Venous thromboembolic disease (VTD) is a recently recognized complication of thalidomide in combination therapy for patients with multiple myeloma (MM). The authors assessed the frequency of VTD and its risk factors in 612 consecutive patients with plasma cell dyscrasia (PCD) who were evaluated and followed from 1991 to 2001.
In the current study, 404 patients were diagnosed with multiple myeloma (MM), 174 with monoclonal gammopathy of undetermined significance (MGUS), and 34 with other forms (excluding amyloidosis). Univariable correlates of VTD were assessed using Kaplan–Meier analysis and Cox proportional hazards analysis.
The authors identified several univariable correlates of VTD in patients with MGUS, including a family and medical history of VTD, immobility, low serum albumin level, and high leukocyte count. Patients with MGUS with immunoglobulin (Ig) G monoclonal immunoglobulin were found to be less prone to develop VTD. In patients with MM, a family and medical history of VTD and the presence of a hypercoagulable state were factors identified in univariable analysis to be associated with an increased risk of VTD. In patients with MM, for each unit increase in serum albumin, the risk of VTD was lower. The type of the treatment regimen did not appear to correlate with the development of VTD.
Venous thromboembolism consists of two interrelated conditions that usually occur during serious illnesses or after major surgeries, namely, venous thromboembolic disease (VTD) and pulmonary embolism (PE). Venous thromboembolism is rare in the general population (odds ratio = 1:2000).1 Malignancy is a prominent risk factor for VTD. In a community-wide survey, Anderson and Wheeler2 reported that 22.3% of patients with a diagnosis of acute VTD had a malignant disease. Other authors reported that the incidence of VTD in this patient population ranged from 1.5% to 25%.3, 4 The incidence of cancer among patients with recurrent VTD was found to be higher (17.1%) than the incidence in patients with secondary and nonrecurrent thrombosis (1.9% and 7.6%, respectively).5 The frequency of postoperative VTD increases two to three times in patients undergoing surgery for malignant disease compared with the same procedure in patients with nonmalignant conditions. Patients with active multiple myeloma (MM) requiring therapy have a 10% chance of developing a VTD during the induction phase of therapy.6 Other investigators reported a higher incidence of VTD in patients with MM who were treated with thalidomide as a single agent or a combination with steroids or chemotherapeutic agents.7–12 To our knowledge, the exact etiology of hypercoagulability in patients with MM and its relation to therapy and the disease is not clear. We initiated a retrospective study to evaluate the frequency of and risk factors for VTD in a series of patients with plasma cell dyscrasia (PCD) identified at the Cleveland Clinic Foundation (CCF; Cleveland, OH) over a 10-year period.
MATERIALS AND METHODS
Methodology of Medical Record Identification: Review and Limitations
We reviewed the complete medical records of all patients with monoclonal gammopathy of undetermined significance (MGUS), MM, smoldering/indolent MM (SMM/IMM), solitary plasmacytoma (SPC), or plasma cell leukemia (PCL) who were prospectively entered in the PCD database. We used the criteria published by the international myeloma working group to define the patients with PCD.13 Patients who were diagnosed with MGUS before a bone marrow trephine biopsy was performed were included if they did not have evidence of end organ damage. When those patients were reevaluated at our institution and the bone marrow showed > 10% plasma cells, we continued to consider them as MGUS provided no end-stage organ damage or therapy was required. Patients were evaluated and followed by staff oncologists at the CCF between 1991 and 2001.
The CCF is an urban tertiary referral center and a teaching hospital. Our proposal was submitted to the CCF institutional review board and we were granted approval based on an expedited review. The requirement for obtaining informed consent was waived based on the study designs (standard medical record-based review) and patient anonymity.
The computerized medical records of patients with PCD were initially reviewed by one of the coinvestigators (M.G.C.). An additional review was performed independently by a second coinvestigator (G.S.). If electronic data were incomplete, the hard copies of charts also were reviewed. When necessary, missing information was sought from referring providers. All but five patients with VTD of the extremities were diagnosed by Duplex ultrasound. These five patients were diagnosed outside of our facility and we could not confirm the type of imaging technique used. Cases of PE were diagnosed by high-probability V/Q scan, contrast computed tomography (CT) scan of the chest, or by angiography.
