Association of elevated serumfree light chains with chronic lymphocytic leukemia and monoclonal B-cell lymphocytosis

Chronic lymphocytic leukemia (CLL) and its precursor, monoclonal B-cell lymphocytosis (MBL), are heritable. Serumfree light-chain (sFLC) measures are a prognostic factor for CLL, but their role in susceptibility to CLL is not clear. We investigated differences between sFLC measurements in pre-treatment serum from five groups to inform the association of sFLC with familial and sporadic CLL: (1) familial CLL (n = 154), (2) sporadic CLL (n = 302), (3) familial MBL (n = 87), (4) unaffected first-degree relatives from CLL/MBL families (n = 263), and (5) reference population (n = 15,396). The percent of individuals having elevated monoclonal and polyclonal sFLCs was compared using age-stratified and age- and sex-adjusted logistic regression models. In age groups >50 years, monoclonal sFLC elevations were increased in sporadic and familial CLL cases compared to the reference population (p’s < 0.05). However, there were no statistically significant differences in sFLC monoclonal or polyclonal elevations between familial and sporadic CLL cases (p’s > 0.05). Unaffected relatives and MBL cases from CLL/MBL families, ages >60 years, showed elevated monoclonal sFLC, compared to the reference population (p’s < 0.05). This is the first study to demonstrate monoclonal sFLC elevations in CLL cases compared to controls. Monoclonal sFLC levels may provide additional risk information in relatives of CLL probands.


Introduction
Chronic lymphocytic leukemia (CLL) has an underlying heritable predisposition, with~10% of individuals affected with CLL reporting a first-degree relative with CLL or a related lymphoproliferative disorder [1][2][3] . Monoclonal Bcell lymphocytosis (MBL), precursor disease to CLL, is an asymptomatic hematologic condition characterized by small absolute levels of blood B-cell clone and no other signs of a lymphoproliferative disorder [3][4][5][6][7] . In the general population, MBL increases with age with a prevalence of 5-9% in individuals over age 60 years 7 . We have reported MBL to be higher among first-degree relatives from CLL families, occurring at a higher rate in high-risk CLL families, implying a shared inherited risk 4 .
Paraproteins, usually aligned with plasma cell disorders, have also been found to be prevalent in other B-cell malignancies, such as CLL 8 . In 2007, Martin et al. 9 evaluated the frequency of monoclonal serum-free light-chain (sFLC) measurements in patients with other B-cell malignancies and established abnormal sFLCs can be detected in a substantial proportion of patients with Non-Hodgkin's lymphoma (NHL) and CLL, and may be a useful clinical tool in the early diagnosis of a B-cell malignancy. Since then other studies have shown that sFLCs and the associated κ/λ ratio (rFLC) can also be used to inform prognosis, survival, and clinical outcomes in patients with CLL [10][11][12][13][14][15][16][17] . Maurer et al. 13 found that both monoclonal and polyclonal sFLC elevations were associated with poor overall survival in CLL compared to patients with normal sFLCs, even after adjusting for Rai stage. A prospective study showed an abnormal rFLC can be detected several years before the actual diagnosis of CLL in a significant percentage of patients 18 . Thus, sFLC measurements have been implicated in detection and prognosis of CLL, however, few studies evaluating sFLCs and risk of CLL (or MBL) have been performed.
Further, there is limited evidence on the differences between sporadic and familial CLL, with most studies focusing on prognostic and clinical factors. Stage at initial diagnosis, need for treatment, and survival over a 10-year period have been reported to be similar in sporadic and familial cases 19 . Similarly, no differences in the expression of ZAP-70, CD38, and CD23, and levels of serum β2microglobulin have been reported 20 . Studies that examined the mutational status of genes in the immunoglobulin heavy chain variable region (IgVH), known to be associated with a better prognosis, found higher frequency of mutated CLL in familial compared to sporadic CLL 21 . Another study found differences in frequency of VH gene segments (4-34, 5-51, 1-69, 4-29, 3-23) in familial vs. sporadic disease 22 . Among familial cases, VH gene segment utilization proved non-random and diverged from the frequencies previously reported among unrelated patients with CLL 22 . However, the association of sFLC with familial and sporadic CLL has not been performed.
In this study, for the first time, we investigated the association of sFLC levels with risk of CLL (or MBL) using CLL cases (familial and sporadic), unaffected relatives, and a large comparable reference population. We hypothesized that pre-treatment sFLC elevation would be higher among familial vs. sporadic CLL, and in CLL cases compared to the reference population. Second, given the known familial nature of CLL, we hypothesized that pretreatment sFLC would be elevated in the family members of patients with CLL (both unaffected relatives and those with MBL) compared to the reference population.

