ERBB2 gene variants and the risk of myeloproliferative neoplasms

The genetic risk factors for myeloproliferative neoplasms (MPN) remain relatively unknown. Familial clustering of MPN cases is a known phenomenon and as seen in ~10% of MPN cases. Patients who have first-degree relatives with MPN have a 5−7-fold increased risk of developing MPN themselves. Affected families often show an autosomal dominant inheritance pattern with incomplete penetrance. There have been advances in identifying germline predisposition variants that are associated with MPN; however, there is a lack of knowledge on the variants with direct influence on MPN.

In a study by Evan Braunstein and colleagues, a rare germline variant was identified in the ERBB2 gene that aligns with the presence of MPN and solid tumors, in a family with autosomal dominant familial cancer syndrome.¹ This mutant was then sequenced in multiple groups of diseased and control patients and analyzed for the association with MPN. The key points from this study are summarized here.

Study design and results

A family with highly penetrant autosomal dominant family cancer syndrome was recruited for the study and whole exome sequencing was carried out on three members of the family to identify the inherited variant.

At 38 years old, the proband was diagnosed with polycythemia vera (PV) that had a JAK2p.V617F variant with an allele frequency of 47%. This patient went on to be diagnosed with melanoma 1 year later, and then 5 years later they also received a diagnosis of renal carcinoma.

Multiple family members were also diagnosed with various malignancies including:

• The mother of the proband who developed ovarian cancer and melanoma.
• Her maternal aunt had essential thrombocytosis and melanoma.
• Her maternal uncle was diagnosed with bladder cancer and melanoma.
• Her brother had melanoma.

Variant analysis oh three family members identified eight single nucleotide variants (SNVs). When two additional family members were genotyped for these eight SNVs, three genes were found that co-segregated with the cancer phenotype in the proband’s family: ERBB2, MCM4, and DOCK5. In this family, the mutation was recorded to be ERBB2 c.3182T > C; p.L1061P.

ERBB2 has not been previously associated with blood cancer but has been recorded in connection to lung cancer, and a potential risk factor for breast cancer and melanoma. ERBB3, on the other hand, has been identified to co-segregate with disease in a family with acute myeloid leukemia (AML)/ myelodysplastic syndrome (MDS), indicating that this gene family may predispose to blood cancers.

An additional 55 cases underwent targeted sequencing to investigate the hypothesis that ERBB2 variants predispose to familial MPN. A predisposition to MPN is defined as:

• A diagnosis of MPN in index case
• And MPN or other myeloid malignancies (e.g., MDS or AML) in a first- or second-degree relative.

In this cohort an extra six cases with a germline variant of ERBB2 was found.

Another two families were genotyped and the probands were found to have the ERBB2 p.W452C variant which is fairly common in African populations with a mean allele frequency (MAF) = 0.049, but predicted to be a damaging variant that acts via PolyPhen-2. This variant only co-segregates with the MPN phenotype in one family and therefore is less like to be involved in a predisposition to MPN.

In the familial MPN cohort, 12.5% were found to harbor rare ERBB2 germline mutations, although none were identical to the index family. One hypothesis is that an inherited predisposition to MPN may be linked to an increase in somatic mutability which leads to an increased incidence of cancer.

To investigate this further, targeted sequencing of ERBB2 was performed in 1,604 patient who were evaluated for hematologic malignancies and the characteristics of these patients are shown in Table 1. In addition to the cancer cases, 256 control cases were included that had no signs of malignancy. Of this cohort of patients evaluated for blood cancer, 5% had a rare germline variant compared to 2.7% in the control cohort.

Two MAF thresholds were examined <0.01 and 0.005. At the more restrictive threshold 1.9% of the malignant cohort compared with 1.2% of the controls having the variant, however the difference was not significant (p = 0.11 and p = 0.41 for MAF < 0.01 and 0.005, respectively).

