Introduction

Multiple myeloma (MM) is a B cell neoplasm characterised by the proliferation and accumulation of plasma cells in the bone marrow and by the overproduction of monoclonal immunoglobulins that can be detected in serum and/or urine (M component) [1–5]. MM accounts for approximately 1% of all neoplasms and 10% of haematological malignancies. The incidence of the disease increases with age and reaches a peak during the seventh decade of life. Males are more frequently affected by MM than females (male to female ratio 3:2). The diagnosis of MM basically relies on the demonstration of bone marrow plasmacytosis or histological proof of a plasmacytoma and/or demonstration of an M component in the serum or urine and/or detection of lytic bone lesions [6–9]. Bone damage is present in more than 80% of patients at diagnosis, resulting in a diffuse osteoporotic pattern or in focal disease (single or multiple lytic bone lesions), causing bone pain, pathological fractures and hypercalcaemia [3, 10–13].

As survival and therapeutic options in patients with MM are extremely variable, there is a need for clinical staging that has a strong prognostic value and can stratify patients into different risk groups [14]. It is quite widely accepted that only patients with active or symptomatic myeloma require systemic therapy. Bone abnormalities (lytic or osteopenic) are considered one of the myeloma-related organ dysfunctions [15]. Moreover, in patients with bone involvement the incidence of vertebral fractures can be reduced by using potent i.v. bisphosphonates, which are now available in the therapeutic armamentarium for MM. Several prognostic parameters have been evaluated and several staging systems have been proposed. The most commonly used staging system is that introduced in 1975 by Durie and Salmon, in which one of the parameters considered is the number of lytic bone lesions detected by whole-body X-ray (WBXR) [16]. This technique is known to be rather insensitive for the detection of bone damage, and bone localisations are detected only in the late phase. Radiographs can significantly underestimate the extent and magnitude of bone and bone marrow involvement. A reliable technique providing both functional and morphological information would therefore be very useful in order to identify bone involvement.

Magnetic resonance imaging (MRI) has been recognised to be the most appropriate tool for this purpose, showing a very good sensitivity compared with WBXR, especially at disease onset [10, 17–21]. The main limitations of MRI are the impossibility of performing the scan in the presence of a metallic prosthesis or severe claustrophobia and, more importantly, the fact that there is a partial field of view (FOV) including only the spine and pelvis. MRI cannot usually explore the skull, ribs, humerus, femur and clavicle, which are frequently affected by lytic lesions. Furthermore, MRI cannot easily distinguish between active lesions and fibrotic lesions, so definition of response to therapy is not always possible. Recently, whole-body MRI has been introduced thanks to the development of ultrafast data acquisition and high-performance gradient systems with faster gradient echo. Whole-body MRI allows the acquisition of images of the whole skeleton in about 60 min, but it is not yet widely available and has low sensitivity for the detection of rib lesions, which are present in a significant number of MM patients [22, 23].

Among other options, low-dose computed tomography (CT) has proved to be more sensitive than WBXR for the identification of lytic lesions and the assessment of fracture risk, but it is not included in the MM staging system [24]. 99mTc-diphosphonate bone scanning and gallium scanning, although occasionally useful, are unreliable in MM owing to minimal osteoblastic activity and hypovascularity of lesions.

18F-fluorodeoxyglucose positron emission tomography integrated with CT (FDG PET/CT) is a total body imaging technique that permits the analysis of tissular glucose metabolism and explores in a single step both bones (spine and pelvis as well as the rest of the skeleton) and soft tissues, which are occasionally the site of MM localisations. The use of a hybrid scanner and the production of a functional image superimposed on a CT morphological map allows exact localisation of small hypermetabolic findings that are not always easy to differentiate from soft tissue lesions on the basis of PET alone. In particular, PET/CT has an increased accuracy over PET alone because, owing to the morphological map, it is possible to recognise even small and/or slightly active lesions that are difficult to identify on the basis of a purely metabolic image.

