Serum immunoglobulin free light chains and their association with clinical phenotypes, serology and activity in patients with IgG4-related disease

The clinical utility of serum immunoglobulin free light chains (sFLC) in IgG4-related disease (IgG4-RD) is unknown. Herein we evaluated their association with clinical phenotypes, serology and activity in patients with IgG4-RD. Cross-sectional study that included 45 patients with IgG4-RD, and as controls 25 with Sjögren’s syndrome (SS) and 15 with sarcoidosis. IgG4-RD patients were classified in clinical phenotypes: pancreato-hepato-biliary, retroperitoneum/aorta, head/neck-limited and Mikulicz/systemic; as well as proliferative vs. fibrotic phenotypes. We assessed the IgG4-RD Responder Index (IgG4-RD RI) at recruitment and measured IgG1, IgG4, κ and λ sFLC serum levels by turbidometry. sFLC levels were similar among IgG4-RD, SS and sarcoidosis groups. Regarding the IgG4-RD patients, the mean age was 49 years, 24 (53.3%) were men and 55.5% had activity. Eight (17.7%) belonged to pancreato-hepato-biliary, 6 (13.3%) to retroperitoneum/aorta, 14 (31.1%) to head/neck-limited, 16 (35.5%) to Mikulicz/systemic phenotypes, whereas 36 (80%) to proliferative and 9 (20%) to fibrotic phenotypes. High κ sFLC, λ sFLC and κ/λ ratio were present in 29 (64.4%), 13 (28.9%) and 13 (28.9%) of IgG4-RD patients, respectively. There were no differences in sFLC among IgG4-RD phenotypes. κ sFLC and κ/λ ratio correlated positively with the number of involved organs and IgG4-RD RI. Patients with renal involvement had higher κ sFLC and λ sFLC. The AUC for κ sFLC and λ sFLC, for renal involvement was 0.78 and 0.72, respectively. Active IgG4-RD had higher levels of κ sFLC and more frequently a high κ/λ ratio. The AUC for κ sFLC and κ/λ ratio for predicting active IgG4-RD was 0.67 and 0.70, respectively. sFLC correlated positively with IgG1 and IgG4 levels. sFLC may be useful as a biomarker of disease activity as well as multiorgan and renal involvement. In particular, a high κ/λ ratio may identify patients with active disease.

Patients with active IgG4-RD had more frequently high κ sFLC compared to patients with inactive IgG4-RD, although without reaching a statistical significance (   www.nature.com/scientificreports/ active IgG4-RD patients with a high κ/λ ratio had normal serum IgG4 levels, four of them had de novo active disease and two relapsing disease. The AUC for κ sFLC and κ/λ ratio, for predicting active IgG4-RD, with a cut-off value of 33.42 mg/L and 1.61, were 0.67 and 0.70, respectively (Fig. 6).
In a sensitivity analysis including only the 14 patients with de novo active disease, we also found a higher level of κ sFLC (33.  .77], p = 0.017). The AUC for κ sFLC and κ/λ ratio for discriminating de novo active from inactive disease with the same cut-off values (33.42 mg/L and 1.61) were 0.75 and 0.74, respectively. On the contrary, when including only the nine patients with relapsing disease, there were no differences compare to the inactive group (data not shown).
IgG4-RD patients with damage at the time of recruitment did not have differences in sFLC levels compare to patients without damage (data not shown).
Patients with biopsy proven disease (n = 35) have no differences in the prevalence κ sFLC, λ sFLC and the κ/λ ratio and their levels compare to patients without biopsy proven disease (n = 10) (data not shown). The number of IgG4 + plasma cells in tissue did not correlate with κ sFLC (rho = 0.23, p = 0.25) λ sFLC (rho = 0.16, p = 0.41) nor the κ/λ ratio (rho = 0.28, p = 0.16). sFLC in IgG4-RD patients under immunosuppressive treatment. There were no differences in the prevalence and levels of κ sFLC, λ sFLC and the κ/λ ratio between patients with and without immunosuppressive treatment (data not shown).
