Use of positron emission tomography with methyl-11C-choline and 2-18F-fluoro-2-deoxy-D-glucose in comparison with magnetic resonance imaging for the assessment of inflammatory proliferation of synovium




To compare positron emission tomography (PET) and magnetic resonance imaging (MRI) in the evaluation of inflammatory proliferation of synovium.


Ten patients (mean ± SD age 36 ± 13 years) with inflammatory joint disease and with clinical signs of inflammation of the joint were studied. A new tracer for cellular proliferation, methyl-11C-choline (11C-choline), and a widely used tracer for the detection of inflammation and cancer, 2-18F-fluoro-2-deoxy-D-glucose (18F-FDG), were applied for PET imaging, and the results were compared with the findings from gadolinium diethylenetriaminepentaacetic acid–enhanced MR images. The uptake of 11C-choline and 18F-FDG in the inflamed synovium was measured and expressed as the standardized uptake value (SUV) and the kinetic influx constant (Ki) obtained from graphic analysis, and these values were compared with quantitative values on MRI. Synovial volumes were measured on the coronal contrast-enhanced T1-weighted MR images using the standard software of the MR imager.


All patients showed high accumulation of both 11C-choline and 18F-FDG at the site of arthritic changes, where quantification of the tracer uptake was performed. The SUV of 11C-choline was 1.5 ± 0.9 gm/ml (mean ± SD; n = 10) and the SUV of 18F-FDG was 1.9 ± 0.9 gm/ml (n = 10) (P = 0.017). The Ki of 11C-choline (mean ± SD 0.048 ± 0.042 minute−1) was 8-fold higher than the Ki of 18F-FDG (0.006 ± 0.003 minute−1) (P = 0.009). Both the uptake of 11C-choline and the uptake of 18F-FDG correlated highly with the volume of synovium; the highest correlation was observed with the Ki of 11C-choline (r = 0.954, P < 0.0001).


In the use of PET scans,11C-choline can be regarded as a promising tracer for quantitative imaging of proliferative arthritis changes. Nevertheless, subsequent prospective studies with larger numbers of patients are necessary to further characterize the relationship between the findings on PET imaging and the clinical and functional measures of inflammation.

In the clinical assessment of patients with inflammatory joint disease, the quantitative evaluation of disease activity is still difficult. No single test is reliable in establishing the extent of disease activity. Clinical evaluation of the patient is the primary method that has been used to judge disease activity. The clear signs of inflammation in the peripheral joint are quite easy to detect, but quantification of the degree of inflammation is not easy. Several factors have been shown to predict the scope of erosions, but a modern quantitative approach is lacking.

Conventional radiography is an insensitive method for the diagnosis of early rheumatoid arthritis (RA). In studies using scintigraphy, new erosions have been especially prone to develop in those joints with persistently high scintigraphic activity (1). Estimation of clinical features, such as joint tenderness and swelling, and laboratory measures, such as the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) level, for the characterization of disease activity are either inexact or weak predictors of the clinical course of RA in an individual patient. Bone erosions and synovitis are visualized with high sensitivity by magnetic resonance imaging (MRI) (2). The best method to detect synovitis using MRI utilizes the intravenous paramagnetic gadolinium chelates such as gadolinium diethylenetriaminepentaacetic acid (Gd-DTPA) (3). The uptake of Gd-DTPA is a time-related phenomenon that is dependent on tissue perfusion and microvascular permeability. Since the inflamed synovium is characterized by hypervascularity and increased vascular permeability, the Gd-DTPA accumulates in the extracellular space, which thus identifies the sites of inflammation. Accordingly, gadolinium-enhanced dynamic MRI allows estimation of synovial volume and the rate of enhancement (4).