Study participants
All participants provided written informed consent. The protocol was approved by the Institutional Review Board at each participating center. Individuals with pretreatment serum samples were eligible. Those known to have died within 1 year of sample date were excluded due to likelihood of underlying disease, which could falsely inflate sFLC levels. Five participant groups (primarily Caucasian) were defined for analyses consisting of three case groups and two control groups: (1) Sporadic CLL cases, (2) Familial CLL cases, (3) Familial MBL cases from CLL / MBL families, (4) Unaffected relatives from CLL/ MBL families; and (5) a reference population (Olmsted County controls) [23][24][25][26] .

Sporadic CLL cases
Sporadic CLL cases consisted of incident CLL cases diagnosed from 2002-2008 without a self-reported family history of CLL and/or MBL and with a pre-treatment sFLC measure in the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource (MER) 13 .

Familial CLL cases
Familial CLL cases were selected from families with at least two reported CLL and/or MBL cases. Families were recruited through the Genetic Epidemiology of CLL (GEC) consortium 27

Familial MBL cases
In our GEC families described above, relatives were screened for MBL in accordance with our previous work 4 . MBL status was determined by flow cytometry and was defined at the time of screening. Any blood relative found to have MBL in a family with two or more CLL and/or MBL cases was eligible for inclusion. We did not require a specific degree of relation between cases.

Unaffected relatives from CLL/MBL families
Family members with no known CLL, MBL, or any other lymphoproliferative disorder who were a firstdegree relative to at least one CLL or MBL case from a familial CLL family were considered unaffected relatives. Primarily, MBL screening was performed on a baseline sample, which was also used for the sFLC measurements; however, there were some samples that did not get screened and were assumed to be unaffected.

Reference population: Olmsted County (controls)
The reference population was a previously characterized population-based cohort of Olmsted County residents age 50 years or older 23 . From 1 January, 1995, to 21 November, 2003, serum samples were obtained from 21,462 (76.6%) of 28,038 enumerated Olmsted County residents and tested for monoclonal protein levels and FLC. We refer to this group as "controls" as they have no known diagnosed CLL or precursor disease, MBL.
Of the 21,462 controls, we excluded 577 individuals with known plasma cell disorders, including those with monoclonal gammopathy of undetermined significance (MGUS), because these are the disorders in which clonal elevations of FLC are known to occur 28 . However, we did not exclude light chain (LC)-MGUS, since FLC is used in the definition of this condition. We also excluded 4,096 individuals who declined authorization to participate in research 29 and 758 who did not have sufficient sample volume to perform the FLC assay. Of the remaining 16,031 subjects, we also excluded individuals with a known lymphoproliferative disorder or precursor (n = 62) or a first-degree relative with this condition (n = 32) from clinical or research databases. After excluding individuals whose death date was within 1 year of sample date (prevalent disease: n = 541), the final control group consisted of 15,396 subjects.

