|Home | About | Journals | Submit | Contact Us | Français|
Serum protein electrophoresis (SPEP) is often obtained at the initial evaluation of a radiolucent bone lesion of unknown etiology. The results are considered convincing evidence of the presence or absence of a plasma cell neoplasm. The sensitivity and specificity of the SPEP have not been reported in this clinical scenario. Our purpose is to assess the diagnostic value of the SPEP in the initial work-up of the radiolucent bone lesion. We identified 182 patients undergoing evaluation of a radiolucent bone lesion that included tissue biopsy and an SPEP value. We then calculated the sen-sitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of SPEP as a diagnostic test for a plasma cell neo-plasm in this clinical scenario. Forty-six of 182 (25.3%) patients in our series were diagnosed with a plasma cell neo-plasm by histopathologic analysis. The sensitivity of SPEP was 71% and the specificity was 83%. PPV was 47% and NPV was 94%. When analyzing only those presenting with multiple lesions, the percentage of patients diag-nosed with multiple myeloma increased to 44.7% (34 of 76 patients). The SPEP, however, did not have a substantially increased diagnostic accuracy with sensitivity of 71%, specificity 79%, PPV 40% and NPV 93%. SPEP lacks sensitivity and positive predictive value to provide a definitive diagnosis of myeloma in radiolucent bone lesions, but has a high negative predictive value which may make it useful in ruling out the disease. We recommend that this test either be performed in conjunction with urine electrophoresis, immunofixation electro-phoresis and free light chain assay, or after biopsy confirming the diagnosis of myeloma.
Patients presenting with a radiolucent bone lesion and no known primary malignancy present a diagnostic chal-lenge. Previous studies suggest a protocol for evaluating such lesions which involves history and exam focusing on thyroid, breast and prostate, advanced imaging including computed tomography (CT) scan of the chest, abdomen and pelvis and laboratory analysis including complete blood count (CBC) with differential, electrolyte and renal function panel, liver function panel, alkaline phosphatase, erythrocyte sedimentation rate (ESR), and serum protein electrophoresis (SPEP)1-4. Using this algorithm in a prospective manner for patients with metastatic disease, Rougraff et al. were able to identify a primary site of malignancy in 85% of patients2.
The aforementioned study, however, excluded patients with SPEP levels suggestive of myeloma as they were presumed to have a lesion due to this disease. There are no studies that we are aware of looking at the clinical utility of the SPEP in this exact setting. It has been our general impression that the eventual histologic diagnosis does not necessarily correlate with the interpretation of the SPEP done at the initial work up of a lytic lesion. Further, it has been suggested that a plasma cell neo-plasm can be accurately diagnosed by histologic and immunophenotyping analysis in isolation, and previous knowledge of the SPEP value is not required.5 As nearly all patients being evaluated for a radiolucent bone lesion of unknown etiology require a confirmatory biopsy, the value of the SPEP in the initial work-up is uncertain.
Our primary question in designing this study was: what is the diagnostic performance of the SPEP in the setting of a lytic lesion of unknown origin? As a sec-ondary outcome we were interested in the diagnostic performance of the SPEP in patients presenting with multiple lytic lesions.
This study was designed as a retrospective chart review. After obtaining approval from our institutional review board, patients were identified by searching our electronic database for all those who had both an SPEP as well as a biopsy performed between the years of 2001 and 2010. Inclusion criteria were patients with a pathology report of the bone biopsy and an SPEP value obtained prior to the biopsy. Exclusion criteria were patients with incomplete records and those younger than 40 years old, as they lie outside the common age range for metastatic disease and multiple myeloma.
After identification of the patients, a limited chart review was performed to identify the location of the le-sion, whether it was solitary or if there were multiple, the histopathologic diagnosis, and the presence of a monoclonal spike. An SPEP value was deemed to be positive, and therefore indicative of a plasma cell neo-plasm (monoclonal gammopathy of undetermined sig-nificance, plasmacytomas, multiple myeloma, plasma cell leukemia), if there was a detectable monoclonal spike in the gamma region of the electrophoresis6.