Positron emission tomography (PET) provides a quantitative method to study the metabolic activity of the target tissue in vivo. The present study was conducted to evaluate the usefulness of PET imaging in the assessment of local inflammatory proliferation of synovium. The value of 2 different PET tracers, i.e., methyl-11C-choline (11C-choline), which is a new marker for cellular proliferation (5), and 2-18F-fluoro-2-deoxy-D-glucose (18F-FDG), which has recently been shown to delineate infectious and inflammatory foci with high sensitivity (6, 7), was investigated, and the results were compared with the findings of gadolinium-enhanced MRI. Based on the knowledge 1) that choline is a precursor for the biosynthesis of phospholipids (in particular, phosphatidylcholine), which is the essential component of all eukaryotic cell membranes (8), and 2) that the proliferative changes occurring in arthritic synovium enhance membrane synthesis which thus increases the demand for phospholipids, we hypothesized that 11C-choline would be a good candidate for in vivo imaging of such processes in arthritic joints. Development of new imaging approaches for the measurement of disease activity, for the prediction of progressive joint destruction, and for the monitoring of the efficacy of treatment is highly valuable.



The inclusion criterion was the presence of clinically active inflammation (swollen and tender joints) in at least one large joint in a patient with inflammatory joint disease. Ten patients with clinically active synovitis (in the knee joint of 9 patients and the ankle of 1 patient) were recruited for this study (6 female, 4 male, mean ± SD age 36 ± 13 years). The mean disease duration was 9 years (range 0.5–23 years). Two patients were diagnosed as having RA, 6 patients had unspecified oligoarthritis, 1 patient had ankylosing spondylitis (AS), and 1 patient had undifferentiated spondylarthropathy (uSpA). Both RA patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for their diagnosis (9), the patient with AS fulfilled the modified New York criteria for AS, and the patient with uSpA fulfilled the European spondylarthropathy study group preliminary criteria for the classification of spondylarthropathy (10). Seven patients were HLA–B27 positive and 1 was positive for rheumatoid factor. The clinical characteristics of the patients are presented in Table 1.

Table 1. Clinical characteristics of the study patients*
Patient/diagnosis/sexAge, yearsDuration of disease, yearsESR, mm/hourCRP, mg/literHLA–B27 statusGlucose, mmoles/literRF status
  • *

    ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; RF = rheumatoid factor; ND = not determined; AS = ankylosing spondylitis; RA = rheumatoid arthritis; uSpA = undifferentiated spondylarthropathy.

  • Average arterial plasma glucose concentration during 2-18F-fluoro-2-deoxy-D-glucose positron emission tomography.

1/unspecified oligoarthritis/F282ND22ND4.9
2/unspecified oligoarthritis/M2585238+5.8
3/unspecified oligoarthritis/F30423<10+4.7
4/unspecified oligoarthritis/F48281+5.2
9/unspecified oligoarthritis/F200.5141+5.2
10/unspecified oligoarthritis/F3923173+5.3

Study design.

All patients underwent PET imaging with11C-choline and 18F-FDG, and with gadolinium-enhanced MRI. The 11C-choline and 18F-FDG PET imaging were performed on the same day and MRI was carried out within 1 week after these radionuclide studies. At least 120 minutes separated the injections of the 2 PET tracers, with 11C-choline always preceding 18F-FDG. For both PET studies, the subjects fasted for at least 6 hours, or overnight. Two catheters were inserted, in an antecubital vein for the injection of PET tracers, and in the opposite radial artery for blood sampling. The research protocol and the consent form were reviewed and approved by the ethics committee of the local hospital district. All patients gave their written informed consent before their participation in the study.

Synthesis of tracers.

The 18F-FDG was synthesized with a computer-controlled apparatus in accordance with a modified method described by Hamacher et al (11). The 11C-choline was prepared with an automated synthesis procedure, as described previously (12).

Dynamic PET imaging.

The PET investigations were performed with a GE Advance PET scanner (General Electric Medical Systems, Milwaukee, WI) operated in 2-dimensional (2-D) mode (13). The imaging field of view (FOV) in axial length is 152 mm. The patients were placed supine on a scanner couch, and the position of the inflamed joint in the center of the FOV was secured by physical examination and review of previous radiographs, if necessary.

