NUCLEAR (SCINTIGRAPHIC) METHODS AND FDG-PET IN RHEUMATOID ARTHRITIS AND OSTEOARTHRITIS

Imaging of inflammatory joint diseases remains challenging. The development in recent years of biologic treatments that are highly efficacious as well as expensive has stimulated the research for imaging of inflammation. This has been particularly true for rheumatoid arthritis (RA), which constitutes the most prevalent inflammatory arthritis, affecting 1% of the population. RA is an autoimmune systemic disease characterized by chronic inflammation of the synovium, with a massive leukocyte infiltration, proliferation of the synovial membrane, and neovascularization. The hyperplastic and hypertrophic rheumatoid synovium, termed the pannus, if left untreated gradually erodes the adjacent cartilage and bone, leading to joint degradation and marked disability. Although conventional radiography displays joint space narrowing and bone erosions, the hallmark of rheumatoid disease, it does not allow evaluation of the inflammatory component of the disease but only its end-stage irreversible characteristics. MRI can detect bone erosions earlier than conventional radiography and also allows the study of cartilage, tendons, and ligaments, as well as the synovium, especially after the injection of gadolinium. However, in gadolinium-enhanced MRI the uptake of the contrast agent by the inflamed synovium is also due to hypervascularization and capillary permeability. Ultrasonography is noninvasive, allows the study of many joints in one time frame, and when associated with power Doppler imaging can also give information about synovial vascularization. Although these two imaging modalities are of high value in the clinical assessment of arthritis (as discussed more fully in other chapters), they remain purely morphologic and do not provide information about cell metabolism. In this chapter we review the nuclear medicine techniques available for assessing arthritis, with a particular interest in positron emission tomography (PET).

BONE SCINTIGRAPHY

Bone scintigraphy is one of the most commonly performed procedures in nuclear medicine and is a classic imaging technique used for the investigation of bone and joint diseases. The first suitable radionuclides developed for bone scintigraphy in the early 1960s were strontium-85m, strontium-87m, and fluoride-18. Technetium-99m ( ^99mTc)-labeled diphosphonates were introduced in the 1970s and have been used since then, the most common being ^99mTc-methylene diphosphonate (MDP) and ^99mTc-hydroxymethylene diphosphonate (HDP). The energy of the photon emitted by the ^99mTc nuclide is well adapted to the physics properties of the current gamma cameras. Both MDP and HDP have a high affinity for bone, and approximately 50% of the dose of ^99mTc-diphosphonate is distributed in the skeleton within 3 hours after intravenous administration, the remainder being excreted in the urine. Their uptake in bone is related not only to osteoblastic activity but also to the regional blood flow. Accordingly, an increased uptake is seen in synovitis where there are increased blood flow, expanded blood pool volume, and vascular permeability. The hyperemia involves not only the synovial membrane but also the juxta-articular bones. Therefore, although this technique is very sensitive for the detection of joint and subchondral bone abnormalities, it cannot discriminate accurately between actively inflamed joints and chronically damaged joints and does not allow correct evaluation of inflammatory arthritis. Other radiotracers have thus been investigated for the evaluation of arthritis: gallium ( ⁶⁷Ga) citrate, indium-111 chloride ( ¹¹¹InCl ₃), ^99mTc-hexamethylpropylene amine oxime (HMPAO)-labeled leukocytes, ^99mTc-liposomes, ^99mTc-polyclonal human immunoglobulin G (IgG) monoclonal antibodies to granulocytes, and ^99mTc-labeled anti-CD4 monoclonal antibodies. Indium and gallium possess a high affinity for iron-binding proteins and probably bind to transferrin receptors abundant in the inflamed synovium, explaining their affinity for the inflammatory site compartment. However, scanning with gallium necessitates a 24-hour delay before imaging, entailing two patient visits, and is associated with a relatively high radiation exposure; in addition, the image quality is limited. Other tracers used for evaluating inflammation include radiolabeled liposomes, which are phagocytosed by macrophages abundant in inflammatory sites, and radiolabeled IgG, which probably leak at the capillary bed and are trapped by Fc receptors of the inflammatory cells, targeting inflammation nonspecifically. These “inflammation-targeted” radiotracers are also used for evaluating infectious diseases and are not specific for any arthritis. The improved understanding of pathophysiologic mechanisms of disease has led to the development of new radiotracers targeting more specific markers of inflammation. Scintigraphy targeting the adhesion molecule E-selectin, expressed by activated endothelial cells in various inflammatory states and, in particular, postcapillary venules of rheumatoid synovium, has been developed and shown to be more specific for targeting active joint inflammation compared with ^99mTc-MDP. On the other hand, ^99mTc-labeled interleukin-8 allows the detection of infection without accumulating in noninfectious inflammatory disorders. Another potential target is the synovial apoptosis, which is shown to be defective in the rheumatoid synovium. During apoptosis, annexin V binds to the cell membrane phospholipid phosphatidylserine, and this binding allows apoptosis imaging. Post and colleagues have shown in collagen-induced arthritis in a mouse model of RA that there is an increased uptake of annexin V in paws of arthritic animals when compared with controls. Treatment with methylprednisolone for 1 week decreased annexin V uptake. Annexin V imaging in human patients with various cancers (Apomate, Theseus Imaging, Boston) has demonstrated the induction of apoptosis in the first 48 hours after chemotherapy. Patients with annexin V uptake at tumor sites had a complete or partial response, in contrast to the progressive disease in those without significant annexin V uptake, thus conferring a prognostic value to annexin V imaging. Apomate could therefore be of use for evaluation of apoptosis in RA patients.