In a sensitivity analysis including the 14 patients with active untreated IgG4-RD, an elevated κ/λ ratio remained more frequent compare to SS and sarcoidosis patients (

Discussion
In this study we demonstrated that a substantial proportion of patients with IgG4-RD had high levels of sFLC, and although their assessment may not be useful for distinguishing from other conditions, it may be valuable as a biomarker of disease activity and multiorgan and renal involvement. Additionally, a high κ/λ ratio may be valuable for the same purposes.
To date, only one study have assessed sFLC levels in patients with IgG4-RD. Grados et al. determined sFLC levels in 16 French patients with IgG4-RD and found that they were higher compared to a group of healthy www.nature.com/scientificreports/ subjects; furthermore, they found that 87% and 53% of the patients had elevated κ and λ sFLC respectively. However, due to the small sample size they were not able to assess the relationship of sFLC with clinical characteristics and disease activity 20 . In our study, we found that 64.4% and 28.9% of our cohort had elevated κ and λ sFLC respectively, which is lower than the frequencies reported by Grados et al. These differences may be explained by the fact that in the French study patients with severe kidney impairment were also included and it is well established that sFLC correlate negatively with the degree of renal insufficiency 21 .
The increase levels of sFLC in patients with IgG4-RD is undoubtedly related to the increase synthesis of immunoglobulins by the B cell lineage which also could explain why we found a positive correlation between sFLC and IgG1 and IgG4 levels. This phenomenon is well described in systemic autoimmune diseases in which B cell over-activation is a component of the pathogenesis of the disease such as SS and SLE 4,5 . It was noteworthy that a considerable proportion of our IgG4-RD cohort (28.9%) had an abnormal κ/λ ratio, in contrast to the lower frequencies found in SS and SLE 4,22 . Skewing of the κ/λ ratio reflects the excess production of one immunoglobulin light chain over the other and is considered a surrogate laboratory parameter indicating clonality in plasmaproliferative or lymphoproliferative disorders 2 . This pattern of κ-restricted pseudoclonality has been previously reported in IgG4-RD and in settings where there is polyclonal IgG4 excess. Grados et al. found an elevated κ/λ ratio of sFLC in 44% of 16 patients with IgG4-RD 20 . Furthermore, two studies have demonstrated that sera from patients with increased serum IgG4 levels evaluated by immunosubstraction, including patients with IgG4-RD, may have a monotypic appearance with a κ-restricted pseudoclonal pattern 23,24 . The observed κ-restricted pseudoclonality may be attributable to the intrinsic characteristics of the IgG4 subclass itself. The ratio of κ-to λ-expressing IgG, commonly 2 to 1, depends on the IgG subclass. IgG1, for example, has a κ/λ ratio of 2.4 to 1, whereas IgG4 has a κ/λ ratio of about 8 to 1 with a heavily skewing towards the κ light chain 24 . This κ-restricted pseudoclonality seen by immunosubstraction and by measurement of sFLC, together with the polyclonal IgG4 serum protein electrophoresis pattern that may mimic a monoclonal gammopathy 23,24 , may lead to a presumed diagnosis of clonality. While some cases of IgG4-related ophthalmic disease with κ light chain  www.nature.com/scientificreports/ restriction evaluated by in-situ hybridization have been reported 25,26 , if tissue specimens from IgG4-RD patients had the same immunoglobulin light chain pseudo-restriction remains to be determined. There is an unmet need regarding biomarkers in IgG4-RD. Up to date, only serum IgG4 and complement levels have a role in diagnosis and/or classification of the disease 20,27 . Furthermore, whereas serum IgG4 levels are a valuable tool to disclose multi-systemic disease and risk of relapse, complement levels may predict renal involvement 16,28 . However, no single serological feature is by itself specific enough for the diagnosis of IgG4-RD. Data on biomarkers to differentiate IgG4-RD from other conditions have been conflicting. For instance, Akiyama et al. described that serum soluble IL-2 receptor was not useful for the differential diagnosis between IgG4-RD and SS 17 . On the contrary, a recent study by Umeda et al. found that serum thymus and activation-regulated chemokine could discriminate between IgG4-RD and SS 29 . In our study we did not find differences in sFLC concentrations among IgG4-RD group and both control groups, however, up to half of the patients with active IgG4-RD had an elevated κ/λ ratio.
