Elevated IgM and abnormal free light chain ratio are increased in relatives from high-risk chronic lymphocytic leukemia pedigrees

Abnormal serum immunoglobulin (Ig) free light chains (FLC) are established biomarkers of early disease in multiple B-cell lymphoid malignancies, including chronic lymphocytic leukemia (CLL). Heavy chains have also been shown to be biomarkers in plasma cell disorders. An unanswered question is whether these Ig biomarkers are heritable, i.e., influenced by germline factors. CLL is heritable but highly heterogeneous. Heritable biomarkers could elucidate steps of disease pathogenesis that are affected by germline factors, and may help partition heterogeneity and identify genetic pleiotropies across malignancies. Relatives in CLL pedigrees present an opportunity to identify heritable biomarkers. We compared FLCs and heavy chains between relatives in 23 high-risk CLL pedigrees and population controls. Elevated IgM (eIgM) and abnormal FLC (aFLC) ratio was significantly increased in relatives, suggesting that these Ig biomarkers are heritable and could offer risk stratification in pedigree relatives. Within high-risk CLL pedigrees, B-cell lymphoid malignancies were five times more prevalent in close relatives of individuals with eIgM, prostate cancer was three times more prevalent in relatives of individuals with aFLC, and monoclonal B-cell lymphocytosis increased surrounding individuals with normal Ig levels. These different clustering patterns suggest Ig biomarkers have the potential to partition genetic heterogeneity in CLL and provide insight into distinct heritable pleiotropies associated with CLL.


Introduction
Chronic lymphocytic leukemia (CLL) is the most common leukemia affecting adults diagnosed in the United States (4.7/100,000 per year), with similar incidence in other western countries. There are very few known risk factors for CLL. These include family history, sex, race/ ethnicity (risk is greatest in non-Hispanic White men 1 ), and exposure to certain chemicals 2 . Of these, family history is the strongest risk factor, with large population-based studies consistently suggesting CLL as among the most familial cancers studied 3,4 and suggesting a striking 5.7-7.8 increased risk in first-degree relatives [5][6][7] .
Biomarkers are biological traits associated with risk to a disease. Heritable biomarkers are those that share germline genetic factors with the disease and can shed light on the steps along disease pathogenesis affected by germline genetics. Heritable biomarkers have the potential to partition genetic heterogeneity (different routes to disease) and identify germline pleiotropies (same heritable biomarker, different disease endpoint). In complex diseases, heterogeneity is a key obstacle that challenges genetic discoveries, and hence the deconstruction of heterogeneity and definition of important pleiotropies will aid in the elucidatation of germline risk factors. Of clinical benefit, knowledge of early disease processes have been shown to be associated with better overall survival in multiple myeloma 8 . Future clinical benefit of heritable biomarkers in CLL could include risk stratification in relatives of diseased individuals toward prevention, early detection, and improved outcomes. A better understanding of heritable immune phenotypes also may have future potential to contribute to the growing knowledgebase concerning immunophenotypes and response to immunotherapies 9 .
High-risk CLL pedigrees are families containing a significant excess of relatives with CLL. Focusing on "atrisk" relatives in CLL families is a powerful design to explore genetically controlled disease initiation. In particular, "at-risk" relatives can provide novel insight into heritable biomarkers. Monoclonal B-cell lymphocytosis (MBL), e.g., is the existence of small clones in peripheral blood (< 5 × 10 9 clonal B-cells/L) and is a precursor to CLL 10 . High-risk CLL pedigrees have shown that MBL is a common occurrence (17%) in otherwise unaffected first-degree relatives 11 , compared with only 3-5% in the general population using comparable laboratory detection methods 12,13 . Thus, MBL is a heritable precursor sharing genetic susceptibility with CLL and indicates that inherited risk includes propensity for clonal development.
B-cells produce immunoglobulin (Ig) as part of the native and adaptive immune system. Aberrant B-cell activation and expansion may be reflected in abnormal polyclonal or monoclonal production of Ig heavy chains (e.g., IgA, IgG, and IgM) and/or light chains (κ and λ). This includes intact Igs comprising bound heavy light chains (HLC) and unbound free light chains (FLC) due to more abundant light-chain synthesis. Abnormal polyclonal (κ + λ) or monoclonal (κ/λ) FLCs in serum have been shown to be biomarkers of prognosis and survival in plasma cell disorders [14][15][16][17][18] and other B-cell lymphoid malignancies, including CLL [19][20][21][22][23][24] . Prospective cohorts have identified elevated polyclonal and abnormal monoclonal FLCs as early biomarkers of CLL, detectable upto 9.8 years before diagnosis 25 . More recently, assays to quantitate serum Ig HLCs have been developed 26 . In plasma cell disorders, HLCs have also proven to be useful biomarkers of progression, prognosis, and survival, independent of FLC markers 18,[27][28][29][30][31][32] . The role of HLC biomarkers has yet to be explored in other B-cell malignancies. Here we explore the hypothesis that inherited risk extends to early steps in CLL pathogenesis, defective immune response (polyclonal expansion), and the development of clonal populations. We investigate the potential for heritability of polyclonal and monoclonal FLCs and HLCs using "atrisk" relatives in CLL pedigrees.