These techniques remain experimental for the time being, and bone scintigraphy with ^99mTc-labeled diphosphonates remains the standard nuclear medicine approach for assessing joint involvement in a whole-body approach. Despite a low specificity and poor resolution, bone scintigraphy reveals the pattern of joint involvement and may therefore be of value, alone or in combination with other techniques, such as MRI, in the differential diagnosis of undifferentiated arthritis. Recent technologic developments, in particular the advent of gamma cameras with tomographic capabilities combined with a CT scanner (SPECT-CT) may improve the clinical usefulness of bone scintigraphy, but no data are available yet.

POSITRON EMISSION TOMOGRAPHY

Positron emission tomography (PET) is a technique using molecules labeled with isotopes that emit positrons from their nucleus. The most commonly used isotopes are fluoride-18 ( ¹⁸F), oxygen-15 ( ¹⁵O), nitrogen-13 ( ¹³N), and carbon-11 ( ¹¹C). They are created in a cyclotron, a device used to accelerate charged particles to create the relatively short-lived positron-emitting isotopes. The most commonly used tracer is 2-deoxy-2-( ¹⁸F)fluoro -d -deoxyglucose (FDG) in which one of the hydrogens has been replaced by the ¹⁸F radioisotope. After intravenous injection, FDG distributes to the extravascular compartments and is taken up by the cells according to their level of glucose metabolism. Indeed, FDG is transported through the cell membrane via glucose transporters (GLUT) into the cytosol, phosphorylated by the hexokinase, and trapped intracellularly as FDG-6-phosphate. Most tumor cells display highly increased glucose metabolism and FDG uptake, hence the numerous oncologic applications of the technique. Activated leukocytes, through different biologic processes, similarly display increased FDG uptake. The spatial resolution of the most recent clinical PET scanners is 4 to 6 mm. It is a fully tomographic technique, offering 3D visualization of the glucose metabolism. FDG-PET, therefore, presents many advantages over bone scintigraphy. Its uptake by inflammatory cells results in the direct visualization of the inflammatory synovitis rather than the indirect hyperemia detected by bone scintigraphy. Its spatial resolution and count rates are also much improved when compared with planar scintigraphy. Finally, the advent of PET/CT allows for a precise anatomic location of the increased metabolic activity.

Methodologic Aspects of PET

The optimal methodology for performing and interpreting the PET/CT studies is not fully defined. To fully contribute to the diagnostic information, the CT scan should be acquired with the appropriate parameters in terms of dose and collimation, possibly with intravenous contrast agents. Whether such a full diagnostic CT is really needed in addition to the low-dose CT remains an unanswered question. Because inflammatory joint diseases often affect young patients, radiation dose has to be taken into account. By using a low-dose CT and injecting 222 to 370 MBq FDG, we obtain good quality images while keeping the equivalent dose in the 8- to 12-mSv range, which is acceptable.