As has been described for other systemic autoimmune diseases such as SLE, SS, RA and SSc 4-7 , measurement of sFLC may be a useful biomarker to assess systemic activity in IgG4-RD, as demonstrated by our findings of higher levels of κ sFLC and κ/λ ratio in patients with active vs. inactive IgG4-RD. Interestingly, only two patients with inactive disease had an elevated κ/λ ratio. Furthermore, half of the patients with an elevated κ/λ ratio had normal serum IgG4 levels, thus, an elevated κ/λ ratio could point to active disease even in those patients with normal serum IgG4. Interestingly, most of the active patients with an elevated κ/λ ratio had de novo active disease.
In certain clinical scenarios after treatment of IgG4-RD, it may be difficult to assess whether a clinical or imaging finding is a reflection of ongoing organ inflammation and activity or permanent scarring and damage; this is especially true for clinical manifestations of the fibrotic phenotype such as retroperitoneal fibrosis and some cases of orbital pseudotumor. In other settings, biochemical findings reflecting permanent damage such as persistent elevation of alkaline phosphatase or urine protein may lead to the false reassurance of ongoing biliary tract and kidney activity followed by unnecessary treatment. Thus, a biomarker for disease activity in IgG4-RD may be particularly beneficial in these situations. The ROC curve generated by the results in the present study showed that the cut-off value of 33.42 mg/L for κ sFLC and 1.61 for the κ/λ ratio may discriminate between www.nature.com/scientificreports/ active and inactive IgG4-RD with a high specificity and positive predictive value, although with a low sensitivity. Interestingly, the AUC for de novo active disease had a better performance, while this was not true for relapsing disease. The latter finding may be due to the smaller number of patients with relapsing disease compare to de novo active disease included in our study.
In agreement with the aforementioned findings, we found a positive correlation between the number of organs involved and the IgG4-RD RI with both the κ sFLC and the κ/λ ratio. Furthermore, we observed higher levels of sFLC in patients with lymph node involvement. As suggested by Grados et al. these findings may correlate with the volume of the polyclonal "tumoral" infiltrate of IgG4-RD 20 , as has been reported for multiple myeloma and AL amyloidosis where sFLC levels correlate with bone marrow infiltration by plasma cells and the number of organs involved by amyloid, respectively 30,31 .
Nowadays, classifying patients with IgG4-RD in different phenotypic groups is appealing 11,23 . Differences in laboratory parameters among different subsets is well stablished in terms of IgG, IgG4, IgE, eosinophils and complement levels. It was tempting to hypothesize that patients belonging to the Mikulicz and systemic and the proliferative phenotype would display higher sFLC levels, as these phenotypes more frequently present with polyclonal hypergammagobulinemia; however, we did not find differences in sFLC levels among the distinct IgG4-RD phenotypes.
Interestingly, we observed that concentrations of both κ and λ sFLC were higher in IgG4-RD patients with kidney involvement, although this observation was not driven by a low eGFR. The ROC curves showed that the cut-off value of 33.42 mg/L for κ sFLC had an acceptable sensitivity and specificity, with a negative predictive value of 90%, rendering κ sFLC as an excellent tool to exclude kidney involvement.
