This study evaluated the prognostic value of a three-grade staging system of spinal involvement using magnetic resonance imaging (MRI) in patients with multiple myeloma and determined its usefulness as an independent parameter in the staging system of Durie and Salmon.
Seventy-seven previously untreated patients with multiple myeloma underwent MRI of the thoracic and lumbar spine with unenhanced T1-weighted spin echo and short-tau inversion time inversion recovery sequences. The patients were evaluated according to their infiltration patterns and the extent of bone marrow involvement was staged using a three-grade scale: Stage I, no focal or diffuse infiltration; Stage II, 1–10 foci or mild diffuse infiltration; Stage III, more than 10 foci or strong diffuse infiltration.
The infiltration patterns had no significant effect on survival. Of 77 patients, 25 would have been understaged using the standard staging system of Durie and Salmon without the findings of MRI and 8 patients would have been understaged if the staging was based only on MRI. The combination of the staging system of Durie and Salmon and MRI was highly significant with respect to survival (P < 0.0001, log rank analysis). MRI staging I–III was independent of the staging system of Durie and Salmon (Cox regression model).
Survival in patients with multiple myeloma is extremely variable.1, 2 New therapy approaches, such as high-dose chemotherapy, stem cell transplantation, and thalidomide, can improve response and/or survival rates.3–7 In addition, the use of bisphosphonates can reduce the incidence of vertebral fractures.8, 9 These facts suggest the need for a clinical staging that has a strong prognostic value and can divide the patients into different risk groups. Several prognostic parameters have been evaluated and new staging systems have been proposed.10–20 However, the staging system of Durie and Salmon, developed in1975, is still in widespread use.21 Patients without significant infiltration and disease activity are prone to a “watch and wait” strategy.7 Chemotherapy does not lead to prolonged survival in these patients. Conversely, patients with strong disease activity and high-grade tumor burden are eligible for high-dose chemotherapy, bone marrow transplantation, or thalidomide therapy.4–6 In older patients or patients with intermediate tumor cell burden, the disease can be controlled by conventional therapy schemes with alkylating agents and prednisone.7 Because bisphosphonates can reduce the incidence of vertebral fractures, early detection of bone marrow infiltration is important. MRI enables the direct visualization of the bone marrow infiltration patterns and extension in patients with multiple myeloma.22–24 Radiographs are negative in many cases of multiple myeloma, even though focal or diffuse bone marrow involvement is present.25–32 The presence or absence of spinal involvement in MRI resulted in significantly different prognosis among myeloma patients.20, 28, 29, 31, 33, 34 In those studies, the presence or absence of spinal involvement was noted as a dichotomous decision. In clinical practice, problems arise when the information provided by the MRI has to be arranged into a clinical and prognostic context because the clinical staging systems consist of three grades. In addition, a more differentiated staging is needed because therapy options have to be chosen carefully and more aggressive therapies result in a higher mortality rate. Our aim was to create a simple three-grade staging system of spinal involvement using MRI. We evaluated the prognostic significance of this MRI staging system and incorporated it into the staging system of Durie and Salmon to determine whether it constituted an independent prognostic factor.
MATERIALS AND METHODS
Over a 5-year period, 85 patients with multiple myeloma were examined with MRI of the spine. Inclusion criteria were patients with histologically proven multiple myeloma without previous therapy. Exclusion criteria were patients with multiple myeloma who had additional malignancies. Three patients were excluded from the survival analysis: two died of coronary heart disease and one committed suicide. Five patients were treated with high-dose chemotherapy and autologous stem cell transplantation, both of which improve survival when compared with conventional chemotherapy schemes.3, 5 Therefore, these five patients were also excluded from the survival analysis. Therefore, 77 patients were included in the statistical evaluation, 29 of whom were female and 48 of whom were male. Their mean age was 60 years (range, 32–86 years). Of the 77 patients, 44 had IgG myeloma, 12 had IgA myeloma, 17 had Bence Jones myeloma, 3 had nonsecretory myeloma, and 1 patient had IgD myeloma. In all patients, the following laboratory parameters were determined: hemoglobin level (mg/dL), calcium level (normal range, 2.05–2.65 mmol/L), M-protein component, and creatinine level (normal range, < 1.0 mg/dL). Using these parameters and radiographs of the skeleton, the patients were staged according to the staging system of Durie and Salmon.21
MRI was performed on a 1.5-T scanner (Siemens Medical Systems, South Iselin, NJ) with the use of a spine array surface coil. T1-weighted spin echo (SE) sequences (repetition time [TR]/echo time [TE] 450/12 ms), and short-tau inversion time inversion recovery (STIR TR/TE/inversion time [TI] 3660/150/60 ms) were performed in sagittal direction (field of view: 40 cm; slice thickness: 4 mm). Two experienced radiologists read the films in a consensus reading and evaluated the infiltration patterns. They determined whether the bone marrow of the spine had a normal appearance with high signal intensity on T1-weighted SE images and low signal intensity on STIR images (Fig. 1). Focal involvement was determined if focal areas of high signal intensity were apparent on STIR images, which corresponded to areas of low signal intensity on T1-weighted SE images (Fig. 2). Diffuse involvement was determined if the signal intensity was homogeneously reduced on T1-weighted SE images (Figs. 3, 4). An inhomogeneous or “salt-and-pepper” pattern was defined if the appearance on T1-weighted images was patchy with small areas of high signal intensity beside small areas of lower signal intensity (Fig. 5). In these cases, no focal areas of signal increase were apparent on STIR images.