Free lightchain testing
Heparinized plasma or serum samples were tested using the sFLC assay (FREELITE™, The Binding Site Ltd., Birmingham, UK) to measure κ and λ immunoglobulin FLC levels at either the Mayo Clinic or National Cancer Institute (NCI) 23 . Prior studies have shown the high intraclass correlation between FLC measures using fresh and stored serum and for the FREELITE™ kappa and lambda free light chains assay; both EDTA-plasma and lithium heparin-plasma can serve as acceptable substitutes for serum 30,31 . As previously described, the analytic sensitivity for the nephelometric FLC assay is 0.1 mg/ L for a monoclonal κ or λ FLC 32 . For samples analyzed at both NCI and Mayo Clinic, elevated FLC was defined as a κ or λ level above the reference range (κ > 19.4 mg/L, λ > 26.3 mg/L) 13,33 . Likewise, an elevated FLC was considered monoclonal if the rFLC (κ/λ) was outside of the normal range of 0.26-1.65 13,33 . Elevated FLC levels with normal rFLC (κ/λ) were considered as polyclonal. Patients with normal FLC levels but an abnormal rFLC (κ/λ) were considered to have ratio-only FLC abnormalities and not considered in the evaluation of elevated FLC 13 . We also evaluated the absolute value or sum of the free light chains, which has utility in non-clonal disease situations and has been shown to predict survival in the general population 23 . The timing of the plasma or serum samples was a median of 2 months between diagnosis date and blood draw date for sporadic CLL cases (N = 302) (range = (0 years-5.7 years), IQR = (2 weeks-9 months)), and a median 18 months between diagnosis and blood draw date for familial CLL cases (N = 154) (range = (0 years-25 years), IQR = (6 months-4.2 years)). Additionally, the FLC was measured in the baseline sample for the majority of subjects, therefore corresponding baseline MBL status was used. Some of the subjects may have had MBL identified in a follow-up sample, but for this study we use the MBL status at the time of the FLC measurement.

Statistical analyses and considerations
Analyses of FLC levels were conducted between the following groups: (1) Sporadic CLL cases vs. familial CLL cases; (2) Sporadic CLL cases vs. controls; (3) Familial CLL cases vs. controls; (4) Familial MBL vs. controls, and (5) Unaffected relatives in CLL/MBL families vs. controls. Age at FLC was summarized as mean ± standard error (SE), and two-sample t-tests were used for age comparisons across groups. Sex and percent elevated FLC were summarized by group as counts and percentages, respectively. The percent of individuals having elevated FLC was compared using age-and sex-adjusted logistic regression models. The FIRTH option 34 was applied to logistic regression to test for significance when there were small numbers in particular cells and estimates were thought to be unstable. Multivariable linear regression models were used to calculate the least squares (LS) means of κ, λ, κ/λ ratios, and FLC sum after adjusting for age and sex; due to the non-normal distributions of the FLC measures, we first log-transformed the values, performed the calculations, then exponentiated the values to summarize.
We performed secondary analyses to ensure that statistical differences were not due to relatedness. We randomly chose one relative from each family from each relative group (either one unaffected or one MBL) and reran analyses above.
Due to the age of the reference population, analyses that compared to the reference group were limited to individuals who were age 50 years and older. Comparisons between sporadic and familial CLL cases included cases who were age 35 and older at sample collection; sensitivity analyses were performed subsetting to age 50 years.
Analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

Study participants
We identified 302 sporadic CLL cases from MER, 154 familial CLL cases, 87 familial MBL cases, 263 unaffected first-degree relatives from GEC, and 15,396 subjects from the Olmsted County reference population, otherwise referred to as controls.

Comparison of sporadic CLL and familial CLL cases
Characteristics and comparisons between sporadic CLL and familial CLL cases are shown in Table 1. Sporadic and familial cases were similar in age, however, sporadic cases were comprised of more males (68%) compared to familial cases (49%) ( Table 1). The adjusted mean FLC sum (familial = 2.75, sporadic = 2.70; p-value = 0.73) and rFLC Likewise, when examining specific type of elevation, there were no statistically significant differences between the familial and sporadic cases (p's > 0.05).

Comparisons of familial MBL, unaffected relatives, and reference population (controls)
Characteristics and comparisons between familial MBL, unaffected relatives from CLL/MBL families, and controls are shown in Table 2. Relatives with MBL were on average older (68.4 years) than unaffected family members (63.6 years) or controls (63.9 years); however, the difference was only statistically significant for familial MBL compared to controls (p < 0.001). The proportion of females was significantly higher among the unaffected relatives (68%) compared to the controls (55%; p < 0.001). Familial MBL cases and unaffected relatives both had a higher age and sex-adjusted mean serum rFLC than the controls, although values were all within the normal range (0.26 to 1.65) 13,33 , (rFLC = 1.04 for familial MBL, 1.00 for unaffected family members, and 0.82 for controls, respectively (p < 0.001)). Similar trends were observed when the analyses were repeated using one randomly selected relative from each family per category (Supplementary Table 2).