All PET studies started with a 9-minute transmission scan for the correction of photon attenuation. Following transmission scanning, the tracer was injected intravenously as a 30-second bolus. The median (±SD) dose of 11C-choline was 430 ± 4.98 MBq (range 424–440 MBq) and the median dose of 18F-FDG was 367 ± 6.45 MBq (range 358–377 MBq). The dynamic imaging started immediately after injection. The acquisition times were as follows: 2 × 15 seconds, 3 × 30 seconds, 3 × 60 seconds, 5 × 180 seconds, and 4 × 300 seconds for a total duration of 40 minutes for PET imaging with 11C-choline, and 4 × 30 seconds, 3 × 60 seconds, 5 × 180 seconds, 4 × 300 seconds, and 2 × 600 seconds for a total examination time of 60 minutes using 18F-FDG. All image acquisition data were corrected for dead time, decay, and measured photon attenuation, and were reconstructed with an ordered subsets expectation maximization algorithm. The final in-plane spatial resolution of the reconstructed images was 5 mm and the axial resolution was 6 mm.

Arterial blood samples were collected at frequent intervals and stored in heparinized tubes for measurement of the input function during both types of PET imaging. The total radioactivity in the plasma was measured with an automatic gamma counter (1480 Wizard; Wallac, Turku, Finland). In addition, arterial blood samples were collected at 1, 5, 10, 15, 20, and 25 minutes after injection of 11C-choline for radionuclide–high-performance liquid chromatography (radio-HPLC) analysis of the metabolites (14). The arterial plasma glucose concentration in the beginning and end of the 18F-FDG scan was determined using the glucose oxidase method (Analox GM7 Analyzer; Analox Instruments, London, UK). The mean of these individual values, expressed in mmoles/liter, was used for further calculations, i.e., the rate of glucose uptake.

MR imaging.

The MR images were obtained using 1.5T Siemens Symphony scanners (Erlangen, Germany). The imaging protocol was as follows. The joints were examined with sagittal T2– and proton density–weighted Turbo Flash sequences, which had a repetition time (TR) of 3,140 msec and echo times (TE) of 83 msec and 17 msec, respectively. The slice thickness in this sequence was 4 mm and FOV was 150 × 150 mm2. Coronal T1-weighted images were obtained before and after contrast enhancement. This spin echo T1-weighted coronal sequence had the following imaging parameters: TR of 660 msec, TE of 16 msec, slice thickness of 3 mm, and FOV of 160 × 160 mm2. The coronal contrast-enhanced T1-weighted images were used when the volume of enhancing synovial membrane was measured with the standard postprocessing software of the imager. These T1-weighted images were transferred to the workstation for the synovial volume measurement. Joint effusion on these images appears gray, while the inflamed synovium appears bright, which allows clear differentiation between effusion and synovium.

The set of contrast-enhanced T1-weighted images was analyzed slice by slice. The inflamed synovium was outlined manually in each coronal slice. The synovial mass in each slice was then calculated by multiplying the outlined area in each slice by the slice thickness. The total synovial mass of the inflamed joint was the sum of the masses of different slices of the slice set covering the entire joint. Between these coronal, stationary T1-weighted images, a dynamic contrast-enhanced sequence was performed with both precontrast and postcontrast images. The coronal dynamic 2-D Turbo Flash sequence had a TR of 800 msec, TE of 2 mm, slice thickness of 4.5 mm, and FOV of 160 × 160 mm2 with temporal resolution of 4 seconds. Using postprocessing software, the positive enhancement integral was calculated from these dynamic images using a circular region of interest (ROI) with a diameter of 8.0–8.7 mm. The circular ROI was placed on a region of synovium in which the greatest increase in signal intensity was assessed visually on the dynamic images. The positive enhancement integral is a calculation of the contrast enhancement that results in a signal increase relative to that in the baseline precontrast images. The contrast agent used was Magnevist at 469 mg/ml (Schering, Berlin, Germany) with a dose of 0.2 ml/1.0 kg of body weight.

A rapid injection was used to obtain a good bolus. The contrast agent was administered via a previously placed 0.9-mm intravenous catheter in the antecubital vein, immediately followed by a 20-ml saline flush at the same injection rate. Imaging was started immediately after finishing the injection of contrast agent.

Data analysis.