Materials and methods
Case definition CLL is characterized by infiltration of the bone marrow by a clonal population of mature, normal appearing but immunophenotypically abnormal B-cell lymphocytes. It is diagnosed by the presence of at least 5 × 10 9 clonal Bcells/L in the peripheral blood, but the lymph nodes, spleen, and liver may also be involved. If diagnosed instead by clonal cells in the lymph nodes, it may be referred to as small lymphocytic lymphoma (SLL). These are considered the same malignancy and we use "CLL" hereafter as an abbreviation to refer to either CLL (International Classification of Disease for Oncology version 3 [ICD-O-3] histology code 9823/3) or SLL (ICD-O-3 9670/3). Relatives of CLL cases identified from the state-wide Utah Cancer Registry (UCR) and who were members in high-risk pedigrees (see below) were the focus of this study.

High-risk pedigrees and controls
We previously ascertained high-risk CLL pedigrees using the Utah Population Database (UPDB) 11 . The UPDB record links a 5 million-person genealogy to the UCR, a state-wide registry since 1966, which has been part of the National Cancer Institute Surveillance, Epidemiology, and End Results Program since its inception. High-risk pedigrees are defined as those containing a statistical excess of CLL (p < 0.05), compared with sex-, birth cohort-, and birth place-matched rates from the full UPDB dataset. Biospecimens for living CLL cases and their relatives were previously collected in 23 high-risk pedigrees for genetic study. Serum samples were available for 117 relatives of CLL cases; mostly first degree (98%). At the time of sampling, some pedigree relatives were known to have other B-cell lymphoid malignancies, solid tumors, or screen-detected MBL 11 . Table 1 summarizes pedigree characteristics and Figure S1 illustrates the 23 pedigrees. Serum samples for 160 population controls with no personal history of hematological malignancies were also available. Some controls were known to have solid tumors at the time of sampling. All sampled relatives and controls provided informed consent and self-reported as White, non-Hispanic. All ascertainment was performed under University of Utah institutional review board approvals.

Ig measurements
Serum was extracted from whole blood samples collected in red-top Vacutainer tubes. Blood samples were allowed to clot at room temperature for~20 min after which the clot was removed by centrifugation at 1000-2000 g-force for 10 min. The resulting serum was transferred to polypropylene tubes and stored at − 20°C. Stored serum was thawed on ice and concentrations of FLCs (κ and λ) and isotype-specific HLCs (IgAκ, IgAλ, IgGκ, IgGλ, IgMκ, and IgMλ) were measured at The Binding Site Ltd (Birmingham, UK) with Freelite® and Hevylite® assays, respectively. The Freelite® is a wellestablished assay that gained wide use in the diagnosis, treatment assessment, and disease monitoring of plasma cell disorders. The newer Hevylite® assay allows for quantification of IgM, IgA, and IgG bound to specific light chains using serum markers for isotype-specific Ig HLC junction epitopes 26 .