The extent of the bone marrow involvement was graded on a three-grade scale. Stage I included normal-appearing bone marrow of the spine (Fig. 1). Nodules were counted when there was focal involvement. The presence of 1–10 nodules was graded Stage II and the presence of more than 10 nodules was graded as “multifocal”Stage III (Fig. 2). When there was diffuse infiltration, a moderate signal intensity decrease on T1-weighted SE images was graded Stage II (Fig. 3). The signal intensity of the bone marrow still had to be higher than the signal intensity of the intervertebral discs. A strong signal intensity decrease of the bone marrow was graded Stage III (Fig. 4). In these cases, the signal intensity of the bone marrow had to be as low as the intervertebral disc or even lower, representing high-grade diffuse tumor involvement. If combined diffuse and focal involvement was present, diffuse and focal involvement was graded separately and staged according to the higher grade of involvement. For example, Stage II diffuse involvement and Stage III focal involvement (> 10 nodules) comprised Stage III.
Statistical analysis was processed with the SAS statistical software package. Survival analysis was made with the log rank test (Kaplan–Meier curves). Multivariate analysis was performed using the Cox regression model.
The following infiltration patterns were found. Normal-appearing bone marrow of the spine was found in 20 patients (Fig. 1), focal disease was found in 22 patients, homogeneous diffuse involvement was found in 24 patients, combined focal and diffuse infiltration was found in 9 patients, and inhomogeneous/salt-and-pepper pattern was found in 2 patients. Twelve patients had 1–10 nodules and 19 patients had multifocal disease (> 10 nodules, Fig. 2). Twenty-two patients showed moderate diffuse infiltration (Fig. 3) and 10 patients showed strong diffuse infiltration (Fig. 4).
Patients with normal-appearing bone marrow and with an inhomogeneously patchy pattern (salt-and-pepper pattern; Figs. 1, 5) showed significantly better survival when compared with patients with diffuse and focal involvement (P < 0.0001). Focal, diffuse, and combined diffuse and focal involvement were not associated with significantly different survival rates (P > 0.05).
According to the staging system of Durie and Salmon, 32 patients were Stage I, 26 patients were Stage II, and 19 patients were Stage III. According to the MRI staging system, 22 patients were Stage I, 28 patients were Stage II, and 27 patients were Stage III. When we incorporated MRI into the staging system of Durie and Salmon, 18 patients were Stage I, 28 patients were Stage II, and 31 patients were Stage III (Tables 1, 2). According to the standard staging system of Durie and Salmon, without the findings of MRI, 25 patients would have been understaged. Fourteen patients were Durie and Salmon Stage I, but MRI Stage II (n = 13) or Stage III (n =1) due to focal or diffuse bone marrow involvement. Eleven patients were Durie and Salmon Stage II, but MRI Stage III due to strong diffuse or multifocal involvement. A total of 27 patients showed no changes on radiographs, but MRI changes of the bone marrow of the spine were Stage II and III. Eight patients would have been understaged if staging depended only on MRI. Three patients showed no changes in MRI of the spine but osteolysis in radiographs elsewhere. Five patients were staged higher with the standard system of Durie and Salmon than with the MRI staging because of laboratory parameters, such as a decrease in hemoglobin level and a high M-protein component.