Quantification of tracer uptake was performed by standard ROI analysis followed by the calculation of the standardized uptake value (SUV) and the kinetic influx constant (Ki) for 11C-choline and 18F-FDG (15). Using the MRI ROIs as reference, circular ROIs with a fixed diameter of 10.6 mm were drawn on 3–4 transaxial PET planes (images were summed over time, and the last 10 minutes of imaging were deemed representative of the entire series), and these PET ROIs corresponded to the areas of hyperplastic synovium. The concentration of radioactivity in all of the ROIs for each PET study was averaged to yield composite time-activity curves, which were used for further analyses.

The graphic analysis approach, as described by Patlak et al (16), was used to analyze the uptake kinetics of 11C-choline and 18F-FDG. Contrary to unmetabolized 18F-FDG, the input function for the uptake of 11C-choline was corrected for measured radioactive metabolite, which constituted a variable but highly significant fraction of total plasma radioactivity. The rate of glucose uptake was calculated by multiplying the Ki of 18F-FDG by the mean plasma glucose concentration during 18F-FDG imaging, to obtain an index kinetic influx constant, Ki[G].

The SUV was calculated as the average radioactivity concentration in the synovial ROI (in MBq/ml) per injected dose (in MBq) of each radionuclide corrected for the body weight of the patient (in grams). Results are expressed as the mean ± SD gm/ml.

The correlation between MRI and PET measures was calculated by linear regression. For comparison of the differences between the 2 populations, a paired t-test was used. P values of less than 0.05 were considered significant. Analyses were conducted with ORIGIN (Microcal Software, Northampton, MA), version 6.


In all patients with clinical symptoms, a high regional uptake of both 11C-choline and 18F-FDG was observed at the site of arthritic synovium as compared with the unaffected joint (Figure 1). The location of the highest uptake of 11C-choline and 18F-FDG corresponded to the presence of proliferated synovium on MR images. The time course of the inflammatory proliferation of synovium revealed a very fast uptake of 11C-choline and 18F-FDG, which reached a plateau after 10 minutes (Figure 2A). Thus, the graphic analysis performed in accordance with the method of Patlak et al (16), which is an established method for the determination of the 18F-FDG uptake rate, was also applicable for the analysis of the uptake of 11C-choline, with metabolite-corrected plasma used as the input function (Figure 2B).

Figure 1.

Images of inflamed and unaffected knee joints (of patient 3 in Tables 1 and 2) by positron emission tomography (PET) with methyl-11C-choline (11C-choline) and 2-18F-fluoro-2-deoxy-D-glucose (18F-FDG) and by gadolinium-enhanced magnetic resonance imaging (MRI). High uptake of both 11C-choline and 18F-FDG is seen at the site of inflammatory proliferation of synovial tissue (indicated on PET images by color-coding according to the amount of radioactivity, from dark blue [lowest] to hot red [highest]).

Figure 2.

Results of dynamic positron emission tomography imaging of arthritic synovium with 11C-choline and 18F-FDG. A, Representative time-activity curves showing that after an initial fast uptake, the uptake of both tracers reached a plateau at ∼10 minutes after injection. % ID/g = percentage of injected dose per gram of tissue. B, Corresponding Patlak graphic analysis plots, with the slope of the regression line representing the kinetic influx constant (Ki) of the tracer. See Figure 1 for other definitions.

Table 2. Uptake of 11C-choline and 18F-FDG by arthritic synovium on PET scan in comparison with MRI measures*
Volume of synovium, cm3Positive enhancement integralUptake rate (Ki), minute−1SUV, gm/ml
  • *

    11C-choline = methyl-11C-choline; 18F-FDG = 2-18F-fluoro-2-deoxy-D-glucose; PET = positron emission tomography; MRI = magnetic resonance imaging; Ki = kinetic influx constant; SUV = standardized uptake value.

  • Values are the Ki and SUV for 11C-choline/18F-FDG.