The aim of this study was to compare the ability of WBXR, MRI of the spine and pelvis, and 18F-FDG PET/CT to detect bone localisations in patients with newly diagnosed MM and to assess the possible added value provided by 18F-FDG PET/CT in the diagnosis of bone involvement in MM. Although low-dose CT was performed for PET attenuation correction, the results of CT alone were not compared with those of PET/CT because CT, unlike WBXR, is not included in the MM staging system.

Materials and methods

We studied 28 consecutive patients (21 males and seven females; mean age 55 years, SD 9 years, range 35–74) with symptomatic MM (according to international guidelines) who had not been previously treated and were referred to our PET Centre by the Haematology Unit of our Hospital from March 2003 to September 2004.

Patients were informed about the PET/CT procedure and signed a written consent form before the tracer injection. To assess bone involvement by the disease, each patient underwent WBXR, MRI (spine and pelvis) and 18F-FDG PET/CT within a maximum interval of 1 month. WBXR included skull, spine, pelvis, ribs, femora and humeri. Specific therapy was started after the performance of these diagnostic procedures in all patients.

MRI was performed using a 1.0-Tesla system with a spinal coil. Sagittal images included a T1-weighted spin echo sequence and a fat-suppressed T2-weighted fast spin echo sequence. The T1-weighted images were repeated after the intravenous injection of gadolinium chelate. MRI studies were reviewed by two radiologists. Myeloma patients are usually classified into three categories according to the MRI pattern of spinal bone involvement [25, 26]: normal pattern, focal pattern and diffuse infiltrative pattern. For the purpose of our study, the same criteria were applied. When the focal pattern was observed, the exact number and site of lesions were reported.

18F-FDG PET/CT scan was carried out according to a standard procedure: after fasting for at least 6 h, and in the absence of antidiabetic therapy, each patient was injected with 5.3 MBq/kg of 18F-FDG. After injection, patients were hydrated and the uptake phase was 60–80 min. The scan was performed with a dedicated PET/CT tomograph (GE, Discovery). Data acquisition was performed in 2D mode, at 4 min per bed position, and attenuation correction was performed with a CT-based method (120 kV, 80 mA) . Skull, upper limbs and femora were included in the FOV. These CT parameters are standard for PET/CT studies and permit differentiation between tissues with good spatial resolution while ensuring that the patient does not receive a high radiation dose. Each PET/CT scan was read by two nuclear medicine physicians who were unaware of the WBXR and MRI results, and the report was written consensually. Each visible area of focal FDG uptake in bone (excluding joints) was considered positive for a myelomatous lesion. The standardised uptake value (SUV) was calculated using the following formula: tissue concentration (MBq/g)/injected dose (MBq)/body weight (g). At the time of PET, none of the patients had been treated with antiblastic therapy. 18F-FDG PET/CT results were compared with WBXR and MRI results in terms of the number and site of lesions detected.

Results

18F-FDG PET/CT vs WBXR

In 16/28 patients (57%), PET/CT detected more bone lesions than WBXR (group 1), while in the remaining 12 patients (43%) the two methods yielded equivalent findings (group 2) (Table 1).

Table 1 Comparison between WBXR, 18F-FDG PET/CT and MRI results
Full size table

In group 1, comprising 16 patients, PET/CT detected more lesions than WBXR: all these lesions were included in the WBXR FOV but were small, and therefore below the contrast resolution limit of the plain film. In nine out of these 16 patients, WBXR was completely negative while PET/CT detected one or more bone lesions. It is also of interest to note that in one patient of group 1 previously diagnosed with plasmacytoma (patient 7), 18F-FDG PET/CT detected a second bone lesion, possibly shifting the diagnosis to MM. In group 2, eight patients had one or more lesions, detected both by WBXR and by PET/CT, while four patients were negative with each method.

18F-FDG PET/CT vs MRI

In 7/28 patients (25%), 18F-FDG PET/CT detected more lesions than MRI, all being located outside of the FOV of MRI [ bone lesions in six cases (Fig. 1) and a hepatic lesion in one case] (group 3). In 14 patients (50%), 18F-FDG PET/CT and MRI detected the same number of lesions (group 4), while in the remaining seven patients (25%) 18F-FDG PET/CT detected fewer pathological findings (group 5) (Table 1).

