Immunotyping Provides Equivalent Results to Immunofixation in a Population with a High Prevalence of Monoclonal Gammopathies

Abstract

Background

Serum immunofixation (IF) is a common laboratory test used to diagnose and monitor patients with monoclonal gammopathies. Similarly, immunotyping (IT) by capillary electrophoresis can confirm the presence of a monoclonal protein (M-protein) and determine its isotype. The goal of this study was to compare the ability of IT and IF to detect M-proteins.

Methods

IT and IF results for 1000 waste clinical serum samples were obtained. All results were interpreted blindly by reviewers who were experienced in each technique. Results were compared by band. Results were also compared to patient history to determine if the original clone was present. We determined the sensitivity of IT and IF alone and in combination with additional tests. Finally, we evaluated the impact of reviewer training on the sensitivity of IT.

Results

IT and IF were concordant in 721/773 (93%) samples with a history of an intact M-protein and in 143/172 (83%) samples with a history of a free light chain (FLC) M-protein. IF was significantly more sensitive than IT for the detection of FLC M-proteins (P < 0.0001). However, IF was not more sensitive than IT for detection of intact M-proteins (P = 0.1272) or when each test was combined with the FLC ratio or urine immunofixation (P = 0.2812 and P = 0.6171, respectively). Finally, after training, inexperienced reviewers improved their IT sensitivity by 19%.

Conclusion

IT provides equivalent results to IF for the detection of monoclonal proteins. Training and experience are critical to the accurate interpretation of IT.

Impact Statement

Gel-based serum immunofixation (IF) and capillary-based immunotyping (IT) are FDA-approved for detecting and isotyping monoclonal proteins (M-protein). Early studies comparing these techniques may have been affected by insufficient IT-reading experience. Therefore, we conducted a large comparison study to reevaluate this technique. We show that reviewer training is critical to the accurate interpretation of IT. When experienced reviewers are used, we find that the sensitivity of IT is similar to IF. Our study is important because IT provides a more automated method for M-protein detection. This paper will benefit patients with monoclonal gammopathies and others who receive testing for M-protein detection.

Introduction

Multiple myeloma and other monoclonal gammopathies are characterized by the proliferation of plasma cells that often secrete a monoclonal immunoglobulin (M-protein). This M-protein biomarker can be detected in serum and urine and is an important component in the diagnosis and monitoring of these diseases.

Several assays are available for the detection, quantitation, and characterization of the M-protein. Protein electrophoresis, either by agarose gel electrophoresis or capillary electrophoresis, is used to detect and quantitate M-proteins in serum and urine. Immunofixation electrophoresis (IF) is most commonly used to confirm the presence of an M-protein and determine its isotype. However, a capillary-based method, called immunotyping (IT), is also available and FDA-cleared for this purpose. IT uses antisera against IgG, IgA, IgM, kappa, and lambda to shift the migration of immunoglobulins on capillary electrophoresis. Disappearance of the abnormality in the antiserum-treated pattern compared to the reference trace indicates the presence of an M-protein and allows identification of its isotype.

Although IT offers several advantages over IF including automated sample analysis, sample tracking, and digital results, fewer laboratories have adopted this technique (1). In addition, the International Myeloma Working Group (IMWG) only specifies the use of immunofixation to evaluate patients’ response to therapy (2). The reason for this may be in part because some (3–5), but not all (6, 7), previous studies have found IT to be less sensitive than IF. However, it has been suggested that the difference in sensitivity may be due to insufficient reviewer training and experience reading IT (3, 8).

To address this issue and evaluate the performance of IT, we conducted a large comparison study using reviewers who were properly trained and experienced in interpreting each technique. Both IT and IF results were interpreted in a blind manner without access to patient history or other laboratory results. We also specifically tested if reviewer training affects the sensitivity of IT. We report our findings for 1000 samples in a population with a high prevalence of monoclonal gammopathies.

