found that after 1 year 54% of ulcers did not heal, 15% of patients died, and 17% of patients required a lower extremity amputation [8]. Osteomyelitis is quite prevalent among DFI with ulcers and is estimated to be present in 10% to 15% of moderate infections and in 50% of severe infections [9]. A study examining the implications of osteomyelitis on outcomes of infected wounds revealed that patents with diabetic foot ulcers were more likely to have minor amputations as compared with those who had soft tissue infections (59.4% vs 13.8%) [10]. Given the high prevalence of osteomyelitis in DFI, accurate diagnosis and treatment are needed to prevent further morbidity.

Radiologic tests are commonly used to detect bone involvement. Plain radiographs are typically used as the initial imaging modality for evaluating a new DFI or suspected osteomyelitis, although they have limitations in terms of sensitivity and specificity for confirming or ruling out osteomyelitis [4]. Early findings of osteomyelitis may be subtle, and changes may not be evident until 30% to 50% of bone mineral content has been affected, which typically occurs after 2 to 3 weeks of infection [11]. Guidelines from the International Working Group on the Diabetic Foot recommend plain films as the initial study, with magnetic resonance imaging (MRI) if suspicion persists [5]. Given the poor sensitivity of the plain film, however, there is a potential to miss early osteomyelitis, unless another diagnostic modality, such as probe to bone, is concerning for osteomyelitis. By contrast, MRI is considered the imaging modality of choice for evaluating diabetic foot osteomyelitis given its higher sensitivity [12]. The guidelines for the diagnosis and treatment of diabetes-related foot infections suggest using MRI when the diagnosis of osteomyelitis is equivocal [5]. A meta-analysis reported a pooled sensitivity of 90% and specificity of 79% for MRI in detecting bone infection in the diabetic foot [13]. A meta-analysis published in 2017 found that the sensitivity of MRI was 93% and the specificity was 75% [14]. The specificity is lower for MRI because reactive bone marrow edema can result from noninfectious processes such as Charcot foot [5]. MRI provides detailed images of the soft tissues, bones, and surrounding structures, allowing for the detection of early inflammatory changes and bone marrow abnormalities associated with osteomyelitis [15].

The objective of this study was to evaluate the association of preamputation MRI and subsequent MRI-guided therapy on reamputation rates in diabetic foot infections complicated by osteomyelitis over the 365 days after the initial amputation.

METHODS

Patients

Patients were identified via ICD-10 codes. Specific date and time of the initial amputation and imaging modalities were identified by CPT codes (Supplementary material). Patient comorbidities were collected via Structured Query Language through the SQL Server Management Studio. Manual chart review was conducted to verify patient records. Preoperative MRI was used to define putatively infected bone. Amputation operation reports and postoperative imaging were used to determine if putatively infected bone was completely resected. These data were reviewed by a podiatric surgeon (C.C.C.). She was blinded to the outcome of the patient. Acquisition and analysis of data were approved by the Research and Development Committee at the Veterans Affairs (VA) Western New York Healthcare System.

Patients were enrolled if they were aged ≥18 years and hospitalized for diabetic foot osteomyelitis at the VA Western New York Healthcare System, Syracuse VA Medical Center, or Albany Stratton VA Medical Center between 1 January 2005 and 1 September 2022. Patients were excluded if they had an initial above- or below-the-knee amputation, since these are generally definitive treatment for osteomyelitis due to diabetic foot infection. Additionally, patients who underwent guillotine amputations were excluded, since these procedures are performed in emergency situations with the intention of subsequent definitive amputation at a later stage. Stratification of enrolled veterans is depicted in Figure 1. The Charlson Comorbidity Index was calculated for all patients.