In contrast to unmetabolized 18F-FDG, the percentage of unchanged 11C-choline in the plasma decreased rapidly. The radioactive metabolite of 11C-choline in plasma, i.e., 11C-betaine, constitutes a variable but highly significant fraction of total plasma radioactivity. According to the results of radio-HPLC analysis, the percentage and time course of unchanged 11C-choline in arterial plasma was 97 ± 3%, 51 ± 12%, 24 ± 7%, 17 ± 6%, 14 ± 5%, and 13 ± 5% (mean ± SD) at 1, 5, 10, 15, 20, and 25 minutes after injection, respectively. The mean (±SD) arterial plasma glucose concentration during the 18F-FDG scans was 5.2 ± 0.4 mmoles/liter.

The SUVs of 11C-choline and 18F-FDG (mean ± SD of 10 samples each) were 1.5 ± 0.9 gm/ml (range 0.1–3.3 gm/ml) and 1.9 ± 0.9 gm/ml (range 0.3–4.0 gm/ml), respectively (P = 0.017). The rates of uptake, or Ki, of 11C-choline and 18F-FDG (mean ± SD of 10 samples each) were 0.048 ± 0.042 minute−1 (range 0.016–0.146 minute−1) and 0.006 ± 0.003 minute−1 (range 0.003–0.012 minute−1), respectively. The Ki of 11C-choline was ∼8 times as high as that of 18F-FDG (P = 0.009). Nevertheless, the Ki of 11C-choline correlated strongly with the Ki of 18F-FDG (r = 0.840, P = 0.002). There was a close correlation between the Ki of 11C-choline and the SUV of 11C-choline (r = 0.885, P = 0.0007). However, the 18F-FDG Ki was not associated with the 18F-FDG SUV (r = 0.528, P = 0.117 [not significant]). The individual findings are shown in Table 2.

The SUVs and Ki values of 11C-choline and 18F-FDG were compared with quantitative measures of inflammation on MRI, i.e., the volume of synovial tissue (in cm3) and the positive enhancement integral (Table 2). Both the uptake of 11C-choline and the uptake of 18F-FDG correlated highly with the volume of synovium. This is plotted in Figures 3A and B. The highest correlation coefficient was observed between the volume of synovium on MRI and the rate of uptake of 11C-choline on PET scan (r = 0.954, P < 0.0001).

Figure 3.

Linear regression depicting the correlation between findings on positron emission tomography and the results of gadolinium-enhanced magnetic resonance imaging. Correlations were observed between the volume of synovium and the kinetic influx constant (Ki) of 11C-choline (A), the Ki of 18F-FDG (B), the standardized uptake value (SUV) of 11C-choline (C), and the SUV of 18F-FDG (D) at the site of inflammatory proliferation of arthritic synovium. See Figure 1 for other definitions.

The glucose uptake rate, i.e., the Ki[G], was 0.032 ± 0.013 minute−1 (mean ± SD of 10 samples; range 0.014–0.055 minute−1). This correlated well with the volume of synovium (r = 0.817, P = 0.004).

The positive enhancement integral obtained from gadolinium-enhanced MRI showed no correlation with either the Ki values or the SUVs on PET. Moreover, neither uptake of 11C-choline nor uptake of 18F-FDG revealed a statistically significant correlation with clinical laboratory measures, i.e., the ESR or CRP level.


In an attempt to explore the use of radionuclides for specifically depicting the inflammatory proliferation of synovium associated with RA and other arthritic diseases, we performed this pilot study which involved imaging of the inflamed joint by PET with 11C-choline. Our first goal was to study the feasibility of 11C-choline PET for the imaging of arthritic joints, and our second goal was to evaluate the relationship between inflammatory proliferation of synovium and uptake of 11C-choline by correlating the findings on PET with the results of gadolinium-enhanced MRI and with clinical laboratory measures. PET with 18F-FDG served as a reference radionuclide method, owing to its increased availability and use in infection/inflammation imaging (6, 7). In the present study, we have shown that 11C-choline accumulates highly in the target tissue, i.e., inflamed synovium, which allows quantitation of the tracer uptake. Because it plays this role, 11C-choline is a possible candidate for functional imaging of arthritic diseases.