Materials and Methods

Patient Samples

Waste clinical serum samples (gold-top, serum separator tubes) were obtained from the clinical laboratory at Memorial Sloan Kettering Cancer Center (MSKCC) with approval from the Institutional Review Board. Any sample that had serum protein electrophoresis (SPEP), IF, and free light chain (FLC) studies ordered for clinical purposes was included in the study. There were no exclusion criteria for sample collection. We stopped collecting samples once we reached 1000. This included samples from 915 unique patients. As we are a cancer center, patients receiving these tests primarily include those who have been diagnosed with a monoclonal gammopathy. Samples were stored at 4 °C for up to 5 days prior to IT analysis and up to 3 days prior to clinical SPEP/IF analysis. Chart review was performed to collect relevant patient information including the isotype and migration pattern of the patient’s original clone, clinical diagnosis, use of monoclonal antibody therapies, and clinical test results including SPEP, IF, FLC values, and urine immunofixation (UIF) results.

SPEP, IF, UIF, IT, and FLC Assays

Serum protein electrophoresis and immunotyping were performed on a CAPILLARYS 3 Tera instrument using the Capi-3 Protein(E) 6 kit and Immunotyping kit, respectively (Sebia, Inc.; Norcross, GA, USA). Serum and urine immunofixation were performed on the Hydrasys 2 analyzer (Sebia, Inc.) using the Hydragel 9 or Hydragel 4 IF kits for serum and the Hydragel 9 Bence Jones kit for urine (Sebia, Inc.). Free light chain quantitation was performed on an Optilite analyzer (The Binding Site). All assays were performed according to the manufacturer’s instructions. Assays were validated and are in clinical use in a CLIA compliant laboratory.

Result Interpretation

This was a blinded study in which the reviewers did not have access to patient history or other test results. All IT results were interpreted by 3 experienced reviewers from the Result Interpretation Escalation Team at Sebia. IF results were interpreted by 3 PhD/MD level scientists from MSKCC specifically for this study. The clinical IF interpretations were not used for comparison with IT. A consensus call was made for each technique when individual interpretations were discordant.

To evaluate the impact of training on IT sensitivity, MSKCC reviewers also individually interpreted the first 360 IT results. MSKCC reviewers were self-taught and had minimal IT-reading experience (<5 IT/week for clinical purposes). Samples 1–120 were used to establish baseline performance of the reviewers. MSKCC reviewers were then trained by the Result Interpretation Escalation Team from Sebia. Training consisted of a 2-hour lecture outlining specific steps for reading IT results. MSKCC reviewers then practiced with samples 121–140 and met for a follow-up session to review mistakes. Samples 241–360 were used to test the hypothesis. The individual MSKCC IT interpretations were compared to the blind, consensus IF results. The individual expert IT interpretations for these 360 samples were also compared to the IF results for reference. Finally, reviewer agreement for each of the techniques was evaluated.

To evaluate performance of IT compared to IF, the blind, consensus interpretations were compared for all 1000 samples. We first analyzed the results by band. That is, all monoclonal bands detected by IT were compared to all monoclonal bands detected by IF. For this analysis, we excluded samples in which the IT or IF result was positive but individual bands were not specified.

We also analyzed results with respect to the “original clone” since the IMWG response classification (ex. complete response, stringent complete response, etc.) is based on the presence/absence of the disease-specific M-protein. We used the patient’s clinical history, which typically included historical IF images depicting the patient’s M-protein (sometimes from years prior to this study), to determine the patient’s original clone. The blind IT and IF interpretations were then analyzed to determine if the original clone was detected. Each sample was categorized as “positive” or “negative” only with respect to original clone. We used the combined results from IF, IT, and FLC/UIF as the reference standard. If the original clone was detected, it was considered a true positive. If the clone was not detected by any test, it was considered a true negative. If the clone was detected by 1 test but not the other(s), it was considered a false negative for the test(s) that did not detect the M-protein. In cases where the original disease included multiple intact M-proteins (ex. IgG kappa, IgA lambda biclonal gammopathy), the sample was considered positive if at least 1 of the proteins was detected. For this part of the analysis, we excluded samples in which the original clone could not be determined because the patient did not have a diagnosis of a monoclonal gammopathy, had nonsecretory disease, or because the original IF was not available to determine the migration of the disease-specific M-protein. Finally, we determined the sensitivity of IT and IF alone and when combined with FLC results or urine immunofixation results. A sample was considered positive if any test in the panel was abnormal. Binomial proportion confidence intervals were calculated for the different sensitivities of individual tests and test panels. Differences in sensitivities were evaluated using the McNemar test. P < 0.05 indicated statistical significance.