The term “immobility” was used for patients who were hospitalized or bedridden for > 7 days at the time of VTD diagnosis. A personal history of VTD was defined as a history of documented thromboembolism before the diagnosis of PCD. Patients were screened for a hypercoagulable state only when they developed what appeared to be a vascular event that needed further investigation.
Data were collected in five categories: demographics, disease and treatment, (thrombosis case information, major risk factors for thrombosis, and baseline laboratory data. Data were manually recorded onto researcher-developed data collection forms and subsequently entered into a computerized database.
Categoric variables are summarized as frequencies and percentages and continuous variables as the median values and ranges. Time to the development of a VTD was estimated using the Kaplan–Meier method. Follow-up was calculated from the date of diagnosis to the date of either the first VTD or the date of final follow-up visit. Cox proportional hazards analysis was used to identify univariable correlates of thromboembolic events. If Cox analysis could not be performed because of the lack of variability, then the log-rank test was performed instead.
Stepwise Cox proportional hazards analysis was used to identify multivariable correlates of VTD among patients with MM. A P value < 0.10 was used to allow variables to enter the model, whereas a P value < 0.05 was used to retain variables in the final model. Multivariable analysis could not be performed for patients with MGUS because of the insufficient number of events (VTD). All statistical analyses were two sided. A P value < 0.05 was used to indicate statistical significance. Analyses were performed using SAS 6.12 software (SAS Institute, Inc., Cary, NC).
Data were available for 612 patients (Table 1), comprising 346 men (56%) and 266 women (44%). Of these patients, 404 patients (66%) were diagnosed with MM and 174 (28%) were diagnosed with MGUS. Of the remaining 34 patients (6%), 14, 18, and 2 were diagnosed with SMM/IMM, SPC, or PCL, respectively. The serum creatinine level was elevated in 20% of patients with MM and in 22% of patients with MGUS. Corrected calcium level was elevated (> 10.5 mg/dL) in 8% of the patients (11% with MM and 3% with MGUS). As expected, the serum albumin level was low (< 3.5 g/dL) in 31% of patients with MM and in 11% of patients with MGUS. In addition, 77% of patients with MM and 53% of patients with MGUS had high levels of β-2-microglobulin (β-2M) (> 1.9).
Table 1. Clinical Characteristics of Patients with Plasma Cell Dyscrasia at the Time of Presentation (n = 612)
MM: multiple myeloma; MGUS: monoclonal gammopathy of undetermined significance; BMPC: bone marrow plasma cells; Ig: immunoglobulin; VTD: venous thromboembolic disease.
Smoldering and indolent multiple myeloma, solitary plasmacytoma, plasma cell leukemia.
b Cleveland Clinic Foundation laboratory reference range, normal values: serum M-protein: 0 g/dL; urine protein: 0–0.15 g/24, β-2-macroglobulin : 0.3–1.9 mg/dL; and serum albumin: 3.5–5.0 g/L.
Median age (yrs) (range)
Serum M-protein (g/L)2
(n = 314)
(n = 131)
(n = 20)
24-hr urine protein (g)2
(n = 296)
(n = 117)
(n = 24)
(n = 365)
(n = 160)
(n = 32)
(n = 362)
(n = 159)
(n = 29)
Serum albumin (g/L)2
(n = 384)
(n = 171)
(n = 32)
Immunoglobulin type (%)
(n = 320)
(n = 163)
(n = 24)
(n = 397)
(n = 174)
(n = 28)
History of VTD
Of 612 patients with PCD, 53 developed VTD, 22 (4%) patients had a personal history of thromboembolism, and 4 (< 1%) had a family history of thromboembolism. Five of the 53 patients with VTD were diagnosed with a hypercoagulable state (2 patients with lupus anticoagulants, 2 patients with antithrombin deficiency, and 1 patient with anticardiolipin).