Table 1. Analysis of the germline ERBB2 variants and cancer cases and controls*

AA, aplastic anemia; AML, acute myeloid leukemia; CMML, chronic myelomonocytic leukemia; MAF, minor allele frequency; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm; vs, versus. *Adapted from Braunstein et al.1
Diagnosis	Cases (n)	ERBB2 MAF < 0.01, %	ERBB2 MAF < 0.005
MPN	236	8.9	4.7
CMML	92	6.5	2.2
MDS/ AML	771	3.9	1.2
AA	61	4.9	3.3
CML	27	3.7	0
Lymphoid	161	3.7	0.06
Control (non-blood cancer)	256	2.7	1.2
All blood cancer	1,348	5.0	1.9
Non-MPN blood cancer	1,112	4.1	1.3
Validation control cohort	1,587	5.2	3.0
Comparison	Odds ratio (95% CI), p value
MPN vs control	—	3.5 (1.4−8.3), p = 0.0053	4.1 (1.1−13.5), p = 0.0313
MPN vs MDS/AML	—	2.4 (1.4−4.0), p = 0.0028	4.1 (1.7−10.1), p = 0.0018
MPN vs non-MPN blood cancer	—	2.3 (1.3−3.9), p = 0.0028	3.8 (1.7−8.6), p = 0.0010
MPN vs validation control	—	1.7 (1.0−2.8), p = 0.0400	1.6 (0.8−3.1), p = 0.1698

The association of MPN variants with diagnosis was investigated next. Patients with MPN showed a significantly increase frequency of ERBB2 variants compared with the controls, both at MAF < 0.01 (8.9% vs 2.7%; p = 0.0053) and MAF < 0.005 (4.7% vs 1.2%; p = 0.031). Germline variants were more frequent in patients with MPN than patients with either non-MPN blood cancer or AML/MDS (Table 1). No association with MPN disease subtype or driver mutation was observed.

A validation cohort of 1,587 individuals was included in this study. This cohort produced similar results as patients with MPN harbored ERBB2 variants more frequently (8.9% vs 5.2%, p = 0.040) with an MAF <0.01. At the stricter threshold of MAF < 0.005, a non-significant increase in frequency of the ERBB2 variant was seen when comparing patients with MPN and the control group (4.7% vs 3.0%, p = 0.169).

In all groups the most frequently encountered variant was ERBB2 c.1960A > G; p.I654V and this occurred in one out of seven (14%) in the family cohort. In the other groups, the frequency of this variant was:

• MPN cohort, 2.05%
• Non-MPN cohort, 1.53%
• Control cohort, 1.07%

The p1654V was found twice as frequently as expected from its known population frequency with an observed/expected (o/e) ratio of 2.13 compared with o/e of 1.59 for the non-MPN cancer group and 1.11 in the control group. This variant has been previously associated with an increased risk of breast cancer and melanoma, which supports the history of the index family.

Other variants that were identified more than once are shown in Table 2. The most frequent in the MPN cohort was the p.E930D variant which had an o/e ratio of 6.67 compared with 1.75 for the non-MPN group and 0.613 in the control group. There was also one outlier in the non-MPN group, a rare I961T variant that was present in two patients which gave an o/e of 113.

Table 2. ERBB2 variants identified in cancer cases*

E, expected; MPN, myeloproliferative neoplasm; NA, not available; O, observed. *Adapted from Braunstein et al.1 †Based on global mean allele frequency.
Variant	MPN cohort (n = 292)			Non-MPN cancer cohort (n = 1,112)			Validation control cohort (n = 1,587)
	O (%)	E†	O:E	O (%)	E†	O:E	O (%)	E†	O:E
I654V	2.05	2.81	2.13	1.53	10.72	1.59	1.07	15.29	1.11
A386D	0.68	2.12	0.94	0.72	8.09	0.99	0.82	11.55	1.13
E930D	0.68	0.30	6.67	0.18	1.14	1.75	0.06	1.63	0.61
P489L	0.34	0.42	2.36	0.27	1.61	1.86	0.13	2.30	0.87
R1161Q	—	0.44	NA	0.27	1.69	1.78	0.13	2.41	0.83
R143Q	—	0.28	NA	0.18	1.10	1.89	0	1.51	NA
I961T	—	0.004	NA	0.18	0.02	113	0	0.03	NA

Conclusion

This study identified a rare ERBB2 mutation within the index family that co-segregated with the MPN phenotype. This variant was demonstrated to be more frequently found in patients with MPN than patients with non-MPN cancer and the control group. The ERBB2 p.I654V variant was associated with ~2-fold increased risk of MPN vs the controls. Therefore, this variant may be associated with a predisposition to MPN and other cancers.