Results

Impact of Reviewer Training on IT Sensitivity

Prior to training, the average IT sensitivity for the 3 MSKCC attendings was 69% (range 65–73%) compared to the blind IF interpretations. After formal training and practice with 120 samples, the average sensitivity of the MSKCC reviewers was 88% (range 86–91%), an improvement of 19%. For comparison, the sensitivity of expert IT review for the test set (samples 241–360) was 95%.

Reviewer Discordant Rates

The 3 MSKCC reviewers were in complete agreement for 271/360 (75%) samples for IT interpretations. At least 1 reviewer had a discrepant interpretation for the remaining 89/360 (25%) samples. The discordance rates for MSKCC IF interpretations and Sebia IT interpretations were 26% and 25%, respectively, for the first 360 samples in this study.

Comparison of Techniques by Band

One-hundred and seventy-six samples were negative for monoclonal bands by both IT and IF. Of the 824 samples that were positive by at least one technique, we excluded 36 samples because the IF or IT result was “multiple bands” but was otherwise not specified. For the remaining 788 samples, a total of 1124 monoclonal bands were detected (Table 1). IT matched IF for 772 bands (69%, 95% CI: 66–71%). IT detected 184 bands (16%, CI: 14–19%) that were missed by IF, and IF detected 168 bands (15%, CI: 13–17%) that were missed by IT. Out of the 352 bands in which the techniques were discrepant (IF+/IT− or IF−/IT+), 339 (96%) did not produce a visible M-spike on SPEP. IT found more IgGκ and IgGλ bands compared to IF. IF found more IgMκ, IgMλ, free κ, and free λ bands compared to IT.

Table 1.

Monoclonal bands detected by IT and IF^a.

Monoclonal bands	Concordant	Discordant		Total
	IF+/IT+	IF+/IT−	IF−/IT+
GK	423	55	89	567
GL	181	21	57	259
AK	67	11	8	86
AL	42	3	2	47
MK	32	10	5	47
ML	4	15	6	25
free K	8	23	10	41
free L	15	30	7	52
Total (%, 95% CI)	772 (69%, 66–71)	168 (15%, 13–17)	184 (16%, 14–19)	1124
M-spikeb	615	1	12	628
No M-spike	157	167	172	496

^aBands were detected in 788 samples. 176 samples were negative by both techniques. 36 samples were excluded because individual bands were not specified in either the IT or IF interpretation.

^bNumber of bands that produced a visible M-spike on SPEP.

Comparison of Techniques by Disease-Specific M-Protein

Original clones could not be determined for 55 samples, which were excluded for this part of the analysis. Of the remaining 945 samples with known original clones, 773 has a history of an intact M-protein and 172 had a history of a FLC M-protein (Table 2 and Supplemental Table 1). We analyzed these groups separately because the IMWG definition of CR varies depending on whether the original clone is an intact or FLC M-protein.

Table 2.

Original clones detected and corresponding IMWG classifications^a.