Patients with MM were staged by the Southwest Oncology Group (SWOG) staging system.14 Of the 361 evaluable patients with meloma, 42%, 34%, 19%, and 5% had Stage I, II, III, and IV disease, respectively. Sixty-six percent of patients with MM had recurrrent/refractory disease and 34% had newly diagnosed disease at the time of evaluation at CCF. Bone survey was positive for lytic lesions in 72% of patients with MM (43% with < 3 lesions and 57% with ≥ 3 lesions). Patients with MM were treated with multiple regimens that included radiotherapy (40%), steroids (94%), vincristine (61%), melphalan (60%), cyclophosphamide (38%), doxorubicin (36%), liposomal doxorubicin (23%), thalidomide (20%), and rituximab (11%). Fewer than 10% of the patients with myeloma received ≥ 1 of the following agents: interferon-α, busulfan, interleukin-2, PEG-L-asparaginase, etoposide, cisplatin, temozolomide, arsenic, methotrexate, trans-retinoic acid, carmustine, cytosine arabinoside, fludarabine, chlorambucil, taxotere, and bortezomib. Fifteen patients with MM were treated with high-dose chemotherapy followed by peripheral stem cell support).
Patients with VTD
VTD was diagnosed by Duplex ultrasound. We could not confirm the type of imaging technique used for five patients. Four patients had atypical venous thromboembolism. In two patients, bilateral renal VTD and multiple pelvic thromboses were confirmed by abdominal contrast CT scan, right atrial thrombosis was diagnosed by echocardiogram in one patient, and ocular venous thrombosis was diagnosed by ophthalmologic examination. We identified nine patients with PE, six of whom were diagnosed by high-probability V/Q scan, two by contrast CT scan of the chest, and one by angiography.
Of the 612 patients in the current study, 53 (29 men and 24 women) (9%) had ≥ 1 episodes of symptomatic, objectively documented VTD after the diagnosis of PCD (Table 2). None of the patients with SMM/IMM, SPC, or PCL developed venous thromboembolism. The incidence of VTD in patients with MM was 10% (40 of 404 patients [10 patients with newly diagnosed disease and 30 patients with recurrent/refractory disease]) and was 7.5% (13 of 174 patients) in patients with MGUS.
Table 2. Clinical Characteristics of Patients with Plasma Cell Dyscrasia Venous Thrombosis
MM (n = 40) (%)
MGUS (n = 13) (%)
MM: multiple myeloma; MGUS: monoclonal gammopathy of undetermined significance; VTD: venous thromboembolic disease; BMPC: bone marrow plasma cells; Ig: immunoglobulin.
The median times from the diagnosis of PCD to development of VTD in patients with MM and MGUS were 8.5 months (range, 0–87 months) and 4 months (range, 0-67 months), respectively. The median time from treatment to VTD in patients with MM was 1 month (range, 0–8 month). Of the 53 patients diagnosed with VTD after the diagnosis of MM or MGUS, 7 (15%) had a medical history of VTD and 2 (4%) had a family history of thromboembolism (Table 2). One of 53 patients (2%) was previously diagnosed with factor V Leiden mutation. Four of 53 patients (7.5%) were smokers and 11% received exogenous estrogens during the course of PCD. Three patients (6%) with VTD had prolonged immobility and 17 (32%) had in-dwelling catheters at some point during their disease. For nine (17%) patients, the development of VTD was believed to be directly associated with in-dwelling catheters. Nonplasma cell-associated concomitant neoplasms were identified in 2 of 53 patients with VTD (4%). Perioperative VTD was observed in six patients. Although patients with newly diagnosed MM had a higher risk of VTD compared with patients with recurrent/refractory MM (hazard ratio [HR] = −1.25), this did not reach a level of statistical significance (P = 0.55). The Kaplan–Meier method was used to estimate the percentage of patients with MM and MGUS who developed VTD (Fig. 1).
Univariable correlates of VTD in patients with MM and MGUS are presented in Table 3. Many factors were considered to be potential correlates of VTD including gender, age at diagnosis, immunoglobulin (Ig) type, a personal and family history of VTD, hypercoagulable state, non–plasma cell-related concomitant neoplasms, smoking, immobility, in-dwelling catheters, estrogen treatment, percentage of plasma cells in the bone marrow biopsy specimen, creatinine level, corrected serum calcium level, serum albumin level, leukocyte and platelet count, hemoglobin level, β-2M level, serum M protein level, and 24-hour urine protein level, as well as individual agents and combinations of chemotherapeutic agents when appropriate. In patients with MGUS, univariable analysis identified personal (HR = 8.95 [P = 0.002]; 95% confidence interval [95% CI], 2.28–35.10) and family (HR = 13.79 [P = 0.014]; 95% CI, 1.69–112.8) history of VTD as significant risk factors for thromboembolism. In addition, immobility (HR = 27.71 [P < 0.001]; 95% CI, 5.41–141.9), decrease in serum albumin level (HR = 4.16 [P = 0.001]; 95% CI, 1.75–10), and increase in leukocyte count (HR = 1.2 [P < 0.001]; 95% CI, 1.08–1.33) were univariable correlates of venous thromboembolism in patients with MGUS. The presence of IgG heavy chain appeared to have a protective role (HR = 0.19 [P = 0.017]; 95%CI,0.05–0.74) in patients with MGUS. In patients with MM, medical history of VTD (HR = 3.93 [P = 0.01]; 95% CI, 1.39–11.1), family history of VTD (HR = 9.59 [P = 0.027]; 95% CI, 1.29–71.20), and known hypercoagulable state (HR = 11.08 [P = 0.019]; 95% CI, 1.49–82.28) were identified as univariable correlates of VTD.
