CORR - 2026-06-15 - Journal Article
Can We Validate the Musculoskeletal Infection Society Criteria in Patients Suspected of Infection After Transfemoral Osseointegration Surgery?
Bozzay AB, Rivera JA, Rabin SE, Souza JM, Potter BK, Forsberg JA
Topics
Key Takeaway
The 2018 MSIS serum criteria failed to discriminate infection from no-infection in transfemoral osseointegration patients (MSIS total score likelihood ratio χ² = 1.6, p = 0.21), while a novel 4-marker composite score achieved AUC = 0.84 with sensitivity 0.88 at a cutoff ≥ 2.
Summary Depth
Choose how much analysis to show on this article page.
Summary
This study asked whether the 2018 MSIS/ICM serum criteria (CRP, ESR, WBC, PMN%) can diagnose peri-implant infection after transfemoral osseointegration, using secondary analysis of 55 patients (18 infected, 37 uninfected) from two Walter Reed FDA IDE studies. None of the individual MSIS serum markers differed significantly between groups except PMN% (66.9% vs. 58.7%, p=0.02), and the MSIS total score was not associated with infection on logistic regression (p=0.21). A novel ROC-derived composite score (1 point per elevated marker, cutoff ≥2) yielded AUC 0.84, sensitivity 0.88, and specificity 0.62 using CDC infection definitions as the gold standard.
Key Limitation
The novel composite scoring system was derived and tested on the same 55-patient dataset without external validation, making its AUC, sensitivity, and specificity estimates highly susceptible to overfitting.
Original Abstract
BACKGROUND
Major complications that may result in the removal of osseointegrated implants in patients with limb loss include peri-implant fracture, severe bone resorption, and infection. It remains challenging to diagnose and manage peri-implant infection in patients who have undergone osseointegration surgery. The Musculoskeletal Infection Society (MSIS) developed criteria to identify periprosthetic joint infection (PJI) after total joint arthroplasty during the 2018 MSIS/International Consensus Meeting; however, there is no such validated system for patients who have received osseointegrated implants, and some of the measures on the MSIS/International Consensus Meeting criteria are not applicable to osseointegrated implants. Nevertheless, establishing an adapted version of the MSIS criteria, or a comparable standardized system, would be valuable for enabling consistent diagnosis of osseointegration-associated peri-implant infections.
QUESTIONS/PURPOSES
(1) Is there a difference in the serum parameters of the 2018 MSIS criteria between a cohort of patients without infection and a cohort with an infection? (2) Can the serum parameters of the 2018 MSIS criteria be validated in a population that has undergone osseointegration? (3) Can we develop a new, valid set of criteria to gauge the likelihood of infection from serum parameters for those who have undergone osseointegration?
METHODS
We performed a secondary (retrospective) analysis using data from the Transfemoral Amputation Osseointegration Study, an FDA investigational device exemption study. We supplemented these data with additional participants from a separate, subsequent longitudinal study who had undergone transfemoral amputations. Both studies were conducted to determine the efficacy of an osseointegrated device in the context of patient-reported outcome measures and adverse events like infections and implant removals. The participants were drawn from both studies, which were performed at Walter Reed National Military Medical Center. Eligible patients were adults who underwent transfemoral osseointegration and who had available preoperative and/or postoperative laboratory data relevant to infection assessment. We excluded individuals with incomplete clinical records that precluded the determination of infection status. We collected serum parameters corresponding to the 2018 MSIS criteria from patients from both studies within 30 days of infection diagnosis, excluding joint aspirate data, as no joints are involved in transfemoral osseointegration. We compared the serum parameters (C-reactive protein [CRP], erythrocyte sedimentation rate [ESR], white blood cell count [WBC], and percentage of polymorphonuclear [PMN] cells) from the 2018 MSIS criteria in the 33% (18 of 55) of patients with infections to the 67% (37 of 55) of patients with no infections. A total of 93% (51 of 55) of patients were male, and the study population had a mean ± SD age of 39 ± 11 years. Most participants sustained traumatic amputations due to combat, whereas the remainder resulted from oncologic or infectious etiologies. All variables were assessed before undergoing any osseointegration surgery as part of routine care (preoperative baseline laboratory values) and during the occurrence of any suspected infection, and the study did not depend on longitudinal trajectories or repeated measures. We used the CDC definition of infection, which classified a superficial infection as an infection limited to the skin and a deep infection as an infection beneath the incision in the muscle and surrounding tissues. We used t-tests to compare serum parameters from the 2018 MSIS variables between the no infection cohort and the cohort that had an episode of an infection. To determine whether the original serum parameters from the 2018 MSIS criteria could be applied to a population undergoing osseointegration surgery, we performed a logistic regression using a generalized linear model with a binomial link. The dependent variable was the presence of any infection (superficial or deep) and the predictor was the serum parameters from the 2018 MSIS total score. Finally, we developed a novel list of criteria and tested them against this dataset to evaluate the sensitivity, specificity, and area under the receiver operating characteristic curve using the CDC definition of infection as the gold standard for the presence or absence of infection. Patient outcomes were followed for a minimum of 2 years after osseointegration surgery (Stage II).
RESULTS
We found no difference between the no infection cohort and infection cohort values for the CRP (2.7 ± 4.0 mg/L versus 6.1 ± 8.4 mg/L; p = 0.14). Likewise, the ESR did not differ between the no infection and infection cohorts (20.0 ± 24.2 mm/hour versus 33.8 ± 27.8 mm/hour; p = 0.11). The WBC also did not differ between the no infection and infection cohorts (4.4 ± 2.4 ×103 cells/µL versus 9.3 ± 3.9 ×103 cells/µL; p = 0.08). However, the serum percentage of PMN was higher in the group with infections compared to the group with no infections (66.9% ± 12.2% versus 58.7% ± 10.7%; p = 0.02). We were unable to validate the serum parameters from the 2018 MSIS criteria in this osseointegration surgery population. Our generalized linear model showed that the MSIS total score was not associated with the presence of infection. Furthermore, the overall model was not different from the null model (likelihood ratio χ2 = 1.6; p = 0.21), suggesting that the serum parameters from the 2018 MSIS criteria did not discriminate between the participants without an infection and those who developed an infection. Finally, we developed a novel list of criteria to assess the presence or absence of infection in patients who had transfemoral osseointegration by assigning 1 point for each laboratory marker (CRP, ESR, WBC, and PMN) above its respective receiver operating characteristic (ROC)-derived threshold (range 0 to 4). The composite score showed excellent discrimination of infections (area under the curve [AUC] = 0.84). The optimal cutoff was ≥ 2, which produced a sensitivity of 0.88 and a specificity of 0.62.
CONCLUSION
The serum parameters of the 2018 MSIS/International Consensus Meeting definition of infection did not differ between patients with and without infection after transfemoral osseointegration surgery, but a new approach that we developed appeared sensitive but not specific for this problem in a secondary analysis of patients from prior research trials. Based on these findings, surgeons should emphasize direct clinical examination, the characteristics of soft tissue changes, and patient-reported symptoms as there is no satisfactory approach using laboratory parameters alone.
LEVEL OF EVIDENCE
Level IV, diagnostic study.