Definition of MRI-Guided Treatment and Proximal Surgical Margins

MRI-guided treatment was defined as (1) the resection of all putatively infected bone as defined by an MRI scan performed within 30 days of surgery or (2) an initial amputation in which all putatively infected bone was not completely resected and the patient therefore received at least 4 additional weeks of culture-driven postoperative antibiotics. Patients were enrolled in the non–MRI-guided group if (1) they did not have preamputation MRI performed or (2) MRI was performed preoperatively but was not used to guide amputation and/or the subsequent antibiotic treatment course. Data were also collected on the pathologic findings of proximal surgical margins. Manual chart review of operative and pathology reports was performed by a podiatric surgeon to ensure appropriate categorization.

Definition of Gram-Positive and Gram-Negative Infection

Gram-positive infection consisted of a culture for methicillin-sensitive Staphylococcus aureus, methicillin-resistant S aureus, Staphylococcus lugdunensis, or β-hemolytic Streptococcus. Gram-negative infection consisted of Enterobacterales, Pseudomonas, or Stenotrophomonas.

Primary Outcome

The primary outcome was time to reamputation for MRI-guided treatment vs non–MRI-guided treatment for up to a year after the initial amputation.

Statistical Analysis

Statistical analyses were conducted in JMP Pro version 18. Categorical variables were compared by χ² tests, ordinal variables with Mann-Whitney U tests, and continuous variables with Student t tests to evaluate differences between MRI-guided and non–MRI-guided therapy groups. Time-to-event data were analyzed by Kaplan-Meier survival curves; group differences between the MRI-guided and nonguided cohorts were assessed by the log-rank test, as were cohorts with and without proximal margin pathology findings indicative of osteomyelitis. To assess potential confounding factors, a Cox proportional hazards model was employed with backward elimination to produce hazard ratios (HRs) with 95% CIs. MRI guidance and pathologic margins were included in the model a priori for backward elimination.

RESULTS

The study enrolled 1522 patients diagnosed with osteomyelitis based on ICD-10 codes, of whom 386 underwent an initial amputation. Among these, 89 patients were assigned to the MRI-guided cohort and 297 to the non–MRI-guided cohort (Figure 1). Nine patients had MRI, but the imaging was not used to guide therapy; therefore, they were included in the non–MRI-guided cohort. The median time of imaging before amputation was 4 days (IQR, 2–12). The majority of patients were male (97.9%) with a mean age of 66.9 years. Baseline characteristics were comparable between the MRI-guided and non–MRI-guided therapy cohorts, with no significant differences in comorbidities, such as myocardial infarction, heart failure, peripheral vascular disease, cerebrovascular accident, dementia, chronic obstructive pulmonary disease, rheumatologic disorders, peptic ulcer disease, liver disease, paraplegia, renal disease, malignant or metastatic cancer, and HIV (Table 1). Both cohorts had a median Charlson Comorbidity Index score of 6 (P = .76). Of the patients who had MRI-guided therapy, 43 (48.3%) had a gram-positive infection, as compared with 94 (31.7%) in the non–MRI-guided therapy group (P = .004). In the MRI group, 19 (21.4%) patients had gram-negative infection vs 68 (22.9%) in the non–MRI-guided group (P = .76).

Significantly fewer patients in the MRI-guided cohort (14/89, 15.7%) underwent reamputation within 1 year of the initial amputation as compared with the non–MRI-guided cohort (96/297, 32.3%; P = .0024). Kaplan-Meier analysis estimated the probability of reamputation over 365 days, showing a significantly lower risk in the MRI-guided cohort vs the non–MRI-guided cohort (P = .0033; Figure 2A).

Although not a primary outcome of this study, pathologic data were available for the proximal margin of resected bone following initial amputation. Among the 386 patients who underwent an initial amputation, 287 had proximal margin pathology not consistent with osteomyelitis, 93 had findings consistent with osteomyelitis, and data were unavailable for 6 patients. The rates of reamputation within 1 year were comparable between patients without proximal margin osteomyelitis (79/287, 27.5%) and those with it (31/93, 33.0%; P = .28). Kaplan-Meier analysis was conducted to estimate the probability of reamputation over 365 days for these cohorts (Figure 2B). The survival probabilities for reamputation were similar between patients with and without proximal margin pathologic findings consistent with osteomyelitis (P = .16).