Original clone		Concordant				Discordant				Total
		IF+/IT+		IF−/IT−		IF+/IT−		IF−/IT+
Intact M-proteins		570		151		32		20		773
	IMWG response classification	IF: non-CR		IF: CR/sCR		IF: non-CR		IF: CR/sCR
		IT: non-CR		IT: CR/sCR		IT: CR/sCR		IT: non-CR
FLC M-protein		16		127		26		3		172
	IMWG response classification	FLC+	FLC−	FLC+	FLC−	FLC+	FLC−	FLC+	FLC−
		16	0	69	58	26	0	3	0
		IF: non-CR	IF: non-CR	IF: non-CR	IF: CR	IF: non-CR	IF: non-CR	IF: non-CR	IF: CR
		IT: non-CR	IT: non-CR	IT: non-CR	IT: CR	IT: non-CR	IT: CR	IT: non-CR	IT: non-CR
Total		586 (62%)		278 (29%)		58 (6%)		23 (2%)		945

^a55 samples with unknown original clones were excluded.

Abbreviations: CR, complete response.; sCR, stringent complete response.

IT and IF were concordant in 721/773 (93%) intact M-protein samples and in 143/172 (83%) FLC M-protein samples (Table 2). IF and IT were discordant in 52/773 (7%) intact M-protein samples and in 29/172 (17%) FLC M-protein samples. IF was significantly more sensitive than IT for the detection of FLC M-proteins (P < 0.0001) but not for the detection of intact M-proteins (P = 0.1272). The IMWG response classification differed for the 52 discrepant intact M-protein samples but not for the 29 discrepant FLC M-protein samples (Table 2). All 29 discrepant FLC samples were positive by the FLC assay and would be classified as non-CR regardless of the IF or IT result (Table 2). Patterns for why certain clones were missed by each technique are shown in Table 3. Examples of small, clinically relevant M-proteins detected by IT are shown in Fig. 1.

Table 3.

Reasons for missed original clone.

Reason for missed original clone on IT	Count
Free LC	26
Beta region	13
Hypogammaglobulinemic sample	9
Broad band	7
Overlapping bands	3
Reason for missed original clone on IF	Count
Broad band	11
Hypergammaglobulinemic sample	5
Cathodal band	4
Overlapping bands	2
Hypogammaglobulinemic sample	1

Fig. 1.

Examples of small, clinically relevant M-proteins identified by IT in 2 patients. (A, D), IT results. Abnormalities are noted with arrows. M-proteins are detected by determining if an abnormality is removed from the reference trace (black) and/or remains in antisera-treated traces (red). (B, E), Corresponding IF results. (C, F), Historical IF results showing migration and isotype of original clone.

When combined with the FLC ratio, the original clone was detected in 763 and 756 samples by IF and IT, respectively (Table 4). IF combined with FLC was not significantly more sensitive than IT combined with FLC (P = 0.2812). Similar results were observed when IF or IT were combined with urine immunofixation (UIF) results. There were 238 patients who had a UIF performed during the study. When combined with UIF, the original clone was detected in 200 and 198 samples by IF and IT, respectively (Table 5). The difference in sensitivity between IF combined with UIF and IT combined with UIF was not significant (P = 0.6171).

Table 4.

Sensitivity of tests with and without FLC.

Assay	Samples with history of intact M-protein (N = 773)		Samples with history of FLC M-protein (N = 172)		Total (N = 945)
	Number of positive samples	Sensitivity (95% CI)	Number of positive samples	Sensitivity (95% CI)	Number of positive samples	Sensitivity (95% CI)
IF	602	91% (89–93)	42	37% (28–46)	644	83% (80–86)
IT	590	89% (87–92)	19	17% (10–25)	609	79% (76–81)
FLC	422	64% (60–68)	114	100%	536	69% (66–72)
IF + FLC	649	98% (97–99)	114	100%	763	98% (97–99) a
IT + FLC	642	97% (96–98)	114	100%	756	98% (96–99) a
IF + IT +FLC	661	100%	114	100%	775	100%

^a(IF + FLC) vs (IT + FLC), P = 0.2812.

Table 5.

Sensitivity of tests with and without UIF.