Statistical analyses
For quality control, individuals with Ig values >5 SDs beyond the mean for their group were excluded as Pedigree structures were defined to include the following: all known CLL cases from the UPDB that determined the high-risk status, all siblings of the CLL cases, all sampled relatives used in this study and their siblings, and all connecting relatives to the common ancestral founder/s. Column headings: Members refer to the total number of individuals in each pedigree structure; Generations is the number of generations from the pedigree founder/s; CLL indicates the number of individuals diagnosed with CLL or SLL in the pedigree; Sampled relatives indicate the number of relatives with serum used in this study; Heme malignancies represent a count of bloodline pedigree members (i.e., not including marry-ins) in each pedigree with a UCR-confirmed hematological malignancy (not including CLL/SLL); B-cell lymphoid malignancies indicate the number of bloodline pedigree members with a UCR-confirmed B-cell lymphoid malignancy (not including CLL/SLL); and Solid tumors indicate the count of UCR-confirmed solid tumors in bloodline pedigree members. Note: individuals with multiple cancers will be counted more than once outliers. Propensity-score adjustment with optimal match selection on sex and age was used to control imbalances and nonlinearity in sex and age distributions between the relatives and the controls 35 . Associations between each Ig variable and "at-risk" relative status were performed using propensity-score-adjusted linear (for continuous polyclonal variables) and logistic (for dichotomous monoclonal variables) regression models. Firth's bias reduction method was employed for estimation of logistic regression parameters to address low observed counts 36 . To explore the independence of multiple Ig variables identified from univariable analyses, multivariable models with correction for age and sex were constructed. Although increases in "at-risk" relatives are of particular interest (consistent with our hypothesis regarding heritability), all tests were two-sided. A nominal type 1 error threshold (a = 0.05) for statistical significance was used throughout. For Ig phenotypes found to be enriched in relatives, we further performed pedigree-only analyses (not including the control data). We categorized pedigrees as harboring aberrant Ig phenotypes or not. Based on these pedigree categories, we explored cancers in the relatives surrounding individuals with abnormal Ig phenotypes and tested differences between the pedigree categories. Such comparisons can identify cancer spectra that share germline susceptibility with biomarkers of CLL and differences may provide insight into CLL genetic heterogeneity. These pedigree comparisons were possible due to the unique UPDB resource, which provides detailed cancer family history for individuals (both sampled and unsampled) within the high-risk pedigrees via recordlinkage to the UCR; a state-wide surveillance of cancer began in Utah in 1966. Cancers observed in pedigree individuals with aberrant Ig phenotypes and their firstto third-degree relatives were counted (individuals that married-in to the pedigree were excluded). These were compared with cancer counts in and surrounding individuals with normal Ig phenotypes in pedigrees without any aberrant phenotypes. First-to third-degree relatives (i.e., out to cousins) were considered to balance contributions across pedigrees, as some pedigrees are much larger than others ( Figure S1). Differences were illustrated using odds ratios (ORs) with 95% confidence intervals (CIs), using Fisher's method to address small counts.

Results
One control individual was removed as an outlier, resulting in a set of 117 at-risk relatives and 159 population controls for analysis. Summaries for the demographics for these two groups are shown in Table 2 and their Ig values are shown in Table 3.

Association of "at-risk" relatives and Ig variables
Polyclonal variables were analyzed as quantitative traits. Results are shown in Table 4. A statistically significant difference in IgM levels was found between high-risk pedigree relatives and controls (p = 0.033, after adjusting for age and sex), indicating a 16.0 mg/dL increase in relatives (95% CI 1.4-30.5). Defining elevated IgM (eIgM) to be a value beyond the 95th percentile from the control reference range (IgM(κ + λ) ≥ 19 mg/dL), 11.1% of the "atrisk" relatives had eIgM (13/117). When individuals with known MBL, hematological malignancies, or solid tumors at the time of sampling were removed, results were similar and remained significant (data not shown). No statistically significant differences were found for polyclonal variables FLC(κ + λ), IgA(κ + λ), or IgG(κ + λ).
Monoclonality was determined by abnormal κ/λ ratios based on the established diagnostic range for the FLC ratio or control 95% reference ranges (for IgA, IgG, and IgM). Results are shown in Table 4. No statistically significant differences were found for monoclonality of any of the HLCs. aFLC ratios were significantly increased in the pedigree relatives compared with controls (OR = 6.38, 95% CI 1.65-34.97, p = 0.0063). The result remained similar and significant when individuals with known MBL or cancers were removed. Figure 1 illustrates FLC(κ/λ) ratio by age at sampling, showing a steeper trajectory in the "at-risk" relatives than in the controls. Increasing FLC ratio with age is highly significant for relatives (increase of 0.10 per 10 years, p < 0.001), a 2.7-fold higher rate of increase than in the population controls (0.038 per 10 years, p = 0.029).
To establish whether IgM level and aFLC ratio were independent biomarkers, we performed a multivariable analysis with adjustment for age and sex. Both eIgM and aFLC ratio were independent predictors and remained significant in the full model: eIgM OR = 1.61 (95% CI