Our study is not exempt of limitations. First, the study had a cross sectional design and thus we did not measure sFLC prospectively to assess their utility to predict relapses and treatment response. Second, serum and urine electrophoresis and immunofixation was not routinely done in our patients to exclude a monoclonal plasma cell disorder, nevertheless, we have no doubt about the certainty of IgG4-RD diagnosis. Third, although some patients were under immunosuppressive therapy at the time of recruitment, we did not find differences in sFLC levels among patients with vs. without such therapy. Finally, longitudinally studies are needed to better define if sFLC levels are sensitive to change under immunosuppressive therapy and for predicting relapsing disease.
Some strengths also need to be mentioned. First, this is the first study to evaluate the usefulness of sFLC in cohort of IgG4-RD according to the clinical phenotype and activity status. Second, we included a control group with other systemic autoimmune diseases that commonly mimic IgG4-RD in order to assess their value as a complementary diagnostic tool. Nevertheless, the performance of sFLC might be further evaluated in other IgG4-RD mimickers such as malignancies or ANCA-associated vasculitides. www.nature.com/scientificreports/

Conclusions
In summary, a significant proportion of patients with IgG4-RD may present with high levels of sFLC and, even though their assessment may not be useful for distinguishing from SS and sarcoidosis. Measurement of sFLC may be useful as a biomarker of disease activity and multiorgan and renal involvement. In particular, a high κ/λ ratio may identify patients with active disease even in those with normal IgG4 levels.

Methods
Patients. This  We retrospectively collected patient information such age at diagnosis, number of organs ever involved, IgG4 serum levels at diagnosis and/or during the disease course, biopsy availability and the use of glucocorticoids and immunosuppressive therapy.
Patients were classified according to clinical phenotypes as described by Wallace et al. in pancreato-hepatobiliary, retroperitoneum and aorta, head and neck-limited and Mikulicz and systemic phenotypes 11 ; patients with only mesentery or mediastinal involvement were included in the retroperitoneum and aorta phenotype; and patients who could not be fitted in one of these clinical phenotypes were termed "unclassifiable" phenotype. Patients were also classified according to the clinical phenotypes described by Zhang et al. in proliferative and fibrotic phenotypes 35 . We assessed the number of involved organs and the IgG4-RD Responder Index (IgG4-RD RI) at recruitment 36 .
We defined active disease as the presence of clinical signs and symptoms, laboratory abnormalities or unequivocal radiological findings attributable to IgG4-RD. Patients with active disease included those with de novo active disease (first episode of active IgG4-RD), relapsing disease (active disease after a period of inactive IgG4-RD) and persistent active disease (active disease despite immunosuppressive treatment). We defined inactive disease as the absence of clinical signs and symptoms, laboratory abnormalities and new unequivocal radiological findings attributable to IgG4-RD. Damage was defined as irreversible organ dysfunction or failure caused by IgG4-RD or as a consequence of surgical procedures performed to diagnose or treat IgG4-RD 36 . sFLC assessment. Blood samples from patients and controls were collected. Serum samples were extracted and frozen at − 20 °C until processing. sFLC levels were measured in the same laboratory using latex-enhanced immunoassay (Freelite, The Binding Site, UK) using turbidimetry (SPA PLUS ). The immunoassay consisted of two separate measurements, one for free κ and one for free λ and then the automatic calculation of the κ/λ ratio. Normal values were set according to the manufacturer's recommendations as follows: normal free κ range: 3.3-19.4 mg/L; normal free λ range: 5.7-26.3 mg/L; and normal κ/λ ratio range: 0.26-1.65 37 . In patients with IgG4-RD, serum IgG1 and IgG4 levels were determined with turbidimetry analyzer (SPA PLUS , Freelite, The Binding Site, UK) and normal values were also set according to the manufacturer's recommendations as follows: IgG1 normal range: 382.4-928.6 mg/dL; IgG4 normal range: 3.9-86.4 mg/dL. eGFR was calculated with the Chronic Kidney Disease Epidemiology Collaboration equation at the time of recruitment.