dGEMRIC (delayed gadolinium-enhanced MRI of cartilage) indicates adaptive capacity of human knee cartilage



Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) is a new imaging technique to estimate joint cartilage glycosaminoglycan content by T1-relaxation time measurements after penetration of the hydrophilic contrast agent Gd-DTPA2-. This study compares dGEMRIC in age-matched healthy volunteers with different levels of physical activity: Group 1 (n = 12): nonexercising individuals; Group 2 (n = 16): individuals with physical exercise averaging twice weekly; Group 3 (n = 9): male elite runners. dGEMRIC was performed 2 hr after an intravenous injection of Gd-DTPA2- at 0.3 mmol/kg body weight. T1 differed significantly between the three different levels of physical exercise. T1 values (mean of medial and lateral femoral cartilage) for Groups 1, 2, and 3 were: 382 ± 33, 424 ± 22 and 476 ± 36, respectively (ms, mean ± SD) (P = 0.0004, 1 vs. 2 and 0.0002, 2 vs. 3). Irrespective of the exercise level, T1 was longer in lateral compared to medial femoral cartilage (P = 0.00005; n = 37). In conclusion, this cross-sectional study indicates that human knee cartilage adapts to exercise by increasing the glycosaminoglycan content. Furthermore, results suggest a compartmental difference within the knee with a higher glycosaminoglycan content in lateral compared to medial femoral cartilage. A higher proportion of extracellular water, i.e., larger distribution volume, may to some extent explain the high T1 in the elite runners. Magn Reson Med 51:286–290, 2004. © 2004 Wiley-Liss, Inc.

It is not known if human cartilage can adapt to mechanical stimulation similar to other skeletal tissues, such as bone and muscle (1–3). An adaptive capacity of human cartilage may open new perspectives with regard to prevention and treatment of osteoarthritis, one of the most common and costly diseases in society (4).

The unique biphasic properties of hyaline cartilage depend on the interaction between the matrix constituents, of which proteoglycans and type II collagen are the most abundant (5). The glycosaminoglycans (GAG) are highly negatively charged sidechains of the proteoglycans. They attract water to the cartilage and create a swelling pressure that is counterbalanced by the rigid collagen network. The chondrocytes, the only cell type in cartilage, maintain the tissue integrity by a balanced synthesis and degradation of matrix molecules.

Animal studies have shown a homogenous cartilage matrix and chondrocyte metabolism at birth (6, 7). Subsequently, during growth the chondrocytes develop different phenotypes by adaptation to local functional requirements within the joint (6, 7). As a result, load-bearing areas in adult animals have an increased GAG content whereas areas subjected to shear stress show different collagen orientation (8–10). In horses, the formation of site differences within the metacarpo-phalangeal joint occurring during the first 5 months was delayed if the animals were kept in boxes (7). In the adult dog, cast immobilization of the stifle joint resulted in a reversible decrease in cartilage GAG content and synthesis (11, 12). With regard to cartilage mechanical properties, cast immobilization for 11 weeks decreased the cartilage stiffness in dogs (13). In vitro, it was shown that enzymatic GAG depletion decreases cartilage indentation stiffness (14). In conclusion, these studies suggest a relationship between GAG content and mechanical stiffness. The load-bearing per se seems necessary for maintaining the cartilage integrity, because joint motion in the absence of weight-bearing causes GAG loss similar to that shown after casting (15).

With respect to cartilage adaptation to exercise, several animal studies have suggested that physical exercise increases the GAG content in weight-bearing cartilage (16–19). Moderate running exercise in dogs increased the proteoglycan content of the weight-bearing medial femoral and patellar cartilage compared with sedentary controls (18). Furthermore, the indentation stiffness of load-bearing cartilage areas in the stifle joint increased after the same exercise regimen (20). The effect of lifelong strenuous exercise in dogs has been studied by Newton et al. (21), who randomized 21 dogs to running 75 min per day, 5 days per week while carrying ∼130% of their body weight, or controls with normal cage activity. No differences were found between the groups with regard to knee (tibial plateau) cartilage thickness, permeability, or stiffness (21).

o far, it has not been possible to demonstrate an adaptive capacity in human cartilage. However, with improvements in the field of MRI, new monitoring techniques evolve. Using a fat-suppressed 3D MRI sequence, Eckstein et al. (22) recently compared knee joint cartilage in physically inactive volunteers and triathletes (nine men, nine women). They found an increased joint surface area in the male triathletes but no differences in cartilage volume or thickness. However, to increase the cartilage volume, the collagen network needs to expand, which is less likely due to the very slow turnover of type II collagen in adult cartilage (23).

