1 Background

As the prevalence of obesity increases in the United States,¹ there is an ever increasing demand on surgeons to perform total knee arthroplasty (TKA) on high-risk patients.^2–4 A catastrophic complication that hinders the success of TKA in this patient population is the development of a surgical site infection (SSI). Body mass index (BMI) is an easily calculable measurement that can be used to assess the patient's risk of an SSI after TKA.⁵ However, BMI does not accurately describe the patient's unique fat distribution or overall body composition.^6,7

In an effort to study the effect of local fat composition on SSI, previous studies have measured subcutaneous fat (SCF) thickness on plain radiographs. These studies have demonstrated that excessive SCF around the knee, represented by prepatellar soft tissue thickness, is a better predictor of wound complications than BMI.^8–10 However, these studies are limited by the two-dimensional nature of plain radiographs and likely misrepresent true SCF around the knee. For example, gravity can affect fat distribution when standing for lateral radiographs leading to an inaccurate measurement of parapatellar fat on radiographic images. Therefore, advanced imaging [computed tomography (CT) or magnetic resonance imaging (MRI)] may be a more accurate and reliably-measured representation of SCF around the knee.

In the last several years, CT imaging for TKA patients has become more common due to their use in robotic-assisted surgery.^11,12 These CT scans are usually performed within 3 months of surgery for pre-operative robotic synchronization and planning and may accurately capture the patient's local fat environment. By utilizing a novel adaptation of a previously published method of quantifying tissue on advanced imaging,¹³ we can quantify the area of the preoperative periarticular subcutaneous fat of a TKA candidate.

The purpose of this study is to answer the following questions: (1) Can SCF around the knee on axial CT images be reliably measured using a dedicated software with a high inter-rater reliability coefficient? (2) Which axial CT cut demonstrates the most accurate predictor of wound complications and how does it compare with BMI and prepatellar thickness? (3) Is there any association between SCF area measurement and other clinical findings such as readmission, periprosthetic joint infection (PJI), and reoperation? (4) What confounding factors are associated with increased wound complications following primary TKA?

2 Materials and methods

2.2

2.2 CT-guided, 2-D SCF measurement

Two independent reviewers measured the SCF area. SCF was defined as the fat superficial to the deep fascia of the thigh or leg and deep to the dermis. Three different axial CT cuts were chosen: 2 cm above the superior pole of patella (proximal), mid-patella (middle), and at the most prominent portion of tibial tubercle (distal). These areas were chosen because their locations encompass the entire incision used in the surgical exposure for TKA (Fig. 1A–D). Boundaries demarcating SCF were easily visualized and drawn using the OsiriX MD 10.0 system (Pixmeo SARL, Bernex, Switzerland). The Region of Interest (ROI) tool was used to select a pixel in the SCF region and the threshold of approximately 200 was selected to give appropriate standard deviation to calculate the SCF area under the ROI (Fig. 1E–G). Adjustment of ±10 in threshold was manually made in order to correct for any different CT quality and individual makeup of pixels in each CT scan. In order to control for variations in patient overall size, the SCF area was normalized by dividing it by the transverse area of distal femur measured 2 cm above the superior pole of the patella (Fig. 1H). Averages of 3 measurements were also calculated. Herein, we will denote normalized proximal, middle, distal and average subcutaneous fat area measured on axial CT scans as pSFA, mSFA, dSFA, and aSFA, respectively (Fig. 2).

Close

Fig. 1 A-J. Subcutaneous fat (SCF) measurement on representative axial CT images of a patient with BMI of 34 kg/m². A-D. Coronal scanogram of lower extremities demonstrating where proximal, middle, and distal axial CT images were selected. E-G. Proximal, middle, and distal SCF areas were semi-automatically calculated using the OsiriX software. H. The transverse area of distal femur is measured to normalize each SCF. I. Prepatellar fat thickness is measured perpendicularly from dermis to outer surface of patella. J. Patella thickness is measured to normalize prepatellar fat thickness.