Although a typical pattern of synovitis is easily recognized on the PET images, quantitative measurements are desired to assess the activity of the disease and its response to treatment. As in oncologic PET, the most appropriate quantitative methods will have to be determined. The simple standard uptake value (SUV) is likely to be the best parameter. The question of which normalization (body weight, body surface area, or lean body mass) should be preferred for calculating the SUV remains to be answered, although there should not be major differences between those methods. There are indications that FDG uptake by inflammatory lesions is more influenced by blood glucose level than cancer cells. In particular, glucose loading decreased the expression of the GLUT-1 transporters in noninfectious inflammatory lesions in a rat model. It may therefore be critical to correct for the patient’s blood glucose level.

A well-known limitation of PET stems from its limited spatial and anatomic resolutions. Although this is not a major issue regarding large joints such as the knees, it may be very difficult to identify synovitis in small joints of the hands or feet. Indeed, these patients often also suffer from inflammation of the tendon sheaths, which may be mistaken for the pannus itself. The same problem may arise when high muscular uptake is observed. In all cases, reading such studies is a tedious and time-consuming process. Nevertheless, the reported intraobserver and extraobserver coefficients of variation are consistently low. The visual identification of the synovitis pattern was obtained with κ values of 0.90 and 0.82 (intraobserver and extraobserver variability, respectively). For measuring the SUVs, the intraobserver coefficient of variation was 3.9%. The extraobserver coefficient was higher, 14.9%, and related to the type of joint, ranging from 0% in the knees to 23% in the metacarpophalangeal joints. These results were obtained with a now outdated PET device, and without the help of the CT information. Palmer and coworkers obtained similarly high interobserver agreement values. These data indicate that, at least in the context of a prospective study, the metabolic activity of the inflamed synovium may be reproducibly quantified. The added value of PET/CT over PET is likely to be significant but remains to be fully evaluated.

PET in Preclinical Models of Inflammation

Although the uptake of FDG by malignant tumors is the basis for FDG-PET in the evaluation of cancer, false-positive findings have been encountered in inflammatory pathologic processes. Animal models have demonstrated that FDG uptake by tumors is not only due to the tumor cells themselves but also to the inflammatory cells appearing in association with growth or necrosis of the tumor. FDG accumulation was higher in macrophages and young granulation tissues than in the tumor cells in mice transplanted subcutaneously with malignant tumors. Up to 29% of the glucose utilization was derived from nontumor tissue in these lesions. Microautoradiographic studies of transplanted tumors in mice revealed a higher and faster FDG uptake in the newly formed granulation tissue around the tumor and in macrophages than in viable tumor cells, allowing a differentiation between neoplastic and non-neoplastic cells using dynamic analysis of FDG uptake. A turpentine-induced model of inflammation in rats also demonstrated a FDG uptake in the inflammatory tissue, in the zone surrounding the abscess wall consisting of young fibroblasts, endothelial cells, macrophages, and neutrophils. This suggests that macrophages and neutrophils in inflammatory tissue utilize glucose as an energy source for chemotaxis and phagocytosis while fibroblasts also use glucose for proliferation. This glucose utilization by inflammatory cells is made possible by their increased expression of glucose transporters when they are activated. In conclusion, these data demonstrate the increased glucose metabolism of many inflammatory cell types and the FDG uptake by inflammatory tissues and are the basis for the use of FDG-PET in the detection and monitoring of chronic inflammatory pathologic processes.

PET in Animal Models

PET can be useful for studying mechanisms of arthritis. Many animal models of RA have been developed and have given an insight into the pathophysiologic mechanisms of arthritis. The K/BxN murine model of RA is characterized by the spontaneous development of a chronic polyarthritis in mice expressing the transgenic KRN T-cell receptor on the nonobese diabetic (NOD) genetic background. The transfer of K/BxN autoantibodies specific for glucose-6-phosphate isomerase (GPI) into naive mice of most strains can induce the acute symptoms of arthritis. This joint selectivity of anti-GPI antibodies is surprising in view of the ubiquitous expression of GPI. By using PET to track the movement and accumulation of GPI-specific antibodies in vivo, Wipke and associates have shown that the autoantibodies localize specifically to distal joints in the front and rear limbs within minutes of intravenous injection, in contrast to control IgG. This joint-specific localization of a ubiquitously expressed self-antigen expands our understanding of autoimmunity; that is, a tissue-specific autoimmune response can be directed against an antigen whose expression is not limited to specific tissues.