Variations in GAG content, as shown in animal exercise studies, are not possible to detect in morphological analysis. Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) is a new noninvasive technique to study the cartilage matrix on the compositional level. In dGEMRIC, a negatively charged contrast medium (Gd-DTPA2-) is injected intravenously. It diffuses, in a dose-dependent manner, into the cartilage in an inverse relationship to cartilage fixed charged density comprised by the GAG (24, 25). A low GAG content yields a high contrast distribution and, hence, a short T1 signal. This method has been evaluated in several in vitro and in vivo studies (24, 26–30). The optimal time for MRI analysis of femoral weight-bearing cartilage is ∼2 hr after injection (29). Using the triple dose (0.3 mmol/kg), dGEMRIC has been shown to be sensitive to small differences in GAG content (29, 31). When the double or single dose was used, T1 measurements in five healthy volunteers that were investigated 2 weeks to 2 months apart were reproducible within 10–15% (25).

The purpose of the present study was to study the relationship between physical activity and cartilage GAG content in humans. We hypothesize that cartilage GAG content is higher in individuals who exercise regularly than in sedentary subjects and that this is reflected in a longer dGEMRIC T1 relaxation time.



Thirty-seven healthy volunteers were categorized according to their level of physical activity during the 2 years preceding the study. No subject had any history of knee injury or knee surgery. Group 1 included nonexercising individuals with a sedentary lifestyle. Group 2 included moderately exercising individuals with regular physical activities, averaging twice a week. The most common activities were jogging and working out at a gym. Group 3 consisted of elite male track and field athletes that were running an average of 90 km per week. Age, body weight, height, and body mass index (BMI) of all subjects are listed in Table 1.

Table 1. Baseline characteristics and T1relaxation times for the individuals in the present study
  Exercise level
(1) sedentary(2) moderate(3) elite
  1. There were no statistically significant differences with regard to age, weight, or BMI between the groups. T1 ratio between groups for both the lateral and the medial compartment is shown in italic (Group 2 vs. group 1, and group 3 vs. group 2).

Number 12 16 9 
Gender male/female 6/6 7/9 9/0 
Age (years, mean ± SD) 25 ± 3 24 ± 2 26 ± 2 
Lenght (cm, mean ± SD) 175 ± 8 174 ± 11 182 ± 6 
Weight (kg, mean ± SD) 70 ± 14 71 ± 16 70 ± 6 
BMI (mean ± SD) 22.7 ± 3.3 22.9 ± 2.5 21.2 ± 1.2 
T1 (ms, mean ± SD) Ratio between groups: laterally (lat) and medially (med)Lateral399 ± 40 432 ± 31(2)/(1) lat1.08P = 0.02486 ± 34(3)/(2) lat 1.13 P < 0.001
Medial364 ± 33 416 ± 24(2)/(1) med 1.14 P < 0.0001465 ± 41(3)/(2) med 1.12 P < 0.001
Ratio lateral/medial1.10, P = 0.0031.04, P = 0.071.05, P = 0.04


The contrast medium Gd-DTPA2- (Magnevist, Schering Ag, Berlin, Germany) is a water-soluble biochemically inert chelate complex that is eliminated through glomerular filtration with a plasma half-life of about 90 min (32). Gd-DTPA2-, at 0.3 mmol/kg body weight, was given slowly (1–2 min) in an antecubital vein with the patient in the supine position to avoid thrombophlebitis at the injection site (29). To optimize the distribution of Gd-DTPA2- into the cartilage, the subjects walked two stories (42 steps) five times up and down at an easy pace, corresponding to ∼7 min of exercise, starting 5 min after injection (29, 31).

