Introduction
Knee osteoarthritis (OA) represents one of the leading causes of chronic disability among the elderly population [Guccione et al., 1994] and it is known to be associated with structural alterations in most articular and para- articular tissues [Poole, 2012]. Obesity is one of the most important risk factors for knee OA and a reduced quality of life [Losina et al., 2011], despite being mostly preventable. It has been recently suggested that the obesity-related risk of knee OA incidence and progression may be conveyed not only by biomechanical factors but also by endocrine and inflammatory mechanisms [Issa and Griffin, 2012; Berenbaum et al., 2013; Pogacnik Murillo, 2015]. Adipokines, a group of adipocyte-derived cytokines, are thought to mediate intra-articular inflammation and upregulation of cartilage matrix degradation [Issa and Griffin, 2012; Richter et al., 2015].
The infrapatellar fat pad (IPFP) represents intra-articular (i.e., intra-capsular, extra-synovial) adipose tissue [Hoffa, 1904; Haug, 2014], with its macroscopic anatomy having been previously examined in relation to knee function, pathology, and surgical approaches by Gallagher et al. [2005]. These authors [Gallagher et al., 2005] found the IPFP to be constant in shape, consisting of a central body with medial and lateral extensions. More recently, the IPFP also has been identified as a source of leptin and other proinflammatory cytokines [Gegout et al., 2008; Distel et al., 2009; Gandhi et al., 2011; Klein-Wieringa et al., 2011; Hui et al., 2012]. In knee OA, the IPFP was found to yield high expression levels of enzymes that are involved in fat metabolism [Gandhi et al., 2011], suggesting that it represents a link between obesity and knee OA. The Intensive Diet and Exercise for Arthritis (IDEA) trial [Messier et al., 2009] showed that an 18-month diet-induced weight loss plus exercise resulted in a significant reduction in knee symptoms, improvement in knee function, and better physical health-related quality-of-life scores compared to exercise alone [Messier et al., 2013]. This improvement was associated with a significant reduction in the mechanistic primary outcomes, i.e., knee compressive forces and systemic inflammatory marker levels, specifically interleukin (IL)-6 [Messier et al., 2013].
Few studies thus far have quantitatively analyzed IPFP morphology. In a recent paper, Diepold et al. [2015] reported that men displayed a significantly greater ratio of IPFP volume/body weight than women, similar amounts of inter-muscular fat, and strikingly less subcutaneous fat than women. Studies examining the relationship between IPFP volume and knee pain or radiographic OA status have provided partially contradictory results [Chuckpaiwong et al., 2010; Han et al., 2014; Cai et al., 2015; Cowan et al., 2015; Eckstein et al., 2015; Pan et al., 2015; Steidle-Kloc et al., 2015; Teichtahl et al., 2015]. Although the IPFP has been reported to be preserved under starvation conditions [Ioan-Facsinay and Kloppenburg, 2013], there appears to be a dynamic window, as a longitudinal increase in IPFP volume was reported in mice with high-fat feeding and weight gain [Chang et al., 2011]. However, to date, no study has examined the relationship between IPFP volume and weight change in humans, and no interventional trial has studied the sensitivity of IPFP morphology to exercise and/or weight loss. Therefore, in this secondary analysis of the IDEA trial [Messier et al., 2013], we examined whether an 18-month diet-induced weight loss (with and without exercise) was associated with changes in IPFP morphology in patients with knee pain and mild-to-moderate OA. Further, we determined whether changes in IPFP morphology differed between groups with exercise intervention only (E), diet-induced weight loss intervention only (D), and a combined diet-induced weight loss + exercise intervention (D+E). Finally, we examined whether changes in IPFP volume correlated with changes in body weight and fat, independently of the intervention.
Materials and Methods
Study Population
IDEA was a single blind, single-center, 18-month, randomized, controlled trial [Messier et al., 2009, 2013], which was conducted from July 2006 to June 2011 at Wake Forest University and the Wake Forest School of Medicine, Winston-Salem, NC, USA. The study was approved by the Human Subjects Institutional Review Board of Wake Forest Health Sciences. Informed consent was obtained in writing from all participants (trial registration No. NCT00381290) [Messier et al., 2009, 2013].