Table 3. Identifying Univariable Correlates of VTD in Patients with MM and MGUS
Among patients with MM, only medical history of VTD (HR = 4.04; P = 0.008) and hypercoagulable state (HR = 12.18; P = 0.015) were found to be independent risk factors for VTD. Two noteworthy observations were made. First, patients who received melphalan and prednisone (MP) appeared to have a lower incidence of VTD (HR = 0.602), but this did not reach statistical significance (P = 0.13). In this group of patients, using a multivariable model that contained MP, β-2M level, creatinine level, and albumin level, each one unit increase in the serum albumin level was associated with a significant decrease in the incidence of VTD (HR = 0.615; P = 0.039). Second, patients who received thalidomide-based regimens did not appear to differ relative to their risk for VTD when compared with patients who received nonthalidomide-containing regimens (Fig. 2).
VTD has been identified recently as a major complication of myeloma therapy.6, 15 We decided to review our PCD data system to assess the incidence of VTD in a large number of patients with PCD in general and myeloma in particular. We also were interested in exploring whether any correlation exists between disease-related factors, different therapeutic agents, and combination regimens.
The current study has the traditional limitations associated with a retrospective review and analysis. The statistical analysis is limited by the finding that several of the variables that were considered to be correlates of VTD were not baseline characteristics. Ideally, these variables would be analyzed as time-dependent covariates. These variables occurred before development of VTD or before the last follow-up among patients who did not develop VTD, but the exact timing is unknown.
We were surprised to discover that 7.5% (13 of 174 patients) of patients with MGUS developed VTD at a median of 4 months (range, 0–67 months) after diagnosis. Similar to findings in the general population, we identified medical history of VTD (P = 0.002), family history (P = 0.014) of VTD, and immobility (P < 0.001) to be significant risk factors for patients with MGUS to develop a thromboembolic event. In addition, a low serum albumin level (P = 0.001) and an increase in leukocyte count (P < 0.001) were found to be correlates for increased incidence of venous thromboembolism in patients with MGUS. The effects of interleukin-6 (IL-6) could explain the correlation between the increased leukocyte count and decreased albumin level with the amplified incidence of thromboembolism in patients with PCD. IL-6 is a potent mediator of inflammatory processes, and it has been proposed that the age-associated increase in IL-6 accounts for certain phenotypic changes associated with advanced age, particularly those that resemble chronic inflammatory disease, such as decreased lean body mass, osteopenia, low-grade anemia, and decreased serum albumin level.16 For this reason, serum albumin level is utilized as part of the new staging system developed by SWOG.14
In addition to being a major growth and survival factor for malignant plasma cells,17 IL-6 has been shown to cause leukocytosis,18–20 and to promote coagulation, without affecting fibrinolysis.21 It activates the coagulation cascade through tissue factor stimulation and increased transcription of factor VIII, up-regulates transcription of fibrinogen, increases von Willebrand factor, and decreases protein S.22 The increased leukocyte count and decreased albumin level in the population of patients with MGUS in the current study could represent a surrogate marker for IL-6 activation and its procoagulant activity. The most likely explanation for the absence of increased leukocyte count as an influencing factor in patients with MM is the negative impact of MM and chemotherapy on the bone marrow.
The current study finding of a lower incidence of VTD in IgG patients with MGUS is well correlated with observations by the Mayo Clinic group. In a study by Kyle et al.,21 IgA and IgM type were found to be associated with a higher incidence of malignant transformation when compared with IgG type, most likely marking a more aggressive disease.