MR examinations were performed 2 hr after the intravenous injection of Gd-DTPA2 using a 1.5 T MR imaging system with a dedicated knee coil (Magnetom Vision; Siemens Medical Systems, Erlangen, Germany). Central parts of the medial and lateral femoral condyles were identified using a routine diagnostic series. In the selected central parts of the cartilage, quantitative T1 measurements were made in 3–5 mm sagittal slices, using sets of turbo inversion recovery images with different inversion times (TR = 2000 ms, TE = 15 ms, turbofactor 11, FOV 120 × 120 mm2, matrix = 256 × 256, TI = 50, 100, 200, 400, 800, and 1600 ms). In each set of images, a region of interest (ROI) was drawn in the medial and lateral femoral weight-bearing cartilage. The ROI was placed between the center of the tibial plateau and the rear insertion of the meniscus and included the full thickness of the cartilage (29, 31). T1 was calculated using the mean signal intensity from each ROI as input to a three-parameter fit (33).

The study was approved by the institutional review board and informed written consent was obtained from the subjects.

A t-test and correlation coefficient (Pearson) were used for the statistical evaluation.


T1 in Relationship to the Level of Physical Exercise

T1 values for lateral and medial weight-bearing femoral cartilage in each exercise group are shown in Table 1. There was a significant difference in T1 between the various levels of physical exercise, with the longest T1 values in the elite runners and the shortest values in the sedentary individuals (Fig. 1). T1 was 11% longer in the moderately exercising compared to sedentary individuals (P = 0.0004, mean of lateral and medial compartment) (Fig. 1). The comparison between elite runners and moderately exercising individuals showed a 12% longer T1 in the elite runners (P = 0.0002, mean of lateral and medial compartment) (Fig. 1).

Figure 1.

T1 relaxation times in relationship to the level of physical exercise. Individual T1 values (ms, mean of lateral and medial weight-bearing femoral cartilage) in men (○) and women (▾), respectively, for each level of physical exercise: sedentary (n = 12), moderate (regular physical exercise in average two times per week) (n = 16), elite (running 90 km per week) (n = 9).

In a separate analysis of each compartment, T1 relaxation time was longer in Group 2 than Group 1 both laterally and medially (P = 0.02 and P < 0.0001, respectively) (Table 1). Similarly, T1 was longer both laterally and medially in Group 3 compared with Group 2 (P = 0.0004 and P = 0.001, respectively) (Table 1).

T1 in Relationship to Gender, Age, and BMI

The elite runners were all male athletes, whereas in the other groups there was an equal number of male and female subjects (Table 1). In the sedentary and moderately exercising individuals, T1 values were similar between genders. Mean T1 in medial and lateral femoral cartilage was 399 ± 30 in men and 410 ± 38 in women (ms ± SD) (P = 0.41).

No correlation was found between T1 and age (R = −0.11, P = 0.50, n = 37) or BMI (R = −0.27, P = 0.10, n = 37).

T1 in Medial Compared to Lateral Femoral Cartilage

With regard to the compartmental evaluation, T1 values were longer in lateral compared to medial femoral cartilage (P < 0.0001, all subjects) (Fig. 2, Table 1). The compartmental difference was a general phenomenon regardless of activity level, although not statistically significant for the moderately exercising individuals (P = 0.07) (Fig. 2, Table 1). The ratio between medial and lateral compartments was 1.10, 1.04, and 1.05 for Groups 1, 2, and 3, respectively (Table 1).

Figure 2.

Compartmental differences in T1 relaxation times for each level of physical exercise. T1 values (ms, mean ± SD) in lateral (○) and medial (•) femoral weight-bearing cartilage, respectively, for each level of physical exercise: sedentary (n = 12), moderate (regular physical exercise in average two times per week) (n = 16), elite (running 90 km per week) (n = 9).


This study indicates a relationship between physical exercise and cartilage structure, as shown by a lower contrast distribution in exercising individuals compared to sedentary individuals. A correlation between contrast agent distribution and GAG has previously been established both in vitro and in vivo, suggesting that the low contrast distribution in the exercising subjects corresponds to a high GAG content (24, 26–28, 33). Despite the limitations of a cross-sectional study, this is the first study to provide evidence for an adaptive capacity of human cartilage. The results are in agreement with cartilage explant and animal studies that have shown induced GAG synthesis and content by mechanical loading. In vitro, the response of cyclic compression at frequencies approaching those of normal walking markedly stimulates GAG synthesis (34). In the knee (stifle) joint of dogs, running 4 km/day for 15 weeks, cartilage GAG content of the femoral condyles increased by ∼30% compared to sedentary controls, more on the medial than lateral side (17). In another dog study that used the same exercise regimen, it was shown that the deposition of newly synthesized proteoglycans occurred in the femoral weight-bearing cartilage, whereas no significant change was detected in the opposing tibial plateau (18). Similarly, horses that were exercised 30 min a day, three times per week for a 6-week period increased the amount of newly synthesized proteoglycans in the cartilage of the third carpal bone compared to sedentary controls (19). As a result of the higher GAG content, the mechanical properties of the cartilage change, as shown by increased indentation stiffness in moderately exercising dogs (20). It can be speculated that increased cartilage stiffness by adding GAG is a functional means to withstand higher mechanical demands.