Export to PPT

Close

Fig. 2 A–F. Each subcutaneous fat (SCF) measurement was divided into four incremental groups to compare each group's wound complication with each other: A, normalized, proximal SCF area (pSFA), B, normalized, middle SCF area (mSFA), C, normalized, distal SCF area (dSFA), D, normalized, averaged SCF area (aSFA), and E, normalized prepatellar thickness (PFT). F, BMI (in kg/m²) was divided into five groups and wound complication rate of each group was compared.

Export to PPT

On axial CT images, prepatellar fat thickness was measured at the mid-patella and was also normalized by dividing it by patellar thickness (Fig. 1I and J). This measurement is denoted as PFT. PFT and BMI were compared with pSFA, mSFA, dSFA, and aSFA as independent variables to study the effect of local fat tissue and obesity on primary and secondary outcomes. PFT, pSFA, mSFA, dSFA, and aSFA were then divided evenly into quartiles based on their increasing values. Thus, Group 1 represents the smallest amount of subcutaneous fat, and Group 2, 3, and 4 represent moderate, large, and the largest amount of subcutaneous fat, respectively. BMI was divided into five groups according to the commonly known classification system used in clinical practice (<25 underweight or normal, 25–29.9 overweight, 30–34.9 class I obese, 35–39.9 class II obese, ≥40 class III obese).

3 Results

4 Discussion

Obesity measured by BMI has been extensively studied as a risk factor for various complications including wound complications following TKA.^5,14–16 However, Waisbren et al. demonstrated that percentage of body fat is a better predictor of surgical site infection than BMI.¹⁷ As a result, the local fat milieu of the surgical site has been studied in general surgery,^18–20 spine surgery,^6,7 and TKA^8–10 as an important component of wound healing. In TKA, standard radiographs have been used to evaluate the local fat content by measuring prepatellar subcutaneous fat thickness.^8–10 This two-dimensional measurement, however, is prone to error and inaccuracy. With recent advancements in robotic-assisted surgery, CT images of TKA patients are now readily available and as suggested in this study, be utilized for accurate and reliable measurement of a patient's periarticular subcutaneous fat area. By focusing on representative axial CT images at surgical sites, we demonstrated that moderate amounts of subcutaneous fat tissue around the knee are associated with better clinical outcomes, less wound complications, and less emergency room visits compared with deficient or excessive amounts of periarticular subcutaneous fat tissue. Furthermore, we found a gender difference in subcutaneous fat distribution: more women were found to have excessive subcutaneous fat around their knees than men despite having similar BMIs.

This study, however, had a number of limitations. First, our cohort of 150 patients had a high average BMI of 34.28 kg/m² with no patients having a BMI of less than 18.5 kg/m². Thus, our findings are specific to a more obese population. Second, we decided to divide each SCF measurement into quartiles to study the effect of deficient, moderate, and excessive fat content on wound complications. Due to the small sample size with high average BMI, our thresholds used to divide the cohort into quartiles are biased and required further studies with a bigger sample. Third, patient reported outcomes (PROs) are not included in this study because they were insufficiently collected. We do not know the effect of SCF on PROs. Fourth, multiple surgeons were involved in this study and strict standardization of perioperative protocol was not followed. However, using multivariate regression analysis, we did not find any statistically significant differences among surgeons. Fifth, the retrospective nature of this study created inherent selection bias and was limited by the information recorded in the electronic chart. Finally, semi-automated subcutaneous fat area measurements using region-of-interest functionality can have systematic errors depending on the quality of CT scan and user errors based upon pixel selection, axial CT slice selection, and adjustment of standard deviation in ROI. That being said, we made every effort to standardize our measurements and had two observers to measure SCF area independently.