PET in Rheumatoid Arthritis

The first reports of accumulation of FDG in inflamed joints were incidental findings in two patients evaluated for cancer screening or follow-up. In a patient with thyroid cancer and RA, the FDG uptake was noted not only in the inflamed joints but also in lung rheumatoid nodules devoid of any malignant or infectious process. Palmer and colleagues reported more than 10 years ago the ability of PET to quantify joint inflammation in patients with inflammatory arthritis of the wrist. Nine patients with RA and 3 with psoriatic arthritis exhibiting active disease and observable synovitis of the wrist were evaluated by clinical examination, MRI, and PET. Patients were evaluated after a washout period for nonsteroidal anti-inflammatory drugs, 2 weeks after treatment with a nonsteroidal anti-inflammatory drug or with low-dose steroid (10 mg prednisone daily), and finally 12 to 14 weeks after treatment with low-dose methotrexate (5-10 mg weekly). All baseline PET images demonstrated increased metabolic activity in the wrist, mainly in the radiocarpal and distal radioulnar joints, but also in the different tendon sheaths. The regions of greatest FDG uptake on PET images corresponded to the presence of enhancing pannus on MRI. Quantitative assessments found a significant linear correlation between the volume of enhancing pannus (VEP) measured by MRI and total uptake value (TUV) as well as regional uptake value (RUV) of FDG measured by PET. Clinical measurements of inflammation in the imaged wrist (pain, tenderness, and swelling) were closely correlated with VEP, RUV, and TUV. This correlation between PET evaluation of synovitis and MRI measurement of pannus in an individual joint has been confirmed in knee synovitis by our group. We evaluated knee synovitis in 16 patients with active RA by clinical examination, PET, ultrasonography, and MRI, all being conducted by independent assessors within 4 days. Physical examination determined the presence of swelling and of tenderness. PET acquisition was started 93 minutes on average after intravenous injection of the tracer (4 MBq/kg), and knee positivity was determined by the uptake of FDG in areas presumed to correspond to joint synovium. The metabolic activity was quantified using the maximum pixel value of the SUV normalized for the lean body mass. Ultrasonography was performed using a B-mode 13-MHz transducer, and positivity was defined as a synovitis of 1-mm thickness. Positivity for MRI was visually appreciated, and synovitis was defined as enhancement of thickness greater than the width of the joint capsule after gadolinium injection. Dynamic MRI allowed the measurement of the relative enhancement at 30 seconds after gadolinium injection (RE ₃₀) in the external and internal recesses, as well as the rate of early enhancement at 55 seconds after gadolinium injection (REE ₅₅), parameters shown to be related to active RA. Knee synovitis was found to be present in 13 of the 16 knees by clinical evaluation, 11 knees by PET ( Fig. 7-1 ), 11 by MRI, and 12 by ultrasonography. Positivity on one imaging technique was significantly associated with positivity on the other two. In comparison with PET-negative knees, PET-positive knees displayed higher SUVs, higher MRI parameters, and higher ultrasound-determined synovial thickness. SUVs were significantly correlated to MRI parameters, synovial thickness, and serum levels of C-reactive protein and matrix metalloproteinase-3, a synovial-derived parameter reflecting joint inflammation. These correlations suggest strong relationships between the size of synovitis measured by ultrasonography, its metabolic activity measured by PET, and also its vascularization and leukocyte infiltration shown to be correlated with the dynamic MRI parameters. Roivanen and coworkers have also shown a correlation between MRI assessment of synovitis and PET assessment using not only FDG, a marker of inflammation, but also ¹¹C-choline, an indirect marker of cellular proliferation. In all 10 patients with clinical symptoms, high uptakes of both ¹¹C-choline and FDG were noted at the site of the arthritic synovium compared with the unaffected joint. The location of the highest uptakes of ¹¹C-choline and FDG corresponded to the presence of proliferating synovium on MR images. PET can therefore assess the metabolic activity of synovitis, not only because the inflammatory cells have an increased uptake of glucose but also because the pannus contains rapidly proliferating cells that contain large amounts of phospholipids, particularly phosphatidylcholine.