The IDEA trial included 454 ambulatory, community-dwelling persons aged ≥55 years with: (1) Kellgren-Lawrence grade [Kellgren and Lawrence, 1957] 2-3 (mild to moderate) radiographic tibiofemoral OA or tibiofemoral plus patellofemoral OA of one or both knees; (2) pain on most days, due to knee OA; (3) a BMI ranging from 27 to 41; and (4) a sedentary lifestyle, i.e., <30 min/week of formal exercise in the past 6 months.
Participants were recruited from the community over a 37-month period (November 2006 to December 2009) [Messier et al., 2009, 2013]. A stratified-block randomization method assigned all eligible persons to 1 of the following 3 intervention arms, stratified by BMI and gender: exercise-only control (E), diet-induced weight loss only (D), and diet-induced weight loss plus exercise (D+E) [Messier et al., 2009, 2013]. The E group was designated as the comparison (control) group because previous work indicated that aerobic walking or resistance training should be part of the standard of care for knee OA patients [Ettinger et al., 1997], and because the original study [Messier et al., 2013] was designed [Messier et al., 2009] to test whether diet alone, or diet in combination with exercise, leads to greater improvements than the standard of care (exercise) alone. Detailed descriptions of the diet-induced weight loss and exercise interventions, the trial design and rationale, and the primary outcomes have been reported elsewhere [Messier et al., 2009, 2013]. In brief, the diet intervention was initially based on partial meal replacements and one daily meal of 500-750 kcal. The exercise intervention consisted of aerobic walking (15 min), strength training (20 min), another aerobic phase (15 min), and a cool-down (10 min), and it was conducted 3 times weekly, for 1 h, over the entire 18-month observation period [Messier et al., 2009, 2013].
Due to budget restrictions, magnetic resonance (MR) images were obtained on a randomly selected subsample of 106 participants at baseline and at the 18-month follow-up [Hunter et al., 2015]. The sample size per group was: E, n = 36; D, n = 35; and D+E, n = 35.
Image Acquisition and Analysis of the IPFP
A 1.5-T (SIGNA HDx; General Electric Medical Systems, Milwaukee, WI, USA) scanner and an extremity coil were used to obtain MR images of the most affected knee [Messier et al., 2013; Hunter et al., 2015]. Segmentation of the IPFP was performed using a sagittal non-fat-suppressed T1-weighted spin echo sequence (time of repetition, 600 ms; time of echo, 11 ms; contiguous slices with a thickness of 4.5 mm; in-plane resolution, 0.625 × 0.625 mm; field of view, 16 cm; and image matrix, 256 × 256 pixels). All segmentations in this study were performed by the first author of this study (A.L.P.M.), who was trained using standardized test data sets [Steidle-Kloc et al., 2016]. Baseline and follow-up images were read as pairs, but the reader (A.L.P.M.) was blinded to the time of image acquisition and the type of intervention during segmentation. Brightness, intensity, contrast, and gray value limits were adjusted manually in each image to warrant optimal contrast between the IPFP and surrounding tissue, with the reader processing all slices (from medial to lateral) that clearly depicted the IPFP. By applying different labels, the reader manually traced the anterior border of the IPFP (the one facing the patellar ligament) and the posterior border (the one facing the knee joint) (Fig. 1a). Clefts (supra- and infrahoffatic recesses) [Roemer et al., 2016] located in the periphery of the IPFP, and the anterior intermeniscal ligament, were excluded from the segmentation. However, small alterations completely surrounded by adipose tissue were included in the volume segmentations. Based on these segmentations, the IPFP volume, the size of the anterior and posterior surface areas, and the mean and maximal depth from the anterior surface to the posterior surface were computed using custom image analysis software (Fig. 1b) [Diepold et al., 2015; Steidle-Kloc et al., 2016], with the intra- and interobserver reliability having been reported [Steidle-Kloc et al., 2016]. For further methodological details, please see the paper by Pogacnik Murillo [2015].
Fig. 1
a Sagittal MR image showing the segmentation of the IPFP anterior surface (red, facing the patellar ligament) and the posterior surface (blue, facing the inferior patellar pole, the distal femur, and the proximal tibia). b Three-dimensional reconstruction of the IPFP (posterior view). The anterior surface area is colored in red and the posterior (femoral and tibial) surface area in blue. MR, magnetic resonance; IPFP, infrapatellar fat pad.