Comparison of cancer configurations within pedigrees
In pedigree-only analyses, we identified the pedigrees that harbored aberrant phenotypes and those that did not. Five pedigrees contained eight members with aFLC ratios. These 8 individuals together had 93 unique firstto thirddegree relatives. Eight pedigrees contained 13 members with eIgM levels, who together had 132 firstto thirddegree relatives. Cancers in these individuals/relatives were compared with 46 individuals with normal Ig phenotype and their 257 unique firstto third-degree relatives residing in 13 "normal" pedigrees (those not containing any members with abnormal Ig phenotypes). Table 5 summarizes cancer configuration comparisons between pedigrees. B-cell lymphoid malignancies (not including CLL or SLL) were 5 times more likely to be observed in the 145 relatives and individuals with eIgM compared with the 303 relatives and "normal" individuals (OR = 5.05, 95% CI 1.13-30.73; 4.8% vs. 1.0%). B-cell lymphoid malignancies were also more likely to be observed in and surrounding individuals with aFLC ratio than in the normal pedigrees (4.0% vs. 1.0%; OR = 4.15 95% CI 0.69-28.83), although not statistically significant. Surprisingly, even though all pedigrees were high risk for CLL by design, there was a greater proportion of relatives with CLL in "normal" pedigrees (14.8% vs. 11.4% and 7.6% surrounding individuals with eIgM and aFLC ratio, respectively). In addition, MBL was more likely to be  observed in the firstto third-degree relatives and individuals with normal Ig phenotypes (4.6% compared with 3.0% and 0.7% in the individuals and relatives with aFLCr or eIgM, respectively). In terms of solid tumors, prostate cancer was over three times more likely to be observed in the male individuals with aFLC ratios and their male relatives (OR = 3.19, 95% CI 1.21-8.40; 21.8% vs. 8.0% in and around "normals"). In contrast, prostate cancer was not increased in relatives and individuals with eIgM (OR = 0.72; 5.9% vs. 8.0%). The observations of increased hematological malignancies (eIgM and aFLC ratio) and prostate cancer (aFLC ratio only) remained very similar when the personal cancer history of the individuals themselves were removed from the comparison.