As expected, a strong correlation between some of the established risk factors for VTD (medical history, familial history, and immobility) and thromboembolic events was observed in patients with PCD. With a median age of 65 years, comorbid conditions and frequent skeletal involvement, which result in decreased mobility, are a considerable factor in this patient population.
In the current analysis, there was no correlation noted between the type of antimyeloma treatment and the rate of VTD. Thalidomide, either used as a single agent or in combination with nonanthracycline chemotherapeutic agents or steroids, was not identified as a risk factor. All our patients had treatment with thalidomide initiated at a low dose (50 mg/day). The dose was increased slowly (50 mg/day every week) until the maximum tolerated dose was reached (400 mg/day). Most of the patients received a dose of thalidomide in the range of 150–300 mg/day.23 When thalidomide was used in combination with steroids (54 of 82 patients), the schedule of dexamethasone, which was the main steroid incorporated in therapy, was used less frequently compared with in other studies. We have traditionally used dexamethasone at a dose of 40 mg a day for 4 days followed by a 10-day rest, after which the schedule was repeated, rather than the 4-day rest. Overall, 82 of the current study patients received a thalidomide-containing regimen. Twenty-eight patients (34%) were treated with thalidomide as a single agent, whereas 54 patients (66%) received thalidomide in combination with dexamethasone-based regimens. In the study by Zangari et al.,24 the combination of thalidomide with chemotherapy that included anthracyclines was found to be an independent predictor for VTD. These patients initially received 200 mg of thalidomide, which rapidly escalated to higher doses. The initial high-dose of thalidomide and the fast escalating schedule would explain the higher incidence of VTD noted among patients in different studies who received thalidomide as a single agent or steroids. Moreover, the combination with anthracyclines, as reported by Zangari et al., is another important factor.
Several different mechanisms were previously described as possible causative factors for VTD in patients with MM. High levels of IL-6 are associated with coagulation cascade activity. The presence of abnormal levels of Ig alters the network structure of gels formed from purified fibrinogen.25 The fibrin fibers that compose these gels have been shown by electron microscopy to be much thinner than normal.26, 27 Plasma clots comprised of thin, small-diameter fibrin fibers are found to be relatively resistant to plasmin degradation.28 Recently, acquired activated protein C resistance (APC-R) was recognized to be common in patients with cancer.29 Zangari et al.30 reported that the incidence of acquired APC-R, in the absence of factor V Leiden mutation, was relatively high (23%) in 62 patients with newly diagnosed MM. However, this event was not confirmed by our group in a larger patient population that had developed thromboembolic events. Using the same methodology as Zangari et al. (i.e., a second-generation APC-R assay), we could not identify any patient with acquired APC–R in the 22 patients with PCD with VTD.31 Four of our 22 patients (19%) were found to be heterozygous for factor V Leiden, which is comparable with findings in consecutive Dutch outpatients with thrombosis.32
Venous thromboembolism appears to be correlated with the duration of disease. In a cohort of 535 patients with MM treated with thalidomide with or without cytotoxic chemotherapy, Zangari et al.24 reported that newly diagnosed disease was an independent predictor of venous thromboembolism. However, exposure bias cannot be completely excluded as a cause for this phenomenon. In the current study, exposure to additional therapy appears to reduce the future risk of developing VTD by 20% (HR = 0.80; 95% CI, 0.38–1.67). However, the differences were not found to be statistically significant (P = 0.55). Larger prospective trials (with a higher number of events) will be needed to confirm or reject this suggestion.
We noted an increased incidence of VTD among patients with MGUS. The risk of this event appears to be directly related to the increase in the leukocyte count and the decrease in the serum albumin level. The known risk factors for the general population (namely, family history, history of hypercoagulation, and immobility) were found to be independent factors that placed patients with MGUS and MM at further risk for VTD. The different laboratory correlates noted for the patients at risk represent circumstantial evidence that the microenvironment cytokine system associated with PCD is involved in the pathophysiology of VTD. Careful attention to the supportive care of the patients is a critical issue in the outcome of therapy for patients with myeloma and MGUS.
Additional studies are needed to further define the different pathophysiologic processes involved. It is most likely that they will be as complex as the pathophysiology of the disease.
The authors thank Ms. Catherine Brand and Ms. Julie Manieri for their editorial efforts.