Two by two tables were then constructed to deter-mine the diagnostic performance of the SPEP in the diagnosis of plasma cell neoplasm. From these tables the sensitivity and specificity were calculated. We esti-mated the prevalence of multiple myeloma in this clinical scenario (lytic lesion of unknown origin) at 17% from previously reported studies 7,8. We then calculated the positive predictive value and negative predictive value9. A second analysis was performed looking at only the group presenting with multiple lytic lesions (patients presumed to be more likely due to multiple myeloma) to see if the diagnostic performance of the SPEP was en-hanced in this group. For patients with a positive SPEP value, we calculated average levels of the monoclonal spike (in g/dL) to compare the value between those who were eventually diagnosed with a plasma cell neoplasm, and those who were not. A two-tailed student’s t-test was used to compare the means to determine statistical significance.
Our retrospective review identified 182 patients who met criteria for the study. Anatomic locations for the lesions are located in Figure 1. Sixty one percent of the lesions were encountered in the axial skeleton (skull, spine, and pelvis) with the remainder being found in the appendicular skeleton. The femur was the most common long bone affected. Of the 182 patients analyzed, 46 patients were diagnosed with a plasma cell neoplasm (38 with multiple myeloma, six plasmacytomas, one plasma cell leukemia and one monoclonal gammopathy of un-determined significance). This indicates a prevalence of 25% in our series (Figure 2), which is larger than the estimated population prevalence of 17% that was used in our calculations.
When evaluating all patients presenting with a lytic le-sion the sensitivity and specificity of the SPEP was 71% and 83%, respectively (Table 1). This indicated a positive predictive value (PPV) and negative predictive value (NPV) of 47% and 94%, respectively. Seventy six of the 182 patients presented with greater than one lytic lesion. In this patient subgroup there were 34 patients who were ultimately diagnosed with a plasma cell neo-plasm, representing an increase in series prevalence to 45%. The sensitivity and specificity in this subgroup were 71% and 79% (Table 2). The PPV and NPV were 40% and 93%, respectively, and were calculated using a prevalence of 17%.
For a sensitivity analysis, we substituted our literature- based prevalence estimation with those found in our series (Table 3) to determine the effect of this estimation on PPV and NPV.
For those patients with a positive SPEP who were eventually diagnosed with a plasma cell neoplasm (n = 33), the average IgG value was 1.6 g/dL (range 0 – 7.2 g/dL). For those patients with a positive SPEP who were not diagnosed with a plasma cell neoplasm (n = 23), the average value was 0.6 g/dL (0 – 6.1 g/dL). The p value comparing these means was 0.07.
The diagnostic algorithm proposed in the classic 1993 study by Rougraff et al. excluded patients with SPEP levels suggestive of myeloma2. The implication is that a positive SPEP indicates a high likelihood of a plasma cell neoplasm and a negative value insinuates absence of this disease. It has been our general impression that this is not the cause. Our goal was to calculate the sensitivity, specificity, positive predictive value, and nega-tive predictive value of serum protein electrophoresis in the population of patients presenting for evaluation of a radiolucent bone lesion of unknown etiology. Determina-tion of the inherent characteristics of this test allows for further discussion regarding the necessity and timing of the SPEP in these patients.
The main limitation of this study is that it is a ret-rospective review. This is problematic because it is dependent upon accurate data entry at the time of patient presentation. Another weakness is the possibil-ity that some selection bias has been incorporated into our patient group. An overall prevalence of a plasma cell neoplasm of 25% is quite high, as previous studies have indicated a prevalence of 17%7,8. It is possible that patients were selected to have an SPEP lab drawn by the practitioner only if there were radiographic and clinical history characteristics consistent with multiple myeloma. Although this would undoubtedly introduce a selection bias, it should theoretically bias our results towards an improved diagnostic performance of the SPEP (i.e. improved PPV).