Discussion
Family history is the strongest known risk factor for CLL [5][6][7] . The high-risk pedigree design is therefore an effective design to explore heritable biomarkers. Previously, a high-risk CLL pedigree design showed that MBL was a heritable precursor, significantly increased in pedigree relatives 11 , suggesting that a component of inherited risk may be to clonal development. Here we used a highrisk CLL pedigree design to investigate polyclonal and monoclonal Ig phenotypes as heritable biomarkers for CLL. We found that aFLC ratios and IgM levels were significantly increased in "at-risk" pedigree relatives compared with controls, and that these were independent factors.
A role for aFLC ratio as a precursor biomarker for CLL was first suggested in a prospective study, with monoclonality observed in 38% of individuals subsequently diagnosed with CLL, and detected upto 9.8 years, before diagnosis 25 . Our observation that aFLC ratio is more prevalent in "at-risk" relatives in pedigrees compared with controls (6.8% vs. 1.3%, OR = 6.38, p = 0.0063) is consistent with aFLC ratio as a heritable CLL precursor biomarker. Abnormal FLC ratio is used clinically to indicate clonal plasma cells in plasma cell disorders. There is also evidence that aFLC ratio is associated with survival and time to first treatment in CLL [19][20][21][22][23] , suggesting it as an important biomarker for clonal B-cell development. In the context of a precursor, a biomarker for active clonal development has the potential to capture important information on progression toward overt disease. Further investigation of aFLC ratio as an effective risk stratification tool for early disease, particularly in atrisk relatives, may be warranted.
HLCs have not previously been studied as a precursor or prognostic indicators for CLL. In this study, we found that polyclonal IgM was elevated in pedigree relatives compared with controls, and that this association was independent from a monoclonal FLC ratio. Serum IgM has several important roles, including the clearance of pathogens and apoptotic cells, and regulation of immune response 37 . Failure to remove dying cells is essential to prevent autoimmunity and inflammation 38 , and a role for natural IgM has been described in various autoimmune and inflammatory diseases 39 , and may contribute to the co-clustering that has been reported for B-cell malignancies and autoimmune disorders [40][41][42] . In Waldenstrom's macroglobulinemia, a B-cell malignancy with similarities to CLL, evidence for polyclonal IgM levels was reported in unaffected relatives in high-risk families, some of which subsequently progressed to monoclonal IgM 43 . Interestingly, the most significant common risk allele for CLL from genome-wide association studies, rs735665-A at 11q24.1 (OR = 1.63, 95% CI 1.53-1.72) 44 , was recently associated with an increase in natural IgM levels 45 . Furthermore, secreted IgM has been shown to accelerate CLL tumor progression in mouse models 46 . It is notable, however, that although CLL tumors are predominantly IgM producing, it is generally membrane bound and not secreted into the serum. The significance of polyclonal IgM as a heritable precursor biomarker for CLL requires further investigation. Attention to IgM-secreting CLL patients regarding progression may also be warranted.
Our evidence for two new heritable biomarkers for CLL, elevated polyclonal IgM and monoclonal FLC ratio, suggests that genetic susceptibility may include defective immune response (polyclonal expansion) and development of clonal populations. Of further interest is whether these two biomarkers represent the same underlying genetic risk observed along the continuum of CLL development or different avenues of genetic risk (genetic heterogeneity). To begin to address this, we explored the individual and family cancer history of individuals with these two abnormal Ig phenotypes. We found that B-cell malignancies were significantly increased surrounding individuals with eIgM, compared with other high-risk pedigrees where eIgM was not apparent. This observation is consistent with overlapping genetic risk factor/s across A "normal" pedigree indicates one that does not contain any members with abnormal Ig phenotypes in this study. Index individuals indicate those with eIgM, aFLC ratio, or those with normal Ig phenotypes (in "normal" pedigrees). 95% CI 95% confidence interval, OR odds ratio, nd not determined (counts too low to assess). All ORs and CIs are from Fisher's exact method. Percent CLL (in firstto third-degree relatives). Percent MBL and percent other cancers (in the sum of index individuals and firstto third-degree relatives). Supplementary Figure S1 illustrates the pedigree drawings and shows all the UCR-confirmed cancer diagnoses from which these summaries are made. All statistics for prostate, breast, and gynelogical cancer are sex specific hematological malignancies. Also consistent with this conjecture, was that pedigrees without eIgM (or aFLC) were more homogeneous for CLL (harboring both a higher proportion of CLL and MBL, and a lower proportion of other B-cell lymphoid malignancies). A unique feature to the family cancer history in and surrounding individuals with aFLC ratio was a significantly increased prevalence of prostate cancer. This was in striking contrast to the reduction of prostate cancer observed in and surrounding individuals with eIgM. Co-clustering of CLL and prostate cancer in families has been observed repeatedly [47][48][49] , and prostate cancers have been observed as increased secondary cancers in individuals with CLL [50][51][52] . Here we find that relatives with aFLC ratio partition our high-risk pedigrees into those exhibiting particularly strong co-clustering with prostate cancer and not, and suggests that it may represent a biomarker that parses shared genetic etiology between CLL and prostate cancer. Based on these observations, we hypothesize that eIgM and aFLC ratios are heritable biomarkers that represent separate genetic susceptibilities for CLL and as such may also be useful as a strategy to reduce heterogeneity in genetic mapping studies.
Our high-risk pedigree study has unique strengths, particularly the rich cancer family history available from the UPDB/UCR and the ability to study large, multigenerational families. It is limited in the fact that it is a single study, which is largely based on a White, non-Hispanic population. All findings will require replication elsewhere. Nonetheless, our novel findings complement other observations in the literature and the hypotheses generated provide new and exciting avenues for the field to consider.
In conclusion, we propose eIgM and aFLC ratios as two new heritable biomarkers for CLL representing separate genetic susceptibilites. Together with MBL, these biomarkers suggest that genetic risk involves susceptibility to polyclonal expansion, clonal development, and clonal expansion. These biomarkers could be useful for developing risk stratification strategies for relatives of CLL cases to identify early disease. In addition, these biomarkers in the relatives in pedigrees may offer a strategy to reduce genetic heterogeneity providing new opportunities to increase power in pedigree-based gene mapping studies. Continued longitudinal studies of high-risk pedigrees has the potential to continue to uncover important steps in the pathogenesis of CLL and other lymphoproliferative disorders.