Table 1. Patients in the Different Staging Systemsa
Durie and Salmon (1975)
Durie and Salmon including MRI
MRI: magnetic resonance imaging
Number of patients who died is shown in parentheses. Due to the higher sensitivity of detecting focal or diffuse involvement of the spine, MRI increases the number of patients in Stages II and III. Five patients had died in Stage I of the standard system of Durie and Salmon.
n = 32 (5)
n = 22 (1)
n = 18 (0)
n = 26 (16)
n = 28 (13)
n = 28 (11)
n = 19 (15)
n = 27 (22)
n = 31 (25)
Table 2. Clinical Staging System for Multiple Myeloma of Durie and Salmon: Supplemented with MRI
> 10 lesions; spine or marked diffuse infiltrationc
At the endpoint of statistical evaluation, 41 patients were alive and 36 patients had died. Mean survival was 1064 days. The MRI staging system showed significant differences in survival for Stage I versus Stage II (P < 0.001), for Stage II versus Stage III (P < 0.001), and for Stage I versus Stage III (P < 0.0001; Hazards ratio 4.7). According to the MRI staging system, survival was independent of the age of the patients and the Durie and Salmon staging system (Cox regression analysis). The standard staging system of Durie and Salmon resulted in significant differences with respect to survival for Stage I versus Stage II (P < 0.001), for Stage II versus Stage III (P < 0.05), and for Stage I versus Stage III (P < 0.0001; Hazards ratio 3.1). The Durie and Salmon staging system, combined with the MRI staging system, showed the strongest correlation with survival. The differences between Stage I and Stage II were significant at a level of P less than 0.001, between Stage II and Stage III at a level of P less than 0.0001, and between Stage I and Stage III at a level of 0.0001 (Hazards ratio 6.1; Table 2). Median survival with the staging system of Durie and Salmon including MRI was 1802 days (59 months) for Stage II and 469 days (15 months) for Stage III (Fig. 6).
A differentiated prognostic evaluation of patients with multiple myeloma is mandatory because survival rates vary greatly and new therapy approaches can improve response and survival rates.1–9 Several laboratory parameters and biologic factors correlate significantly with survival.10, 11 The most important are β-2-microglobulin (β-2M) and the plasma cell labeling index. β-2M reflects tumor cell burden and the plasma cell labeling index is a measure of plasma cell proliferation. Different cutoff levels have been defined with respect to survival.12, 15, 16 In addition to the staging system of Durie and Salmon, other prognostic staging systems have been proposed. Bataille et al.13 used a combination of β-2M and C-reactive protein (CRP) for a three-grade staging system. A cutoff level of 6 mg/L was used for both CRP and β-2M. Greipp et al.12 replaced CRP with the plasma cell labeling index. Cutoff levels of 1% for the labeling index and 2.7 mg/dL for β-2M divided the patients into three risk groups. The British Medical Research Council18 presented a three-grade staging system based on blood urea concentration, hemoglobin level, and the World Health Organization performance status. Using multivariate analysis, Weber et al.20 found that an M-protein level above 30 g/L, IgA type, and Bence Jones protein level above 50 mg/dL were strong and independent risk factors that predicted early progression. In the staging system of Bladé et al.,19 blood urea concentration and the serum level of albumin were used for staging.
Multiple myeloma is a heterogeneous disease and requires a staging system that takes into account the diversity of this pathology. In large cohorts, the combination of two laboratory parameters may correlate significantly with survival and may be important for dividing patients into subgroups. However, they do not take into account the different aspects of the disease. Laboratory parameters alone are always subject to errors. For example, the CRP level may be increased due to infectious disease and the β-2M level may be increased due to renal failure, which is common in multiple myeloma. The M-protein component may be relatively low in low differentiated plasma cells or it may be missing in patients with Bence Jones myeloma and nonsecretory myeloma. The staging system of Durie and Salmon is still in wide clinical use and it represents a good basis for staging patients because it evaluates many aspects of the disease process. First, the hemoglobin level detects bone marrow failure, which is a sign of active disease. Second, the M-protein level reflects the secretory activity of the plasma cells and hypercalcemia reflects a worse complication of the disease. Third, it takes into account imaging methods. Radiologic imaging methods, like radiographs and computed tomography and MRI scans, are standard in the staging of almost all malignant diseases.