Assay	Samples with history of intact M-protein and UIF results available (N = 189)		Samples with history of FLC M-protein and UIF results available (N = 49)		Total (N = 238)
	Number of positive samples	Sensitivity (95% CI)	Number of positive samples	Sensitivity (95% CI)	Number of positive samples	Sensitivity (95% CI)
IF	160	92% (87–96)	16	57% (37–76)	176	88% (82–92)
IT	155	90% (84–94)	7	25% (11–45)	162	81% (74–86)
UIF	140	81% (74–86)	28	100%	168	84% (78–88)
IF + UIF	172	99% (97–100)	28	100%	200	100% (97–100) a
IT + UIF	170	98% (95–100)	28	100%	198	99% (96–100)a
IF + IT + UIF	173	100%	28	100%	201	100%

^a(IF + UIF) vs (IT + UIF), P = 0.6171.

Discussion

Identification of monoclonal immunoglobulins in serum is an important component in diagnosing and monitoring monoclonal gammopathies. Previous studies found that IT correctly identified M-proteins at quantifiable concentrations, but missed low concentration bands compared to IF, particularly IgMs, IgAs, and free light chains (3–5). In addition, 1 study found IT interpretation to be more subjective and more inconsistent compared to IF (4).

In our study, we found that reviewer training is critical to the accurate interpretation of IT. Training improved IT sensitivity by 19%. Even after formal training, the sensitivity of MSKCC interpretations was slightly less than that of experienced Sebia reviewers suggesting that sensitivity continues to improve with experience. In addition, we did not find IT interpretations to be more subjective or variable compared to IF, even prior to training. Reviewer discordance was about 25% regardless of the technique or reviewers.

As with IF, we found that the most challenging part of interpreting IT is distinguishing between the signal and noise. Training improved our ability to identify small bands because it emphasized the importance of reviewing results electronically. Software features, such as zooming and unsmoothing functions, aid in the identification of small abnormalities. Training also helped dispel a major myth about IT. As a subtractive technique, it is often assumed that an abnormality must be visible in the reference trace for one to notice its removal (9). In fact, the subtracted traces contain important information about what proteins remain. As shown in Fig. 1, D, some abnormalities are only visible after proteins have been removed. In this example, an abnormality in the κ-subtracted trace indicates that a λ-containing M-protein remains. Since the IgG- and λ-subtracted traces are normal, the M-protein was identified as an IgGλ, which matched clinical history (Fig. 1, F). Based on our results and experience with formal training, the lower sensitivity and inconsistent interpretations reported in previous studies are not surprising if the reviewers were less familiar with interpreting IT.

When IT results were interpreted by experienced readers and compared to blind IF results, we found that IT performs similarly to IF. As observed in previous studies, IT correlates very well with IF when the M-protein is at high concentration (Table 1). Discrepancies are observed but occur with low concentration bands and are balanced. Out of 1124 bands detected by either technique, IT found 184 (16%, CI: 14–19) bands that were missed by IF. IF found 168 (15%, CI: 13–17) that were missed by IT. There was no statistically significant difference between the number of discrepant bands found by each technique.

In addition to comparing techniques by band, we also analyzed results with respect to original clone. We did this for 2 reasons. First, the presence or absence of the original clone is the most clinically relevant piece of information provided by these tests. Other monoclonal bands may be detected, but only the original M-protein matters in the IMWG definition of complete response (10). Second, by referring to the original clone, we were able to determine the clinical significance of discrepant results.

Many bands detected in our study, particularly IgGκ and IgGλ bands, are not original clones. This is not surprising since patients can develop oligoclonal responses during treatment, which often are IgG isotypes (11), and monoclonal antibody drugs are frequently used in our patient population. For the bands that are original clones, we found that IT and IF agree for 721/773 intact M-protein samples and for 143/172 FLC M-protein samples. Both techniques miss low concentration, clinically relevant bands that the other technique detects (Table 2). Compared to IF, IT is better at detecting bands that migrate cathodally or in a hypergammaglobulinemic background, particularly IgG M-proteins (Fig. 1, Table 3). IF is better at detecting FLCs, bands in a hypogammaglobulinemic background, and small bands that migrate in the beta region, which often include IgA and IgM M-proteins. Both techniques can miss M-proteins when they overlap another band. Both techniques can miss broad bands, which are sometimes are interpreted as being polyclonal. IT is slightly better at detecting these broad bands perhaps because it provides better resolution than IF. Finally, it is important to note that IgD and IgE testing is only available by IF.