Analysis of Body and Thigh Composition
The BMI of all of the participants was calculated as mass (measured in kg on a standard calibrated scale) divided by body height squared (measured in m). The whole body fat mass was measured at baseline and at the 18-month follow-up using dual-energy X-ray absorptiometry (DXA) in a subsample of the participants (n = 88). DXA scans were obtained with a fan-beam scanner (Delphi A™; Hologi, Waltham, MA, USA) using the manufacturer's recommendations for patient positioning, scan protocols, and scan analysis, with the reliability (coefficients of variation of repeated measurements) being reported as 1.2% for whole-body fat mass [Messier et al., 2009]. Further, baseline and 18-month follow-up computed tomography (CT) scans of the thigh were obtained in a subsample (n = 82), and measurements of subcutaneous thigh fat, intramuscular thigh fat, and total thigh fat were obtained [Messier et al., 2009].
Statistical Analysis
Baseline anthropometric measures and quantitative measures of the IPFP are reported as means and SD. Within- and between-group comparisons of longitudinal changes in IPFP morphology were evaluated using ANCOVA, adjusting for baseline values of the outcome, baseline BMI, and sex, but not adjusting for multiple between-group comparisons in this post hoc exploratory analysis. IPFP volume was considered the primary analytic focus of the current analysis, and IPFP anterior surface area, posterior surface area, mean and maximal depths were exploratory measures. Correlations between IPFP change and BMI change, fat mass change (DXA) or thigh composition change (CT) were studied across the entire sample, independently of the randomized controlled trial design and the specific type of intervention, using Spearman correlations. The level of significance for all comparisons was set at 0.05. All analyses were performed using SAS v9.4 (SAS Institute, Cary, NC, USA).
Results
There were no significant differences in age, sex, BMI, or IPFP morphology measures between the 3 intervention groups at baseline (Table 1). There was a significant (p < 0.01) reduction in IPFP volume in each of the 3 intervention groups over the 18-month observation period (Table 2; Fig. 2) (E: -2.1%, 95% CI -0.8 to -4.8%; D: -4.0%, 95% CI -1.9 to -5.7%; and D+E: -5.2%, 95% CI -3.5 to -7.5%). As a reference, the mean percent body weight loss from baseline amounted to -1.0% in the E group, -10.5% in the D group, and -13.0% in the D+E group. Changes in kilograms and their 95% CI are shown in Table 2. There was a significant (p < 0.01) within-group decrease in IPFP posterior surface area for each intervention group, and a significant (p < 0.001) within-group decrease in IPFP anterior surface area in the D+E group; the change in IPFP volume therefore appeared to be driven more by a change in surface areas than by a change in IPFP depth (Table 2). Between-group comparisons revealed a significantly greater reduction in IPFP volume and IPFP posterior surface area in the D+E group than in the E (control) group (p < 0.05); other between-group differences in the longitudinal changes in IPFP morphology did not reach statistical significance (Tables 2, 3).
Table 2
Changes in BMI, body weight, and IPFP morphology in the E, D, and D+E intervention groups between baseline and the 18-month follow-up
Table 3
Pairwise comparisons between intervention groups for mean changes in IPFP morphology between baseline and the 18-month follow-up, adjusted for baseline values of the outcome, baseline BMI, and sex
Fig. 2
Mean change in IPFP volume between baseline and the 18-month follow-up in the E, D, and D+E groups, adjusted for baseline values of the outcome, baseline BMI, and sex. Error bars represent 95% CI. IPFP, infrapatellar fat pad; E, exercise intervention group; D, diet intervention group; D+E, diet-plus-exercise intervention group.
Across the 3 intervention groups, there was a significant (p < 0.01) correlation between IPFP volume change and weight loss (r = 0.40), BMI change (r = 0.39), total body fat mass change (r = 0.44, n = 88; DXA), subcutaneous thigh fat change (r = 0.32, n = 82; CT), and intermuscular thigh fat change (r = 0.29, n = 82; CT) (Table 4). Across all intervention types, each percent of weight loss was related to a 0.27% reduction in IPFP volume.
Discussion
In this first study investigating the effect of diet-induced weight loss, with and without exercise, on IPFP morphology, we observed a significant reduction of IPFP size in all intervention groups. Further, the findings suggest that the combination of D+E was more effective in reducing IPFP volume than E alone and that, independently of the type of intervention, the reduction in IPFP volume was significantly correlated with loss of body weight and fat mass.