Fig. 1
figure 1

FDG PET coronal slices and corresponding PET/CT fused transaxial slices showing MM bone lesions outside the MRI FOV

Full size image

Among group 3 patients, one had been classified as having plasmacytoma after WBXR and MRI, while 18F-FDG PET/CT showed a second lesion, shifting the diagnosis from plasmacytoma to MM (patient 14, Fig. 3). In two patients (nos. 11 and 17) in whom MRI detected no lesions, PET showed findings consistent with disease areas, mainly located outside of the MRI FOV (liver and ribs).

Fig. 2
figure 2

18F-FDG PET/CT showing increased uptake in the pelvis, which was assumed to be solitary plasmacytoma of bone after WBXR and MRI. PET showed a second lesion, shifting the diagnosis to MM

Full size image

In group 5, MRI identified an infiltrative pattern in the spine in all seven patients without evidence of focal lesions (four of the seven had disease in all the vertebrae with a typical infiltrative pattern, while three had an infiltrative pattern limited to some vertebral bodies): this kind of pattern was never documented by 18F-FDG PET/CT scan (Fig. 2). In one of the four patients with a diffuse infiltrative pattern, MRI detected an osteoporotic pattern with three vertebral pathological fractures that were negative at PET.

Fig. 3
figure 3

T1- and T2-weighted MRI of the spine showing a diffuse infiltrative pattern. T1-weighted images show hypointense bone marrow, and T2-weighted images show hyperintense bone marrow; in normal conditions, T1-weighted images give a more intense signal, and T2-weighted images are hyperintense. The corresponding sagittal PET slice shows no pathological findings

Full size image

The attenuation correction CT scan did not show any anatomical abnormality all the seven patients with an infiltrative pattern on MRI.

SUVmax

SUVmax based on body weight was calculated for each lesion. Table 2 shows, for each patient, the SUVmax of the least active lesion, the SUVmax of the most active lesion and the disease stage.

Table 2 SUVmax of the least and the most active lesion and disease stage for each patient
Full size table

SUVmax showed a high varibility within each patient and from patient to patient, ranging from 1.9 to 7.2. Although we did not find any correlation between SUVmax and the clinical stage of the disease, it is important to note that SUVmax is significantly underestimated for lesions smaller than 1 cm owing to the partial volume effect. In our opinion, this could have been the cause of such high intra-patient variability in SUVmax.

Discussion

In MM the correct identification of bone involvement is part of the Durie and Salmon staging system. Regarding bone imaging, this system takes into consideration WBXR, the only technique available in 1975 [16]. Radiographs remain a good method for surveying the skeleton and evaluating fractures and bone lesions at risk of fracture. However, they can easily underestimate bone involvement, especially in the spine, where overlying tissue and the rib cage hinder the assessment of osteolysis. In addition, WBXR cannot distinguish between idiopathic osteoporotic vertebral fractures and fractures due to MM. For these reasons, imaging techniques with high sensitivity for bone lesions acquire more importance both for the exact evaluation of bone disease extension at disease onset and for the assessment of response to therapy.

MRI enables direct visualisation of the bone cavity content, which accounts for the superiority of MR images over plain radiographs for the detection of spinal lesions in haematological malignancies [2]. Furthermore, the MRI pattern has prognostic implications [27–31]. In particular, MRI patterns before and after treatment are used to demonstrate the effect of therapy and to confirm response, especially in patients with equivocal clinical findings or non-secretory disease [31]. In addition, several authors have recently suggested that MRI could play a role in disease staging [25, 28, 32]. Most of these studies have demonstrated the superiority of MRI over plain radiographs for the detection of lesions in the spine and pelvis. The main limitation of MRI is the incomplete field of view including only spine and pelvis, which account for only part of the bone involvement in MM patients. Baur et al. recently proposed a combination of an MRI staging system with the Durie and Salmon staging system, and found that this combination showed the strongest correlation with survival [26]. In addition, it should be mentioned that both MRI and FDG-PET have been integrated in a new staging system, called the Durie and Salmon PLUS, contained within the recent myeloma guidelines developed by the Scientific Advisors of the International Myeloma Foundation [15].

