Multiple myeloma (MM) remains an incurable cancer. Treatment often is initiated at the time patients experience a progressive increase in tumor burden. The authors of this report investigated magnetic resonance imaging of the bone marrow (BM-MRI) as a novel approach to quantify disease burden and validated a staging system by correlating BM-MRI with common clinical and laboratory parameters.
The extent of bone marrow involvement was evaluated by BM-MRI. Clinical and laboratory parameters were assessed in patients with active MM, and correlations between variables were assessed statistically. Bone marrow involvement by BM-MRI was defined as stage A (0%), stage B (<10%), stage C (10%-50%), and stage D (>50%).
In total, 170 consecutive patients were evaluated (77 women and 93 men), including 144 patients who had active MM. The median age was 61 years (age range, 35-83 years). Advance stage disease (stage >I) based on Durie-Salmon (DS) staging or International Staging System (ISS) criteria was observed in 122 patients (84%) and 77 patients (53%), respectively. Lytic bone disease was noted in 120 patients (83%). There was a significant association between BM-MRI involvement and DS stage (P = .0006), ISS stage (P = .0001), the presence of lytic bone disease (P < .0001) and mean β-2 microglobulin levels (P < .0001). Among the patients with previously untreated MM, there was a significant association between BM-MRI stage and overall survival (OS) (univariate P = .013; multivariate P = .045). Plasmacytosis on bone marrow biopsy at diagnosis was not predictive of OS (P = .91).
Multiple myeloma (MM) is the most common malignant plasma cell disorder. In the United States, approximately 20,000 new patients were diagnosed with MM in 2008.1 Clinical characteristics of MM include bone marrow infiltration by malignant plasma cells, aberrant production and accumulation of monoclonal immunoglobulin, lytic bone disease, renal dysfunction, and an overall immune compromised state. Despite the availability of several new therapeutic agents, myeloma remains mostly an incurable disease. The current recommendations to initiate treatment are directed primarily by the extent of disease in the bone marrow2 (the primary site of disease) or the development of end-organ damage (such as renal failure, myelosuppression, and/or lytic bone lesions).3-5 Thus, accurate assessment of the extent of disease remains an important factor in determining need for therapeutic intervention. Because bone marrow is the primary site of disease involvement, histologic evaluation of the bone marrow has remained the primary mode of determining disease extent. An obvious and commonly recognized limitation of this process is sampling error. This is because bone marrow involvement in MM is noncontiguous (disease-involved areas interspersed with areas of normal bone marrow). Levels of the monoclonal proteins in the serum and/or urine also are used to estimate tumor burden in MM. This biochemical approach is of no value in patients with nonsecretory or oligosecretory MM.6, 7 Therefore, novel strategies that can assess disease burden accurately and more predictably will be helpful in clinical decision making.
Radiologic evaluation of the bone marrow using magnetic resonance imaging (BM-MRI) is a novel, noninvasive technique that can provide valuable information regarding the extent of myeloma infiltration in a large volume of bone marrow. MRI can identify abnormalities in the bone marrow with greater sensitivity than radionuclide bone scans, conventional radiography, and computerized tomography (CT).8 Thus, a detailed mapping of bone marrow infiltration can be done by BM-MRI that accurately reflects disease burden and can help in clinical decision making. Previous reports on BM-MRI evaluation have used a subjective classification of bone marrow infiltration with “focal,” “patchy/variegated,” and “diffuse” patterns.9-12 To our knowledge, a more objective BM-MRI examination and staging technique has not been investigated formally in patients with MM, particularly its correlation with other standard clinical parameters of the disease and its correlation with overall survival. In the current study, we prospectively evaluated the role of BM-MRI in patients with MM to objectively quantify the extent of bone marrow infiltration and correlated this with standard clinical and prognostic markers of disease.