Although IF is better at detecting FLC bands compared to IT, neither technique is particularly good for this purpose (Table 4). The free light chain ratio was abnormal in 114/172 samples with a history of a FLC M-protein. In comparison, IF was positive in 42/172 and IT was positive in 19/172 samples. There were no instances in which the FLC ratio was normal when IT or IF detected a monoclonal FLC band (Table 2). Similarly, urine immunofixation detected FLC bands better than either IF or IT and there were no instances in which urine immunofixation was normal when IT or IF detected a monoclonal FLC band (Table 5).

It is important to remember that serum assays should not be used in isolation for detecting and monitoring monoclonal gammopathies and, instead, should be used in combination with free light quantitation or urine immunofixation (12). Therefore, it is best to evaluate the performance of IT as part of a panel. When combined with the FLC ratio or urine immunofixation, IF is not more sensitive than IT (P = 0.2812 and P = 0.6171, respectively). This finding has been observed by a previous study that showed IT with FLCs was a suitable alternative for the evaluation of patients with AL amyloidosis (6).

Our results are relevant to laboratories looking for more automated methods for detecting M-proteins. IT provides a possible solution and we show that the analytical and clinical performance of IT is similar to IF, provided that reviewers are properly trained. Of course, many other factors come into play when deciding to switch methods. For M-protein detection, the use of these tests in clinical trials is also important. The IMWG specifies the use of IF to evaluate patient response to therapy in clinical trials (2). Based on results from our study, we conclude that IT can be used interchangeably with IF and believe that IMWG guidelines should be updated to reflect this.

In summary, we found that training and experience are critical to the correct interpretation of IT results. When experienced reviewers interpret IT, we find that the sensitivity is similar to IF for the detection of monoclonal proteins. Discrepancies between IT and IF are observed at lower M-protein concentrations when an apparent M-spike is not visible by SPEP. However, these discrepancies are balanced and each technique finds clinically relevant bands that the other misses. Similar sensitivities are observed with IF and IT when each test is used in combination with serum free light chain quantitation or with urine immunofixation. It may also be useful for laboratories to reflex to IF in cases of hypogammaglobulinemia or beta-region-migrating M-proteins. Workflows and reflex strategies will vary depending on the patient population and whether the laboratory has access to patient history and medication information. Further work is needed to evaluate the practical and clinical utility of various strategies and the performance of IT longitudinally.

Supplemental Material

Supplemental material is available at The Journal of Applied Laboratory Medicine online.

Nonstandard Abbreviations

M-protein, monoclonal immunoglobulin; IF, immunofixation; IT, immunotyping; IMWG, International Myeloma Work Group; MSKCC, Memorial Sloan Kettering Cancer Center; SPEP, serum protein electrophoresis; FLC, free light chain; UIF, urine immunofixation.

 

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest: Employment or Leadership: None declared. Consultant or Advisory Role: None declared. Stock Ownership: None declared. Honoraria: K.L. Thoren, Sebia; S.I. McCash, Sebia. Research Funding: This study was supported by Sebia, Inc. (to K.L. Thoren) and in part by the Memorial Sloan Kettering Core Grant (P30 CA008748) funded by the National Cancer Institute. Expert Testimony: None declared. Patents: None declared.

Role of Sponsor: The funding organizations played a direct role in the design of study, review and interpretation of data, and final approval of manuscript. The funding organizations played no role in the choice of enrolled patients or preparation of manuscript.