The recent introduction of whole-body MRI has strengthened the value of this technique in the evaluation of MM patients. The main limitations of whole-body MRI are that it is not yet widely employed or available in clinical practice and that it seems to have a low sensitivity for the detection of rib lesions, which are quite frequent in MM. Furthermore, the acquisition time, though acceptable, is still long (at least 60 min), preventing claustrophobic patients from undergoing the scan [22, 23]. Finally, like conventional MRI, this technique cannot be used in patients with metallic prostheses.

18F-FDG PET/CT is a metabolic imaging technique that includes the whole skeleton in the FOV. Experience with FDG-PET in MM is considerably more limited than that with MRI [33–35]. In MM, as with most malignant tumours, PET is a useful tool to evaluate disease activity, to detect extra-osseous disease involvement, to direct local therapy such as radiation, to assess patients with non-secretory myeloma and to evaluate response to therapy by measuring visually and semi-quantitatively the FDG uptake in each neoplastic lesion. The major advantages of FDG-PET are the ability to perform whole-body examinations, the potential to detect medullary and extramedullary lesions in a single examination and the possibility of distinguishing between new active disease and old disease, scar tissue, necrotic tissue, radiation changes and separate benign disease.

To date, limited data have been published comparing FDG-PET and MRI in patients with MM. In this study, therefore, we compared 18F-FDG PET/CT with WBXR (the only skeletal survey examination included in MM staging system, and widely used) and MRI results in 28 patients with newly diagnosed symptomatic MM. We found that PET showed a higher number of lesions in 16/28 (57%) patients compared with WBXR and in 7/28 (25%) patients compared with MRI. WBXR proved to be much less sensitive than 18F-FDG PET/CT for the detection of myelomatous bone lesions as it is a planar technique: small lytic lesions and early lesions (those with less than 30–50% of bone resorbed) are not detected and the number of lesions is consequently underestimated.

According to published data and to our results, MRI is very sensitive for the localisation of small myelomatous bone lesions. In our study, MRI detected more pathological findings than 18F-FDG PET/CT in 7/28 (25%) patients, showing an infiltrative pattern in the spine in all seven cases (diffuse or limited to some vertebrae). In our population, PET did not detect this kind of pattern as there were no focal lesions. In these patients, PET showed a very mild and diffuse increase in FDG uptake by the spine: this is a very common finding that must be considered physiological because it is frequently found even in the healthy population (for example in young or mildly anaemic patients). In this group of patients, low-dose CT also did not show morphological abnormalities. In the above-mentioned 7/28 patients in whom 18F-FDG PET/CT detected more lesions than MRI, this was because lesions were located outside of the MRI FOV. It is important to note that PET changed the diagnosis of plasmacytoma to MM in two patients and showed bone lesions in two patients with negative MRI, potentially changing the therapeutic approach.

Although it has been proved that low-dose CT (in this case used for PET attenuation correction as well as a morphological map) is accurate and more sensitive than WBXR for the evaluation of bones in MM [24], we did not include it in this protocol. CT, in fact, is not yet part of the MM staging system, which still considers WBXR to be the gold standard for exploration of bones. Of course, by reading the attenuation correction CT as an independent examination it may be possible to skip WBXR, preventing the patient from being subjected to unfruitful radiation exposure.

In conclusion, this is a preliminary study showing that 18F-FDG PET/CT can be a useful tool to evaluate bone involvement in MM patients at disease onset, even though MRI of the spine is important to identify the infiltrative pattern usually missed by FDG-PET and the morphological studies. At the same time, radiographs and low-dose CT permit estimation of fracture risk, which is not possible on the basis of the results of FDG-PET or MRI. Therefore, we consider that careful evaluation of MM bone disease at diagnosis should now include 18FDG-PET/CT together with WBXR and MRI of the spine.