Potential confounding variables were compared between the MRI-guided and non–MRI-guided therapy cohorts (Table 2). Vascular status and interventions, which could influence outcomes, were similar between the groups. Among all patients, 122 underwent arterial Doppler studies, and 48 underwent revascularization procedures. The rates of revascularization were comparable between the MRI-guided (8/89, 9.0%) and non–MRI-guided (40/297, 13.5%) cohorts (P = .26). Similarly, the frequency of infectious disease consultations did not differ significantly between cohorts (P = .62). Antibiotic durations exceeding 4 weeks were significantly more common in the MRI-guided cohort than the non–MRI-guided cohort (P = .02). However, this was expected, as patients in the MRI-guided cohort with incomplete resection of putatively infected bone received at least 4 additional weeks of postoperative culture-driven antibiotics. Proximal margin pathologic findings consistent with osteomyelitis were significantly more common in the MRI-guided cohort than the non–MRI-guided cohort (P = .003). Additionally, cultures were obtained significantly more often in the MRI-guided cohort, which may have influenced antimicrobial regimen design and efficacy.

A Cox proportional hazards model demonstrated that MRI-guided therapy was associated with a significantly reduced risk of reamputation within 1 year (HR, 0.47; 95% CI, .26–.83; P = .0098) (Table 3). The following were not significant in the model for hazard of reamputation within 365 days: antibiotic duration >4 weeks (HR, 0.82; 95% CI, .48–1.4; P = .48), gram-negative infection (HR, 1.23 95% CI, .8–1.9; P = .35), and gram-positive infection (HR, 1.0; 95% CI, .67–1.5; P = .97). Other variables, including age, Charlson score, and revascularization, were not significant in the final model. Proximal margin pathologic findings, whether consistent with osteomyelitis or not, were included in the model a priori but showed no significant effect on reamputation risk at 1 year. Initial margins consistent with osteomyelitis had an HR of 1.25 (95% CI, .81–1.93; P = .31) for risk of subsequent amputation. These results suggest that proximal margin pathologic findings are not a reliable predictor of the risk for subsequent reamputation.

DISCUSSION

This study demonstrated that at 1 year, 15.7% of patients who had MRI-guided therapy had subsequent amputation, as opposed to 32.3% of patients who had non–MRI-guided therapy (HR, 0.47; 95% CI, .26–.83; P = .0098). Therefore, preamputation MRI and MRI-guided therapy can be considered an approach to improve clinical outcomes in the setting of DFI complicated by osteomyelitis. Based on this study, MRI holds the promise of staging the extent of osteomyelitis. If surgical intervention is to occur, MRI staging can guide the extent of amputation to increase the chance of achieving a surgical cure of the osteomyelitis. If surgery does not resect all the infected bone as defined by MRI, then a subsequent prolonged course of antimicrobial therapy can be used to maximize the chances of cure. Of note, our MRI group was more likely to have a prolonged course of antimicrobials.

MRI is the most accurate of the imaging modalities for diagnosing osteomyelitis and establishing the extent of involvement, with a sensitivity of 0.90 and specificity of 0.79 [13]. Despite MRI having excellent sensitivity, anatomic definition, and avoidance of radiation for the patient, it is often underutilized. Findings from this study demonstrate the potential benefits of incorporating preamputation MRI into the management plan for DFI complicated by osteomyelitis. The utilization of MRI guidance for the initial amputation was associated with a significantly lower rate of reamputation as compared with cases where MRI guidance was not utilized. At 1 year after the initial amputation, MRI-guided therapy was associated with a decreased risk of subsequent amputation at 1 year (HR, 0.47; 95% CI, .26–.83; P = .0098). If residual infected bone remained after surgery as defined by MRI, a ≥4-week antimicrobial course was incorporated into the MRI-guided therapeutic plan; in this circumstance, the optimal antimicrobial duration has not been defined, but a prolonged course is reasonable given the consequences of failure. By contrast, proximal margin pathologic findings consistent with or without osteomyelitis did not have a significant association on decreasing the risk of reamputation at 1 year (HR, 1.23; 95% CI, .81–1.93; P = .31). Improved outcomes of DFI is unequivocally needed; our study supports the use of preamputation MRI and MRI-guided therapy as an important tool to achieve that goal [6].