Due to the high uptake of glucose in inflammatory cells, 18F-FDG is an appropriate tracer for the evaluation of inflammation or infection. Generally, 18F-FDG PET allows visualization of highly stimulated glucose metabolism in specific cells during acute and chronic inflammation (17). Accumulation of 18F-FDG has been observed in studies of experimentally induced inflammation and in inflammatory processes such as joint inflammation or RA (18–20). Such accumulation is attributed to the hyperperfusion and hyperemia associated with regional inflammation, and to the increase in glycolytic activity in leukocytes, macrophages, and other cells of inflamed tissue. Using 18F-FDG PET, Polisson et al and Palmer et al have previously demonstrated high glucose metabolism in the joints of RA patients, which correlated well with MRI measures and with the effects of treatment (19, 20). In addition, Danfors et al have demonstrated that PET with 11C-D-deprenyl allows visualization and evaluation of joint inflammation, although the mechanism for high uptake of the tracer in affected joints is not known (21).

Choline is a precursor for the biosynthesis of phosphatidylcholine, which is an essential component of cell membranes (8). Rapidly proliferating cells, e.g., tumor cells, contain large amounts of phospholipids, particularly phosphatidylcholine, compared with that in normal tissue (22–24). This high phospholipid content is believed to be reflected in the increased uptake of choline, making 11C-choline a useful marker for PET imaging of cell membrane proliferation. Previously, 11C-choline PET has been applied for the detection of various tumors, e.g., brain tumors, esophageal carcinoma, prostate carcinoma, and bladder cancer (25–30). Yet, our recent studies on brain tumors have revealed that the rate of 11C-choline uptake is associated with the level of cellular proliferation, but not with the histologic grade of malignancy (31). The uncontrolled, aggressive, and highly invasive destructive growth of synovial cells resembles that seen in neoplasms and characterizes the pathophysiology of RA. Although several attempts have been made to explore the proliferative capacity of synovial cells, this is the first study focusing on the in vivo quantitative imaging of proliferation of synovium.

Intravenously administered 18F-FDG appears to be unmetabolized in plasma during PET imaging. Yet, the major radioactive metabolite of 11C-choline, i.e., 11C-betaine, constitutes a variable but highly significant fraction of total plasma radioactivity and can easily be quantified by radio-HPLC (14). From the radio-HPLC analysis, it was clear that only metabolite-corrected plasma radioactivity could be used as the input function when applying the graphic analysis method to evaluate the rate of 11C-choline uptake and metabolism in tissue. To our knowledge, this report is the first to present Ki values for the illustration of 11C-choline kinetics. Previously, other groups have reported only semiquantitative SUVs for 11C-choline uptake, in which the radioactivity in the plasma was not taken into account, i.e., studies that were performed without taking blood samples during PET imaging.

For the present study, patients underwent 2 sessions of PET imaging, i.e., with 11C-choline (T1/2 = 20 minutes) and with 18F-FDG (T1/2 = 110 minutes). In general, the effective dose of radiation for the patient in PET depends on the injected dose and the radioactive half-life of the tracer. If the administered doses are 440 MBq and 370 MBq for 11C-choline and18F-FDG, respectively, the estimated total radiation dose for the patient is ∼9 mSv (32). The annual dose of natural background radiation is 4–5 mSv. Thus, the 9-mSv total dose of radiation in an 11C-choline/18F-FDG PET study is acceptable and comparable with that of ordinary diagnostic procedures involving x-rays; e.g., whole-body computed tomography gives an effective dose that ranges from 6 to 15 mSv.

We conclude that 11C-choline PET is a technique that has the potential for functional in vivo imaging of arthritic changes at the molecular level. It may be valuable in the future for use in studies assessing the progression of disease in patients with recent-onset RA. Possible changes in 11C-choline influx into the synovium may relate to changes in membrane synthesis as an indication of growth activity, e.g., the level of synovial hyperplasia. However, it must be kept in mind that PET has a lower resolution than that in MRI, and therefore the more important erosions can not be depicted or measured with PET. Subsequent prospective studies with larger numbers of patients are necessary to further characterize the relationship between quantitative 11C-choline PET imaging and clinical and functional measures of inflammatory joint disease.