MATERIALS AND METHODS
This study was conducted after obtaining approval from the institutional review board. All patients with plasma cell cancers who were seen in the MM clinics at Roswell Park Cancer Institute from January 1, 2004 to June 30, 2007 were included in this analysis. Patients were eligible for this study irrespective of treatment status. Analyses were conducted on all patients together and on subsets of previously treated patients and previously untreated patients separately. All patients with active MM received treatment with regimens that contained novel agents, including proteasome inhibitors or immunomodulatory drugs. None of the patients who were included in the analysis had undergone stem cell transplantation.
The extent of bone marrow involvement was evaluated in each patients by BM-MRI. Imaging was conducted on a 1.5 Tesla GE MRI scanner (GE Healthcare [formerly GE Medical Systems], Waukesha, Wis) using an 8-channel GE cervical-thoracic-lumbar spine coil. The imaging technique included sagittal spin echo T1 and fast spin echo inversion recovery (minimum echo time) sequences of the cervical, thoracic, and lumbosacral spine. Coronal T1 and fast spin echo inversion recovery sequences of the sacrum, pelvic bones, and proximal femurs were also performed. No intravenous contrast medium was administered to the patients. The areas of active red bone marrow that were not evaluated by this technique included the sternum and ribs, because these areas do not constitute a significant fraction of active bone marrow in adults compared with the other imaged areas. The imaging sequences described above are performed routinely at our institute and did not require any special equipment or hardware for the current study. Clinical staging for the MM patients was performed according to the Durie-Salmon (DS) staging system2 and the International Staging System (ISS).13 Skeletal radiographs were used to assess presence of lytic bone lesions for all patients.
A predefined scoring system (Table 1) was used to prospectively evaluate the extent of bone marrow infiltration with MM. The extent of bone marrow involvement by disease was reported as a percentage of the total extent of bone marrow assessed for each patient. A semiquantitative system was used for the purpose of this evaluation. Patient-specific bone marrow involvement patterns (confluent and/or solitary or multiple focal lesions with a normal bone marrow background) were assessed, and the percentage involvement was calculated taking into account the total amount of bone marrow assessed by BM-MRI. Thus, for each patient, a BM-MRI percentage score was calculated as follows:
On the basis of the percentage score, a BM-MRI stage (see Table 1) was assigned to each patient. Because the BM-MRI stages spanned a range of bone marrow infiltration extent rather than an exact number, this semiquantitative method was deemed reproducible. Bone marrow infiltration lesions with a greatest axial dimension ≥0.5 cm were included in the final BM-MRI assessment. Representative BM-MRI stage images are shown in Figure 1. Two expert radiologists were involved in reading the BM-MRI studies (A.N.A. and R.A.A.) and were blinded to the results of the patient's clinical tests, including results of the bone marrow trephine biopsy (BM-Bx).
Table 1. Magnetic Resonance Imaging of the Bone Marrow Staging System
Bone Marrow Involvement, %
BM-MRI indicates magnetic resonance imaging of the bone marrow.
Clinical, Radiologic, and Pathologic Parameters
Patient demographics, disease staging (both DS and ISS), skeletal bone survey for lytic lesions, laboratory biochemical markers, and bone marrow biopsy specimens were included in the analyses for patients with active MM as long as they were recorded within 4 weeks of the BM-MRI procedure.
To study statistical correlations between pairs of ordinal variables, a test corresponding to the Spearman correlation was used, and the statistical significance of correlations between nominal and ordinal variables was determined by using the Wilcoxon or Kruskal-Wallis test. Statistical assessment of observed differences in the survival distributions was performed using the log-rank test in conjunction with a Bonferroni adjustment for multiple comparisons. A Cox proportional hazards model was used to calculate the hazards of death for previously untreated patients who had stage C or D disease according to the BM-MRI staging system compared with patients who had stage A or B disease, first in a univariate analysis and then in a multivariate analysis that incorporated ISS stage and the presence of lytic lesions for each patient. A Cox proportional hazards model was also used to test the association between the percentage of plasmacytosis on BM-Bx in previously untreated patients and overall survival. A .05 nominal significance level was used in all hypothesis testing.