Plain radiography is often more accessible and less costly than alternative imaging modalities. It can be used as part of the initial evaluation of infection [5]. Yet, use of plain radiography has many limitations. The pooled sensitivity and specificity of plain radiography are 0.54 and 0.68, respectively [13]. Additionally, radiographic changes may not be present until 2 to 3 weeks of infection [5]. Therefore, due to its decreased sensitivity for detecting the presence and extent of osteomyelitis, this modality cannot serve as a substitute for MRI. Computed tomography (CT) can provide more early information on bone involvement than plain films, although it has less soft tissue contrast than MRI, making it less desirable [11]. CT is able to detect elevated periosteum and has better visualization of soft tissue, intramedullary gas, and sinus tracts. It is, however, unable to detect bone marrow edema, making it less able to identify early osteomyelitis [15]. In chronic osteomyelitis, CT had a sensitivity of 0.67 and a specificity of 0.5 [11]. CT may be useful for patients where MRI is not feasible or is contraindicated, but the clinician needs to be aware that CT is imperfect for detecting the presence and extent of osteomyelitis. Single-photon emission CT/CT SPECT/CT hybrid imaging has been compared with bone biopsy for the diagnosis of osteomyelitis in diabetic foot infections. SPECT/CT and MRI have similar accuracy: the sensitivity was 89% for both, and the specificity was 35% and 37%, respectively [16].

Probe to bone is commonly used for diagnosis of osteomyelitis. A study found that the pooled sensitivity and specificity were 0.87, meaning that it can rule in osteomyelitis for a patient at high risk and rule out osteomyelitis for a patient considered low risk [17]. The sensitivity of probe to bone is 0.6 and the specificity is 0.9 [13]; another study reports a sensitivity of 0.87 and a specificity of 0.83 [17]. The probe to bone must be done by a clinician who is experienced in this technique [5]. Furthermore, the probe to bone cannot determine if infection has contiguously spread to other bones. This modality was not evaluated in our study.

Bone biopsy is an accepted reference standard via the International Working Group on the Diabetic Foot [18]. It can, however, have sampling error, leading to a missed osteomyelitis diagnosis. Similarly, the efficacy of proximal surgical margins evaluated by pathology for osteomyelitis is controversial. There has been low interrater reliability among pathologists [19]. In our study, proximal margins without evidence of osteomyelitis were not associated with a significantly decreased risk of amputation.

Peripheral vascular disease is a risk factor for the development of DFI and treatment failure. It can affect antibiotic delivery to the site of infection and wound healing after amputation [20]. Therefore, this represents a potential confounding variable for outcomes. A total of 122 patients in our study were assessed for peripheral vascular disease, and 48 patients underwent a subsequent revascularization procedure, which was similar in the MRI-guided (n = 8, 8.99%) and non–MRI-guided (n = 40, 13.47%) cohorts. Therefore, this intervention did not bias the decreased risk of subsequent amputation in the MRI-guided cohort. Although revascularization was not identified in the Cox proportional hazard analysis as a factor that further decreased the risk of reamputation in combination with MRI-guided therapy, the small sample size and disproportionate number of patients who underwent revascularization in each cohort may have affected this result. Future studies are warranted to determine the role of vascular studies and revascularization in the management of DFI complicated by osteomyelitis. Likewise, antibiotic duration was not identified as a factor in the Cox proportional hazard model that further decreased the risk of reamputation in combination with MRI-guided therapy. However, the MRI-guided cohort was significantly more likely to receive >4 weeks of antimicrobial therapy as compared with the non–MRI-guided cohort, possibly because antibiotic duration is incorporated into the MRI-guided therapy management plan. In the MRI-guided cohort, all patients in which putatively infected bone was not completely resected received at least 4 additional weeks of culture-driven postoperative antibiotics, which is predicated to maximize the chances of cure. In the non–MRI-guided cohort, there was uncertainty if there was residual osteomyelitis after the first amputation. If this was the case and a shorter course of antibiotics was used, it is tempting to speculate that this could be a factor that increased the risk of reamputation. There were 9 patients who had MRI evidence of incomplete resection and received a short course of antimicrobials; they were therefore included in the non–MRI-guided group and 44.4% underwent subsequent amputation.