It is not known whether the longer T1 relaxation times found in exercising individuals reflect the physical activity during the years preceding the study, or are influenced by the activity level already during childhood. In comparison, studies of bone mineral density have shown that adolescence is very important with regard to bone mineral acquisition (35).

The optimal type or level of exercise is not known. In dogs that ran excessively (40 km/day for 1 year), the femoral cartilage GAG content as well as indentation stiffness decreased significantly (36, 37). Eventually, alterations may occur in the collagen network, as shown in strenuously exercising horses (38). In contrast, Newton et al. (21) have shown that lifelong running exercise in dogs does not cause degenerative cartilage changes, such as fibrillations, chondrocyte cloning, or loss of Safranin O staining from the matrix.

The elite runners in the present study had a very low and homogenous contrast distribution, indicating that 90 km of running every week was well tolerated by the cartilage in these individuals. In support, most human studies have not found an increased prevalence of osteoarthritis in runners (39–41). However, as in this study, such studies may include a selection bias. Only those that can and have chosen to run during their active life are included. A population-based randomized study of long-distance running is not easily performed, but may reveal that not all individuals can withstand such strenuous exercise.

In all three groups there was a longer T1 in lateral compared to medial femoral cartilage, indicating a higher GAG content laterally. Similarly, Lyyra et al. (42) found a 12% higher indentation stiffness in lateral compared to medial femoral cartilage in humans.

The lack of female elite runners is a potential weakness of this study. However, no gender difference has been observed in previous dGEMRIC studies (29, 31). In accordance, in this study T1 was similar in men and women (Fig. 1). As expected, no age-dependence was found since a homogenous age group was studied.

In the present study, as in other dGEMRIC studies, the contrast medium was administered per kg body weight, assuming a similar distribution of Gd-DTPA2- in all subjects. However, elite runners have less body fat and a larger proportion of body water than “normal” individuals despite similar BMI (43). When our groups of sedentary individuals and elite runners are related to studies on large normal populations and runners comparable to ours, a body fat percentage of 18.5 and 11.5, respectively, can be assumed (44, 45). These fat percentages are consistent, with ∼8% more extracellular water (ECW) in our group of runners compared to the sedentary men (43). Since Gd-DTPA2- almost solely distributes in the ECW, the elite runners may have received a relatively lower dose of the contrast medium per kg ECW (32). The linear relationship between the injected dose of Gd-DTPA2- and cartilage 1/T1-values can be used to calculate T1 from different contrast media concentrations (29). Accordingly, 8% more ECW corresponds to 4.5% lower T1 relaxation time than we report (29). Differences in ECW cannot explain the 25% T1 difference between the elite runners and the sedentary individuals or the difference between sedentary and moderately exercising individuals in this study. However, it implies that the true T1 value of the elite runners is somewhat lower than what we report and further studies are required with regard to the influence of body fat/water content on dGEMRIC.

Adaptive capacity of human cartilage may have implications in the health perspective, especially with regard to osteoarthritis. It can be speculated that a low cartilage GAG content and biosynthetic activity could develop into “chondroporosis,” with increased susceptibility for osteoarthritis. In such a case, exercise may show to be preventive or even therapeutic once the disease is established. Longitudinal effects of training are needed to confirm this hypothesis.

In conclusion, this study indicates that human knee cartilage adapts to exercise by increasing the GAG content. Exercise level is probably a factor to consider in dGEMRIC, for example, in the selection of controls for clinical studies. Furthermore, results suggest a compartmental difference within the knee with a higher GAG content in lateral compared to medial femoral cartilage.


Grant support was provided by Swedish Medical Research Council (K99-73X), Swedish Center for Research in Sports, Medical Faculty of Lund University and the Swedish Rheumatism Association.