One hundred seventy consecutive patients with plasma cell disorders who had BM-MRI performed at our institute were evaluable for this analysis. Of these, 90 patients (53%) had previously untreated disease. Demographic and disease-specific characteristics of these patients are summarized in Table 2. The median age of enrolled patients was 61 years (range, 35-83 years). Among all patients with plasma cell disorders, 144 patients had active MM and were included in the analyses for BM-MRI staging with clinical parameters. The incidence of patients who had DS stage I, II, and III disease was 12.9% (n = 22), 14.1% (n = 24), and 57.6% (n = 98), respectively. According to BM-MRI staging criteria, the incidence of stage A, B, C, and D disease was 15% (n = 22), 22% (n = 31), 27% (n = 39), and 36% (n = 52), respectively.
Table 2. Clinical Characteristics of Patients at the Time of Bone Marrow Magnetic Resonance Imaging Evaluation
DS indicates Durie-Salmon; ISS, International Staging System; Ig, immunoglobulin.
Included patients who had a diagnosis of monoclonal gammopathy of unknown significance, smoldering myeloma, plasma cell leukemia, or plasmacytoma.
Association Between BM-MRI Stage and Clinical Stage
Statistical analyses evaluating the dependence of the extent of disease defined by BM-MRI stage and various clinical and prognostic markers were completed as described above. First, we investigated whether there was any correlation between the radiologic stage as determined by BM-MRI stage and DS stage. For this analysis, only patients with active MM (n = 144) were selected. We noted a significant correlation between BM-MRI stage and DS stage (P = .0006) (Fig. 2 Top). It is interesting to note that this correlation between clinical stage and BM-MRI stage remained consistently significant even with the newer ISS, which incorporates different staging parameters (P = .0001) (Fig. 2 Bottom). This analysis also was restricted only to patients with active MM (n = 144).
Association Between BM-MRI Stage and Other Clinical Features
Next, we investigated whether there was any association between BM-MRI stage and other clinical parameters of disease, such as β-2–microglobulin (β-2M), lytic bone disease, immunoglobulin subtype, and patient age. Among patients who had no lytic bone disease on skeletal surveys, 52% had positive involvement on BM-MRI studies. The β-2M level at the time of MRI evaluation was correlated significantly with BM-MRI stage (P < .0001) (Fig. 3 Top). Furthermore, the presence of lytic lesions assessed by plain radiographic skeletal survey also was found to have a significant correlation with BM-MRI stage (P < .0001) (Fig. 3 Bottom). We did not observe any association between BM-MRI stage and patient age (P = .792) or immunoglobulin subtype (P = .812).
Association Between BM-MRI Stage and Bone Marrow Biopsy
All patients who had undergone a BM-Bx within 4 weeks of BM-MRI were included in this analysis. Fifty patients were qualified, including 23 women (46%) and 27 men (54%). All 50 patients had previously untreated MM. Evidence of bone marrow involvement on BM-Bx and on BM-MRI was concordant in 23 patients (46%) and discordant in 27 patients (54%). Of the patients who had discordant results, 89% had more extensive BM involvement detected by BM-MRI, and 11% were found to have a higher reading on BM-Bx. The estimated kappa statistic for the measurement of disease extent by BM-MRI and BM-Bx in these 50 patients was 0.2876 (95% confidence interval [95% CI], 0.1238-0.4514), indicating that there was not much agreement between the 2 techniques in assessing disease burden. Furthermore, there was no significant association noted between the percentage of plasmacytosis on BM-Bx in these patients and the clinical MM stage (P = .27).
Association Between BM-MRI Stage and Overall Survival
Finally, we investigated whether the extent of disease determined by BM-MRI was predictive of survival. For this analysis, all previously untreated eligible patients (n = 90) were evaluated. This was done to get a more homogenous population for the survival analysis. Their overall survival and BM-MRI involvement stage (determined by a BM-MRI obtained within 4 weeks of the date of diagnosis) were studied.