A limitation of the current analysis is that only a subsample of the IDEA cohort was examined [Hunter et al., 2015]; however, the subsample was reasonably representative of the larger sample and sufficient to demonstrate responsiveness of the IPFP to diet and exercise intervention. A strength of the current study is its randomized controlled design, which allowed analysis of the effects of diet-induced weight loss, with and without exercise, on IPFP volume. Further, randomization of the subsample to the different intervention groups appeared to be without bias, as anthropometric and IPFP morphometric baseline values were similar between the intervention groups. Another limitation of the current study is that a standard no-treatment reference group (with neither exercise nor diet intervention) was not available because the intent of the original study was to test whether diet, in combination with exercise or not, was more effective in reducing clinical symptoms than exercise, which is part of the standard of care for people with knee OA, and that no additional reference group with or without a known OA status in other joints, e.g., those of an upper limb, was available.
Interestingly, an exercise intervention alone resulted in a small, but statistically significant, within-group reduction in IPFP volume despite a minimal and statistically nonsignificant reduction in body weight. Diet in combination with exercise was more effective in reducing the IPFP volume despite reductions in body weight being similar in the D and D+E groups. Although the mechanism behind exercise-related effects on the IPFP is unclear, this could potentially be due to modifications in inflammation and edema within the IPFP. These findings indicate that exercise may be effective in modifying IPFP morphology, at least partly independently of weight loss. Further interventional studies should examine whether specific types of exercises are more effective than others in reducing the IPFP volume, whether reduction of the IPFP volume is associated with less inflammatory activity and release of adipokines, and whether modification of the IPFP (including surgical reduction) is related to joint health.
Previous studies examining the relationship between IPFP size and knee OA have revealed partly contradictory results; Chuckpaiwong et al. [2010] found no differences in IPFP volume between patients with knee OA and healthy controls in a small (and likely underpowered) sample. A smaller IPFP size was found to be associated to cartilage defects compared to the control group, but only in women [Duran et al., 2015]. Furthermore, differences in IPFP volume between patients with cartilage defects and healthy controls disappeared after adjustment for age and BMI [Duran et al., 2015]. The IPFP volume in joints with patellofemoral OA was reported to be greater than in those without patellofemoral OA [Cowan et al., 2015]. A greater IPFP maximal sagittal cross-sectional area, however, appeared to be beneficial in terms of structural changes and knee pain, but these observations were limited to a single slice [Han et al., 2014; Pan et al., 2015; Teichtahl et al., 2015]. In this context, a strength of the present study is that the complete IPFP volume was studied (not only a single slice, a 2-dimensional representation of its tissue) [Han et al., 2014; Pan et al., 2015]. This type of volumetric analysis is reproducible within and between observers [Steidle-Kloc et al., 2016]. Also, volumetric analysis permitted us to examine the responsiveness of different measures of IPFP morphology, indicating that a reduction of the (posterior) surface area may be more sensitive to exercise and diet-induced weight loss than IPFP depth.