Our primary goal was to evaluate the diagnostic per-formance of the SPEP in the setting of a lytic lesion of unknown origin. It has been suggested that the SPEP alone is not a definitive test for myeloma, as a review of 1027 patients from the Mayo Clinic noted a positive SPEP in only 83% of the new diagnoses10. We found a sensitivity of 73% and a specificity of 81% when analyzing patients presenting with a lytic bone lesion who were eventually diagnosed with a plasma cell neoplasm. That same study from the Mayo Clinic, however, did note an improved diagnostic accuracy when incorporating the urine protein electrophoresis (UPEP), immunofixation electrophoresis (IFE) and free-light chain (FLC) assay10. Although there are no defined “acceptable” values for a diagnostic test, obviously the goal is to optimize both the sensitivity and specificity to prove that the test is reliable, reproducible, and accurate. For example, a widely used diagnostic tool, computed tomography (CT) scanning, has been shown in some studies to have 100% sensitivity and 98.9% specificity in the diagnosis of acute appendicitis11. Our results indicate that the SPEP has suboptimal performance as a definitive diagnostic test in isolation.
However, when compared to screening tests (studies performed in an asymptomatic population at risk of a disease), it performed similarly. For example, a prostate specific antigen (PSA) value of greater than 4 ng/mL has a sensitivity of 21% and specificity of 91% in the screening of prostate cancer12,13. PPV and NPV are reported at 30% and 85%, respectively. Mammography for breast cancer screening has been shown to have a sensitivity of 70% and specificity of 92% with a PPV of 5%9. For a screening test to be a good one, it should have a high specificity, or very low false positive rate. Although the SPEP is not truly a screening test, as it is being performed because a clinical finding prompted further work-up, it is help-ful to consider its performance in this context. When viewed as such, it performs reasonably well to “rule out” multiple myeloma. A negative predictive value of 94% shows that, when negative, the SPEP reliably indicates a low likelihood of the diagnosis of myeloma in patients presenting with at least one lytic bone lesion.
As a secondary outcome we were interested in the diagnostic performance of the SPEP in patients present-ing with multiple lytic lesions. It would seem logical that this would enhance the diagnostic performance of the SPEP as those with multiple lytic lesions would be more likely to have a diagnosis of myeloma; however, our data indicates that this was not the case. The performance was essentially unchanged, with sensitivity and specific-ity of 71% and 79%, respectively.
Additionally, we found that there was a trend towards larger absolute values of monoclonal spikes in the patients with a “positive” SPEP in the patients with a diagnosis of a plasma cell neoplasm compared to those with a false positive (p=0.07). Although the total num-bers are relatively small, the wide range in values for the SPEP for each group (true and false positives) and the presence of several histologically confirmed cases of a plasma cell neoplasm with no measurable SPEP further supports the notion that this test is lacking substantially in terms of diagnostic performance.
It is interesting to note that our comparison screening tests (PSA and mammography) are quite controversial in their clinical utility. Both have both been publicly criticized as unhelpful in the overall care of patients, even though their inherent characteristics may be acceptably accurate14,15. One may make a similar argument in this example, as an SPEP value at the initial evaluation of these patients means little by itself. Our data indicate that SPEP lacks the sensitivity, specificity, PPV, and NPV characteristic of a good diagnostic test; however, it may have some utility in ruling out myeloma at initial presentation. Given that plasma cell neoplasms are readily characterized on histopathologic exam without knowledge of the SPEP result, our findings raise the question of whether this test should be used at all prior to a histologically confirmed diagnosis of a plasma cell neoplasm. This test may be best utilized as a baseline, after a confirmatory biopsy, by which to measure the response to therapy16. Alternatively, if the SPEP is to be used in the work-up of a patient with a lytic bone lesion, it should be combined with UPEP, IFE and FLC assay so as to avoid the diagnostic errors that our data indicates would ensue from reliance upon SPEP alone.