The analysis revealed that there was a significant difference in survival distribution between various BM-MRI stages (P = .01) (Fig. 4 Top). Pair-wise comparisons between different BM-MRI stages for this analysis did not reveal any significant differences. This suggests that BM-MRI stage can be used reliably for prognostication at the time of initial diagnosis. For the purpose of calculating hazards ratios for death, patients in BM-MRI stage A or B were compared with patients in BM-MRI stage C or D. Log-rank testing revealed that there was a significant difference in the overall survival distributions for BM-MRI stage A or B and BM-MRI stage C or D (P = .001) (Fig. 4 Bottom).
On univariate analysis, the hazards of death for patients with BM-MRI stage C or D disease was significantly higher than the hazards of death for patients with stage A or B disease (P = .013). The estimated hazards of death for patients who had BM-MRI stage C or D disease was 13.12 times (95% CI, 1.69-101.92 times) the estimated hazards of death for patients who had stage A or B disease. Furthermore, a multivariate analysis that incorporated BM-MRI stage, ISS stage, and the presence of lytic bone lesions revealed that this trend remained consistent. The hazards of death for patients who had BM-MRI stage C or D disease was found to be significantly higher than the hazards of death for patients who had BM-MRI stage A or B disease (P = .0454). The estimated hazards of death for patients who had BM-MRI stage C or D disease was 8.41 times (95% CI, 1.05-67.7 times) the estimated hazards of death for patients who had BM-MRI stage A or B disease when ISS stage and the presence of lytic lesions were analyzed along with BM-MRI stage.
Furthermore, we tested the association of the percentage plasmacytosis on BM-Bx (n = 50) with overall survival in patients who had previously untreated MM, and there was no significant association noted (P = .91). This suggested that BM-MRI staging, and not the percentage of BM-Bx plasmacytosis at the time of diagnosis, was predictive of overall survival in patients with MM. Results of these survival analyses are summarized in Table 3.
Table 3. Overall Survival Analysis Using the Cox Proportional Hazards Model
HR indicates hazards ratio; 95% CI, 95% confidence interval; BM-MRI, magnetic resonance imaging of the bone marrow; BM, bone marrow; ISS, International Staging System.
BM-MRI: Stage C/D vs A/B
Percentage plasmacytosis on BM biopsy
BM-MRI: Stage C/D vs A/B
Presence of lytic lesions
Extent of disease in the bone marrow is an important factor that determines the need for therapeutic intervention in patients with MM.2, 5, 13, 14 MRI is an important tool that can identify areas and patterns of disease involvement in patients with MM and is being used increasingly in the workup of these patients.15, 16 In this context, the primary use of MRI to date has been to assess bone disease, extraosseous metastases, or paraspinal tumors to look for cord impingement.16-18 Despite its increased use, to our knowledge there is no defined system to incorporate bone marrow imaging for routine patient evaluation and staging. The imaging techniques are not only subjective but also have not been validated uniformly with other established clinical or laboratory markers of disease. Infiltration of the bone marrow by myeloma has been described previously as broad categories of focal, patchy/variegated, or diffuse infiltration.16-22 The current study describes a semiquantitative way to assess bone marrow infiltration by MRI that may be more reproducible.