The IPFP is a local source of proinflammatory mediators, including a variety of cytokines and adipokines, such as leptin and adiponectin [Gegout et al., 2008; Distel et al., 2009; Gandhi et al., 2011; Klein-Wieringa et al., 2011; Hui et al., 2012]. The inflammatory activity of the IPFP affects synovial inflammation [Eymard et al., 2014], while high levels of inflammatory mediators are associated with cartilage degradation [Ding et al., 2008; Bao et al., 2010; Kang et al., 2010; Stannus et al., 2015]. The close spatial relation of the IPFP to the synovium-lined knee joint cavity suggests that it may more directly influence inflammatory mediator levels involved in the pathogenesis of knee OA than systemic factors [Clockaerts et al., 2010]. Findings that the IPFP of patients with knee OA display high expression levels of enzymes involved in fat metabolism led to the conclusion that the IPFP may be responsive to the overall nutritional state; a reduction of the IPFP volume may thus be associated with a reduction of the IPFP inflammatory activity and the secretion of inflammatory mediators into synovial fluid, and potentially a reduction in cytokines involved in cartilage degradation, inflammation, and joint pain. The role of the IPFP in inflammatory processes of the knee, however, is still controversial [Berenbaum et al., 2013]. However, conditioned adipose tissue medium of the IPFP of OA patients was found to block catabolic processes induced by IL-1β in articular cartilage [Bastiaansen-Jenniskens et al., 2012]. Further, local production of inflammatory mediators in the knee was attributed not only to the IPFP but also to osteophytes and the synovium [Presle et al., 2006; Gegout et al., 2008]. No intra-articular inflammatory or other cytokine markers were examined in our current study; hence, follow-up studies, preferably with intra-synovial aspiration, will need to address whether a direct relationship exists between diet- and exercise-induced reduction in IPFP volume and inflammation. Messier et al. [2013] reported, among overweight and obese adults with knee OA, knee compressive forces to be lower after 18 months of diet intervention compared to exercise, concentrations of IL-6 to be lower with diet-and-exercise intervention compared to exercise, and knee pain levels to be lower (and knee function better) with diet-and-exercise intervention compared to both diet and exercise interventions. The D+E group also had better physical health-related quality of life scores than the E group. In extension to the original study [Messier et al., 2013], it was further shown that: (1) weight loss led to a reduction in bone mineral density in the femoral neck and hip, irrespectively of whether the diet was combined with an exercise intervention or not [Beavers et al., 2014]; (2) a high amount of subcutaneous fat had significant associations with knee joint forces similar to those of abdominal fat despite its much smaller volume, and thus this could be an important therapeutic target for patients with knee OA [Messier et al., 2014]; (3) intentional total body fat mass and abdominal fat volume reductions were independently associated with significant reductions in inflammatory serum markers [Beavers et al., 2015]; and (4) despite the potent effects of weight loss on symptoms and mechanistic outcomes, there was no statistically significant difference between the 3 active interventions on radiographic joint space width change, quantitative cartilage thickness loss, or semiquantitative measures of Hoffa-synovitis and joint effusion [Hunter et al., 2015]. However, whether the symptomatic improvement and increased physical health-related quality-of-life scores observed with the diet-and-exercise intervention [Messier et al., 2013] is mediated by longitudinal changes in IPFP volume, or whether the latter should rather be considered an “innocent bystander” in this process, is beyond the focus and design of this report and may be elucidated in further studies.
In summary, the IPFP, a potential link between obesity and knee OA, was sensitive to intervention by diet-induced weight loss and/or exercise in overweight or obese patients with symptomatic knee OA. The reduction in IPFP volume correlated with the individual level of weight and body fat change. Diet in combination with exercise appeared to be more effective in reducing IPFP volume than exercise alone.
Acknowledgment
Support for this study was provided by grants from the National Institutes of Health (R01397 AR052528-01, P30 AG21332, and M01-RR00211) and General Nutrition Centers, Inc. Support for the image analysis of the IPFP was received from the Paracelsus Medical University Forschungsförderungsfond (PMU FFF; R-14/ 0/061/WIT).
The statistical analysis and writing of this article was independent from and not contingent upon approval from the study sponsors. The funding sources were not involved in the study design or in the statistical analysis of the data.
We would like to thank Anja Ruhdorfer, MD, PhD, Eva Steidle, PhD, and Torben Dannhauer, PhD, at the Institute of Anatomy at Paracelsus Medical University for their support and critical discussion of IPFP segmentation in this study.
Author Contributions
• Substantial contribution to the study conception and design: all authors
• Substantial contribution to the acquisition of data (IPFP): A.L. Pogacnik Murillo, F. Eckstein, and W. Wirth
• Statistical analysis: D. Beavers
• Analysis and interpretation of data: all authors
• Writing of the first manuscript draft: A.L. Pogacnik Murillo and F. Eckstein
• Critical manuscript revision and approval of the final manuscript: all authors
Disclosure Statement
F. Eckstein is CEO of Chondrometrics GmbH, a company providing MR image analysis services to academic researchers and to industry. He has provided consulting services to MerckSerono, Abbvie, and Servier, has prepared educational sessions for Medtronic, and has received research support from Pfizer, Eli Lilly, Merck Serono, GlaxoSmithKline, Centocor R&D, Wyeth, Novartis, Stryker, Abbvie, Kolon, Synarc, Ampio, Orthotrophix, and Samumed.
W. Wirth has part time employment with Chondrometrics GmbH and is coowner of Chondrometrics GmbH.
The other authors declare no competing interests.
Get Permission
PubMed ID