In 1975, only radiographs were available for imaging the skeleton. They remain a good method for surveying the skeletal structure and evaluating the bony structure and fractures. However, in the spine, detection of osteolysis is limited because of the complex anatomy and overlying tissue, such as the abdomen and the rib cage. Only large osteolysis with destruction of pedicles prove tumor involvement. In addition, on radiographs, vertebral fractures are subject to errors because they can also occur due to osteoporosis. The spine is the most affected site of involvement in patients with multiple myeloma.35 In this location, MRI is of great advantage because it can visualize directly bone marrow and bone marrow changes with a high resolution.22 Twenty-seven of our patients showed MRI involvement, but no changes were evident on the radiographs. In several other studies, MRI also proved to be more sensitive than radiographs in the detection of myelomatous infiltration.25–32 In these studies, radiographs resulted in 30–60% false-negative results. Conversely, MRI of the spine cannot replace the skeletal survey because some patients have osteolysis elsewhere. This was the case in three of our patients and in 8 of 80 patients (10%) in the Lecouvet et al.32 series. MRI of the spine generally shows two infiltration patterns, focal or diffuse involvement.22–24 Focal infiltration is detected best by sequences with fat–suppression because normal bone marrow has a low signal intensity and tumor nodules are demonstrated as focal areas of strong hyperintensity36, 37 (Fig. 2). Diffuse infiltration is displayed as a homogeneous decrease of signal intensity, which is detected best on T1-weighted SE images38 (Fig. 3). When there is high-grade diffuse infiltration, the signal intensity of the bone marrow is as low as the intervertebral discs, which contain water (Fig. 4). The intervertebral discs can be relatively brighter in signal, which has been described as the “bright disc sign.”39 The inhomogeneous appearance of the bone marrow, which has also been described as a salt-and-pepper pattern, represents normal bone marrow with only sparse infiltration beside the fat islands.23 These are shown as areas of high signal intensity on T1-weighted SE images and as areas of low signal intensity on STIR images. This appearance should not be misinterpreted as multifocal involvement. The fat-suppression techniques, like STIR in our study, can prove that there are no foci of high signal intensity (Fig. 5). With spine array coils, nearly the entire spine can be covered in one examination. The combination of an unenhanced T1-weighted SE sequence and a fat-suppression sequence, e.g. STIR, is a quick and relatively cheap method. Acquisition time is about 5–10 minutes.
In the current series, patients with normal-appearing bone marrow of the spine showed a significantly better survival rate than patients with diffuse or focal involvement. This is consistent with the findings of other authors.20, 28, 29, 31, 33, 34 In the current study, the focal, diffuse, and combined focal/diffuse infiltration patterns did not show significantly different survival rates among patients. The extent of infiltration is the most important determinant for prognosis.
Spinal involvement in MRI has proven to be a significant determinant for disease progression.28–31, 34 However, until now, no attempt has been made to include MRI in a widely used clinical staging system for myeloma patients. Our three-grade staging system with MRI for myeloma involvement of the spine has proven to be a strong and independent parameter for survival. Because the spine covers large portions of the red bone marrow and it is the major site of involvement in patients with multiple myeloma, the spine is ideally suited for staging the extent of bone marrow involvement. MRI can directly visualize tumor cell burden. On the basis of the standard staging system of Durie and Salmon, 25 of 77 patients (32%) would have been understaged without the MRI findings. This might be the reason why in some other trials, the patients in Durie and Salmon Stages I and II showed no statistical difference in survival.10, 40 Turesson et al. reported the prognostic significance of different staging systems in 583 patients, which included the staging system of Durie and Salmon. Only patients in Durie and Salmon Stage III had a significant shorter survival than patients in Stage I or Stage II. The Finnish leukemia group evaluated the long-term survival of 324 patients. Patients in Durie and Salmon Stages I and II did not show a significant difference in survival rates. Spasov and Goranova41 and Löffler et al.42 found no significant differences in survival rates among myeloma patients in Durie and Salmon Stages II and III. This agrees with our findings. Differences in survival between patients in Durie and Salmon Stages II and III reached statistical significance but at a low level (P = 0.048). The inexact criterion “advanced lytic bone lesions” may contribute to the uncertainties in staging the patients into Stage II or III. The findings of MRI may help in making those decisions.
In conclusion, great variances in survival of myeloma patients and new therapy options imply the need for a differentiated staging of myeloma patients. In former studies, the discrimination of patients in Stage I versus Stage II or Stage II versus Stage III sometimes resulted in no significance with respect to survival. This problem may be due to the false-negative results with X-rays, especially in the spine, where overlying tissue and the rib cage hamper the assessment of osteolysis. In addition, diffuse infiltration may appear as osteoporosis and cannot be distinguished reliably from idiopathic osteoporosis. MRI offers a direct visualization of tumor infiltration and proved to be more sensitive in detecting myeloma infiltration. A differentiated staging of spinal MRI on a three-grade basis was a highly significant determinant for prognosis. The combination of MRI in the staging system of Durie and Salon provided the best results with respect to survival. Therefore, we propose the implementation of MRI in the staging system of Durie and Salmon. In future studies, a larger series of patients and different staging systems inclusive of MRI should be evaluated and compared with staging without the inclusion of MRI.