Discussion
We demonstrate increased monoclonal elevations for CLL cases, regardless of family history, compared to a reference population. Monoclonal elevation in both sporadic and familial CLL cases was higher than in the reference population. We also found evidence for increased monoclonal elevation in relatives, including unaffected family members as well as those with MBL, compared to that of the reference population, which may provide additional risk information in unaffected and MBL relatives of CLL probands.
We report no differences in sFLC elevation between familial and sporadic CLL cases, which is consistent with prior reports that have demonstrated little or no difference in other clinical and prognostic markers between these two groups 19,20,[35][36][37][38] . For example, no differences in the expression of biologic markers (ZAP-70, CD38, and CD23), levels of serum β2-microglobulin, CXCR4 expression, or chromosome 13q deletion have been reported 20,37,38 . Thus, our results from this study add to evidence for similar biology underlying sporadic and familial CLL.
Our finding of an increased monoclonal, but not polyclonal, protein elevation among unaffected family members compared to the controls is provocative. Polyclonal FLC elevation, where the rFLC is normal, can be due to a number of causes including renal dysfunction, inflammation, or immune stimulation 39,40 , whereas a monoclonal elevation is suggestive of clonal plasma cell proliferation (e.g., multiple myeloma/MGUS) 41,42 . In a prior study of CLL, the monoclonal elevated light chain in the serum (kappa or lambda) agreed with the clonal Bcell's lightchain restriction by flow cytometry in 96% of cases 13,18 . In addition to elevated sFLC measures in our unaffected relatives, we also observed elevated sFLC measures in the familial MBL cases compared to a population-based reference. It has been shown in a prospective study that an abnormal rFLC can be present several years before the actual diagnosis of CLL in a significant percentage of patients 18 . Taken together, these findings suggest that monoclonal sFLC elevation may identify a precursor clonal process prior to the onset of CLL. Confirmation in other studies is necessary, but results could suggest incorporating FLC measures and MBL status (enhanced testing for MBL clones) to inform family members at greatest risk of progression to CLL, especially among those who are 60 years and older. Our study has a number of strengths, including a wellcharacterized collection of familial and sporadic CLL cases, all of whom had validated CLL diagnoses through medical record review, and a large comparable and systematically screened population-based reference group.
Limitations include the small sample size for some of the age groups examined and generalizability to other ethnicities due to the primarily Caucasian nature of our families and Olmsted County reference population (controls). Additionally, we make the assumption that the Olmsted County population is unaffected, as they are not screened for MBL, however to the best of our knowledge they have no known reported CLL. Regarding the unaffected relatives from the GEC study sites, primarily the same baseline sample was used for the sFLC measurements and the MBL screening, however, there were some samples that did not get screened for MBL and were assumed to be unaffected. Finally, we did not have total serum protein or the flow cytometry results available to correlate the lightchain restriction pattern with the sFLC results. However, prior literature suggests high concordance between monoclonal elevated light chain in serum and the CLL clone's lightchain restriction 13,18 . Another important consideration is we were unable to perform analyses including other prognostic parameters due to missingness for familial CLL cases. Future studies should aim to consider prognostic parameters (e.g., beta-2-microglobulin, Rai Stage, IgHV mutational status, FISH cytogenetics, CLL-IPI index) with sFLC in sporadic vs. familial context.
Our study showed that sporadic and familial CLL have similar sFLC but increased monoclonal sFLC elevation compared to controls. In addition, monoclonal elevation of sFLC was higher among unaffected relatives and relatives with MBL from CLL families than a general population. To our knowledge, this is one of the first studies to examine sFLC among relatives from CLL/MBL families and provide further evidence of the potential of monoclonal sFLC elevations as a valuable prognostic factor in CLL/MBL. Follow-up studies are needed to replicate our findings and determine the relationship between elevated sFLC levels, MBL, and future CLL risk.