Evaluating a CLL susceptibility variant in ITGB2 in families with multiple subtypes of hematological malignancies

To the editor:

The application of rapid and affordable genome sequencing technologies to families with a high incidence of hematological malignancies (HMs) provides significant opportunities for understanding the genetic causes of these cancers.¹  In one of the few studies published to date using this approach, Goldin et al recently reported that a rare variant (rs2230531) within ITGB2 is associated with genetic predisposition to chronic lymphocytic leukemia (CLL).²  In this study, whole-exome sequencing was performed in a discovery set of 20 high-risk CLL families. Analysis proceeded on a “per family basis,” to identify rare (minor allele frequency [MAF] <1% in population databases), nonsynonymous variants shared in the CLL cases, as well as obligate carriers and non-Hodgkin lymphoma cases. In 20 families, 926 variants were prioritized, of which 2 were identified in multiple families, rs2230531 in ITGB2 and rs201725177 in SWI5.

The integrin gene ITGB2 was selected for resequencing in 39 additional small CLL families and 173 unrelated CLL patients from a clinical cohort. Carriers of the variant were identified in 4 families and 3 of 173 unrelated CLL patients. In summary, this variant was detected in 6 of 59 CLL families and 3 CLL cases from a clinical cohort, all with a family history of lymphoid malignancy.² 

Integrins are broadly known for their role in cancer³ ; however, mutations in ITGB2 have not previously been implicated in CLL. Variation in ITGB2 presents as an interesting biological candidate for the genetic susceptibility of CLL. Mutations in ITGB2 have been shown to cause leukocyte adhesion deficiency type 1 (OMIM: 116920), an autosomal-recessive disorder where neutrophil function is affected and patients experience recurring bacterial and fungal infections.⁴  Across 53 tissues in the Genotype-Tissue Expression (GTEx) RNA-sequencing expression database, ITGB2 is primarily expressed in blood,⁵  supporting a role for this gene in diseases of this tissue type. Additionally, Rotunno et al have previously reported this variant in a whole-exome sequencing study of familial Hodgkin lymphoma.⁶ 

Here, we examine the ITGB2 variant rs2230531 in an independent cohort of HMs from the Tasmanian Familial Hematological Malignancies Study (TFHMS). Our aim was to follow up the findings of Goldin et al in relation to the role of this variant in the genetic predisposition to HMs. The TFHMS cohort, previously described in detail,^7-9  is a collection of familial and nonfamilial HM cases and unaffected relatives from Tasmania, the island state of Australia. Using TaqMan genotyping (ThermoFisher Scientific, assay ID: C__15852956_10), we genotyped rs2230531 in 71 affected and 228 unaffected first-degree relatives from 47 families, in addition to 80 nonfamilial HM cases from the TFHMS cohort (n = 379, 48.3% male), together with 2 Tasmanian population control datasets (n = 862), detailed further in Table 1.

We identified 4 HM cases as carriers of the rs2230531 variant in the TFHMS cohort. Two HM cases carrying rs2230531 had a family history of HMs, and 2 were nonfamilial cases. Three unaffected relatives of HM cases were also identified as variant carriers. An unaffected relative and a familial HM case were identified in the same family; however, the 2 are only distantly related and separated by 12 meioses (based on genealogical records). We also detected the variant in our Tasmanian population non-HM cohorts (TPC and TCC; Table 1) with an overall MAF of 0.011. None of the 25 CLL cases screened, or their first-degree relatives, were found to carry the variant. Table 1 describes the characteristics of the individuals carrying this variant. Supplemental Table 1 (available on the Blood Web site) details HM subtypes included in these analyses.

To appropriately test the enrichment of rs2230531 in HM families compared with population controls, there is a need to account for the nonindependence of related samples in TFHMS. To do this, we used a maximum likelihood estimate (MLE) of the MAF in the TFHMS families, taking into account the pedigree structure, using Sequential Oligogenic Linkage Analysis Routines software.¹⁰  MLEs were compared across groups using a Wald test (with one degree of freedom). As an additional population control group for comparison, allele frequency data were also leveraged for rs2230531 from 26 133 samples in the non–The Cancer Genome Atlas, non–Finnish European component of the Exome Aggregation Consortium (ExAC), where the MAF is 0.007.¹¹  We additionally compared the Tasmanian population control cohort alone, and in combination with the TFHMS samples as a single Tasmanian group, against the ExAC sample set.

Table 2 shows the results of Wald test MLE comparisons. Overall, rs2230531 was not enriched in HM families in comparison with both Tasmanian population controls (P = .671) and publicly available population data from ExAC (P = .235). Additionally, rs2230531 was not enriched in Tasmanian controls or all Tasmanian samples together (TFHMS/TPC/TCC) in comparison with the non-TCGA, non–Finnish European component of ExAC (P = .141 and 0.069, respectively). A post-hoc power analysis was performed to consider the power within the TFHMS dataset to detect a difference in allele frequencies from the observed allele frequency of 0.011 in the 862 Tasmanian controls. The TFHMS dataset (of sample size 379) has 80% power to detect a difference of ∼0.023 in allele frequency, 50% for a 0.014 difference, and 25% for a 0.008 difference. The measured difference was only 0.0022923, suggesting that no biologically significant difference exists between sample groups.

These data provide evidence that the ITGB2 variant rs2230531 is not enriched in Tasmanian families with HMs in comparison with both Tasmanian and larger international population cohorts. In our cohort, we observe the variant in HM cases other than CLL, including both lymphoid and myeloid HMs, the latter being a novel observation. This suggests that if this variant does indeed contribute to HM risk (in other populations), it may contribute more broadly across HM subtypes and thus may warrant further analysis in other familial HM cohorts and HM subtypes.

An alternative interpretation of these data is that rs2230531 is a rare polymorphism not associated with disease and present in the population at large. This may be a feature of the historically stable and isolated Tasmanian population. The established founder effect in Tasmania can lead to the observation that some rare variants, reported in large publicly available databases, are enriched in the Tasmanian population (occur at a slightly higher frequency in the Tasmanian population; unpublished observation). In our non-HM Tasmanian samples, rs2230531 has a MAF of 0.011, whereas in 26 133 samples from ExAC the MAF is 0.007. This difference, however, is not statistically significant.

Our findings do not necessarily rule out a role for ITGB2 in CLL or HM genetic predisposition. HM is a multifactorial disease, and mutations in ITGB2 may play but one part in the larger context of HM risk. Our study does serve to provide a necessary sequel to the study of Goldin et al, as the evaluation of a disease-associated variant in multiple cohorts is required to establish a new genetic cause of disease, as outlined in guidelines produced by MacArthur et al.¹²  We propose that the current conflicting findings of the rs2230531 in familial HMs, together with the biological role of ITGB2 in blood cells, creates an a priori hypothesis that genetic variation in ITGB2 contributes to HM risk. Further, this finding should encourage other researchers in the field to step beyond the single variant and perform gene-based screening of ITGB2 to determine whether multiple, rare, deleterious variants contribute to HM risk.

The TFHM study was approved by the Health and Medical Human Research Ethics Committee (Tasmanian network; reference number H8551), and written informed consent was obtained from all participating individuals. The Tasmanian population samples used in this study were approved by the Health and Medical Human Research Ethics Committee (Tasmanian network) reference number H9999 and H7740.