This study has various limitations. First, the retrospective design inherently introduced limitations, such as documentation bias, the risk for incomplete or missing information, and confounding variables not identified that could affect the results. In addition, the study primarily enrolled older male veterans, which may restrict the generalizability of the findings to a broader population, although non-VA studies have similar demographics [18]. While the findings may not be fully generalizable to the general population, they contribute to the existing body of knowledge and highlight the need for further research in different patient populations and health care settings. Second, treatment failures can occur for a variety of reasons, one of which is an inadequate blood supply for tissue repair and delivery of antibiotics. As discussed, not all patients underwent vascular assessment and revascularization as needed. However, this should not bias our results since a greater proportion of patients in the non–MRI-guided cohort underwent revascularization as compared with the MRI-guided cohort; it is possible that a uniform use of vascular studies and interventions could further improve MRI-guided outcomes. Third, determining the offending pathogens and instituting appropriate treatment is important, especially in the setting of residual osteomyelitis postamputation. This can be challenging when wound cultures may not reflect the pathogens causing the osteomyelitis and when bone cultures obtained during therapy are imperfect. Furthermore, data on anaerobic pathogens were not collected. Therefore, we cannot rule out that the antimicrobial regimens were inequitable between study cohorts. Furthermore, the efficacy of an oral vs parenteral regimen (or combination) on outcome is an area of uncertainty for DFI complicated by osteomyelitis; this potential confounding variable also could not be accounted for in this study. Fourth, MRI-guided therapy was based on MRI data obtained within 30 days of initial surgery. It is possible, in fact likely, that at the time of surgery the extent of osteomyelitis was more extensive than defined by the earlier MRI for some patients. This could result in some cases deemed a surgical cure having residual osteomyelitis, which in turn could result in incorrect management decisions (eg, a short postoperative course of antimicrobials) and increased failure rates. It is predicted that if MRI were obtained as close to the initial surgical intervention as possible, then results could be further improved. Last, we were unable to distinguish whether reamputation was due to failure of the initial treatment course or to reinfection. Addressing these important variables in future studies hold the promise of further improving outcomes.

CONCLUSION

This study supports that incorporating preamputation MRI and MRI-guided therapy into the diagnostic and treatment approach for osteomyelitis associated with DFI may reduce the risk for subsequent amputations over 1 year.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. K. A. M.: manuscript preparation, project design, and statistics; J. X.: data collection and manuscript preparation; A. K. C.: data collection and manuscript preparation; C. C. C.: data collection and manuscript preparation; B. A. W.: data collection and manuscript preparation; A. L. O.: data collection and manuscript preparation; A. G. P.: manuscript preparation; M. D.: data collection and manuscript preparation; J. M. N.: manuscript preparation; A. H.: statistics; T. A. R.: project design and manuscript preparation.

Disclaimer. The funders had no role in the decision to publish or the preparation of this manuscript. The contents do not represent the views of the US Department of Veterans Affairs or the US government.

Financial support. This work was supported by the US Department of Veterans Affairs VA Merit Review (I01 BX004677-01 to T. A. R.).

References

Author notes