The results of the current study indicate that the extent of disease in patients with MM can be determined accurately by MRI and that this correlates with bona fide clinical and prognostic factors of this disease. The DS staging system represents disease burden measurement, whereas the ISS adds prognostic information to disease staging. Correlation of our proposed scoring system with both of these staging systems suggests their overall utility with the BM-MRI technique and the ability to assess disease burden more accurately. A previously reported study has addressed the sensitivity of BM-MRI in detecting focal bone marrow lesions and its value in predicting patient prognosis compared with metastatic bone surveys.16 Although that report studied a large patient cohort, it compared patients who had >7 focal lesions with patients who had <7 lesions without addressing the continuum of bone marrow infiltration, which can range from a solitary lesion to diffuse bone marrow involvement. Furthermore, to our knowledge, there have not been any data published to date regarding the importance of the size of individual focal lesions versus their number. Thus, merely enumerating the focal lesions may not provide a true measure of disease burden, because it would not quantify the extent of bone marrow infiltration. Our proposed scoring system addresses this and validates the staging of bone marrow infiltration as classified by BM-MRI. There was a small subset of patients in our study (6%) who had stage A bone marrow infiltration according to the BM-MRI score but had clinically advanced stage (stage >I) disease. This may be because of some heterogeneity in the appearance of active red bone marrow on BM-MRI studies. Despite this small variation, the overall statistical associations, including the survival analyses, were valid. This suggests that the proposed BM-MRI technique and the scoring system can be used as a valid tool for patients with active MM despite minor variations. In contrast to previously used terms such as focal, multiple focal, patchy, diffuse etc,9-12, 16 the proposed system determines BM-MRI stages based on the percentage involvement and therefore potentially reduces interobserver and interpopulation variability during interpretation of the BM-MRI. This scoring system likely will add to the currently used DS and ISS staging systems by assessing bone marrow infiltration in patients with MM more accurately, thus aiding better clinical decisions.
In a subset of our patients, all of whom had previously untreated, active MM, we demonstrated that BM-Bx may not be an accurate measure of the extent of bone marrow infiltration, because it may underestimate the extent of disease burden. Plasmacytosis on BM-Bx did not appear to have a significant association with disease stage, suggesting that plasmacytosis on a random BM-Bx may not be a good representation of actual disease burden. The measurement of bone marrow disease burden with the 2 techniques of BM-Bx and BM-MRI was discordant in a substantial number of patients. The BM-MRI scoring estimates bone marrow infiltration based on a larger sample of the active bone marrow and correlated well with various clinical and prognostic markers of disease, thus providing a better measure of disease extent.
Furthermore, in the current study data, BM-MRI stages were found to be independent predictors of differences in survival for patients with MM at the time of diagnosis with active disease. We were able to demonstrate that patients with BM-MRI stage C or D disease have a significantly greater hazards of death than patients with BM-MRI stage A or B disease. This analysis remained significant whether BM-MRI staging was used as the only variable (univariate) or was combined with ISS stage and the presence of lytic bone lesions (multivariate) in a Cox proportional hazards model. These results support the finding that the noninvasive imaging technique of BM-MRI and the proposed scoring system used in this study complement other parameters that are used currently to correctly define the extent of bone marrow infiltration by MM and that, by itself, can predict overall survival from the time of diagnosis. We conducted a similar analysis of the percentage plasmacytosis on BM-Bx in patients with previously untreated MM who underwent the BM-Bx within 4 weeks of the BM-MRI. The BM-Bx was not found to be correlated significantly with overall survival, thus indicating that the BM-MRI stage at diagnosis was superior to BM-Bx in predicting patient survival.
Clinical heterogeneity among different patients with similar MM stages is noted commonly and may reflect biologic differences within the disease entity. Although cytogenetic analysis and gene expression profiling have been able to identify patient subsets with an adverse prognosis,23-28 there currently is no consensus regarding the most optimal time of therapeutic intervention for these patients. The extent of tumor burden is the most reliable clinical parameter that directs the need for therapeutic intervention. Thus, an accurate estimate of the extent of bone marrow involvement in patients with MM is necessary. The noncontiguous nature of the disease limits accurate and dependable estimation with BM-Bx, as discussed earlier. BM-MRI helps evaluate large volumes of bone marrow to accurately assess the extent of bone marrow infiltration by MM.
Results from our investigation demonstrate that the novel BM-MRI staging system presented herein is accurate for predicting the extent of disease burden, is more reproducible than the currently used terminology for BM-MRI reporting, correlates well with currently used clinical and laboratory parameters for the assessment of disease, and correlates with survival outcome in patients with myeloma. Larger prospective studies are warranted for the further validation of BM-MRI staging in patients with MM, including its correlation with cytogenetic and molecular markers of disease.