Despite these limitations, we were able to demonstrate highly accurate subcutaneous fat area measurements with excellent interobserver reliability. All the normalized SCF area measurements at 3 different levels had ICC greater than 0.986. However, when we attempted to measure normalized prepatellar fat thickness at the mid-patella level, ICC inter-observer reliability was only moderate (0.741). There may have been inconsistency in selecting CT axial slices that match the “mid-patella” as well as in measuring a perpendicular distance from the patella to the skin. Presence of osteophytes and eburnation may have contributed to the inconsistency as well. Watts et al. measured prepatellar and pre tubercular fat thickness using lateral radiographs with surprisingly high inter-rater reliability (Pearson's coefficient of 0.92 and 0.96, respectively).⁸ On the other hand, inter-rater reliability was not evaluated in Wagner and Yu studies.^9,10 We believe that measurement of the entire SCF area is less prone to technical error and more representative of the local fat milieu than simply measuring fat thickness. We suspect that development of an algorithm to calculate the entire three-dimensional volumetric measurement of SCF would lead to an even more precise, clinically relevant measurement.

Our study is the first study to show that both small and large amounts of subcutaneous fat are associated with wound complications following TKA. All previous studies linked excessive adipose tissue to higher wound complications and infection.^{6–10,18–20} Complications associated with smaller amounts of subcutaneous fat in our study were minor wound complications as reoperation did not occur in this group of patients. Cellulitis and suture abscesses constituted the majority of wound complications within this group. This finding implies that a certain amount of fatty tissue is protective with respect to wound complications. On the other end of the spectrum, excessive subcutaneous tissue in the leg was associated not only with an increase in wound complications but also with major complications requiring reoperation. Compromised immune system function has been theorized as a leading cause of increased infection in patients with abundant subcutaneous tissue as described by both Watts et al. and Wagner et al. Furthermore, a growing body of evidence has shown that malnutrition and surgical site infection are associated.^21–23 Both deficient and excessive subcutaneous fatty tissues should be considered when evaluating the malnutrition of a patient with chronically compromised immunity. Further studies are needed to determine the upper and lower thresholds of subcutaneous fat that result in a higher risk of wound complications.

Surprisingly, our study also demonstrated that the pSFA Group 2 had a significantly lower emergency room visit compared with other groups. The rate of 90-day emergency room visits following primary TKA in our study was comparable to the study by Kelly et al. (15.33 % vs 13.8 %).²⁴ Our study showed an even distribution of orthopedic-related (47.8 %) and medical-related (52.2 %) reasons for ED visits with the majority of the orthopedic-related reasons being post-operative pain. To the authors' knowledge, there is no study demonstrating a certain body composition as having lower complications (such as ED visits) following primary total joint arthroplasty. However, Black et al. found that malnutrition is associated with increased readmission and ED visits following total joint arthroplasty.²⁵ Again, if we consider deficient and excessive subcutaneous fatty tissue around the knee as two sides of malnutrition, our findings are consistent with Black et al.’s findings. Further investigation is needed to confirm this relationship.

Finally, our study demonstrated a gender difference in body composition – females demonstrated a greater localization of excessive fat tissue in their lower extremities when compared to that of males. This is consistent with other studies that have demonstrated that females have up to twice as much thigh subcutaneous fat than males which is speculated to contribute to gender differences in physical performance as well as knee loading.^26–31 Differing body composition, however, did not predict differences in surgical outcomes as our study did not find any gender differences in either wound or any other complication. Further studies are warranted to look at gender differences in wound complications following total knee arthroplasty.

In conclusion, the subcutaneous fat area around the knee can be accurately and reliably measured on CT axial images using dedicated computer software. Our study further demonstrated that there is a gender difference in thigh subcutaneous fat area and that both deficient and excessive fatty tissue at the surgical site of total knee arthroplasty are associated with an increase in wound complications and emergency room visits.

CRediT authorship contribution statement

Young M. Jee: Conceptualization, Resources, Data curation, Writing – review & editing. Mazen Zamzam: Methodology, Investigation, Writing – original draft, preparation. Sazid Hasan: Software, Validation. Muhammad A. Waheed: Formal analysis, All authors have read and agreed to the published. Ehab S. Saleh: Validation, Visualization. Abdullah M. Omari: Conceptualization, Validation, Supervision, Project administration.

Ethics approval and consent to participate

Institutional Review Board approval was obtained to perform a retrospective review at Corewell Health Hospital with reference number 2019-067.

Consent for publication

Not applicable.

Availability of data and methods

All data generated or analyzed during this study are included in this published article.

Funding

No funding was needed for this study.