Fumarate hydratase FH c.1431_1433dupAAA (p.Lys477dup) variant is not associated with cancer including renal cell carcinoma

Fumarate hydratase (FH) is an enzymatic component of the Krebs cycle catalyzing fumarate to malate. FH deficiency is a severe autosomal recessive disorder resulting from homozygous or compound heterozygous mutations in FH (Ewbank, Kerrigan, & Aleck, 1993). Hereditary leiomyomatosis and renal cell cancer (HLRCC) is an autosomal dominant disorder caused by FH heterozygous loss of function mutations. HLRCC may manifest with multiple cutaneous leiomyomas, uterine leiomyomatosis, and/or renal cell cancer (Badeloe et al., 2006; Tomlinson et al., 2002; Toro et al., 2003). Historically, many terms have been cited in the literature for this syndrome including multiple cutaneous leiomyomatosis, multiple cutaneous and uterine leiomyomatosis (MCUL), and/or leiomyomatosis cutis et uteri. The association of cutaneous and uterine leiomyomas has been referred to as Reed's syndrome (Reed, Walker, & Horowitz, 1973). These are all descriptions of the same disorder, HLRCC, although there are significant intra and interfamilial variation in terms of diseases severity (Wei et al., 2005).

To date, the same FH variant has been classified uniformly for both disorders regardless of its status as a homozygous, heterozygous, or compound heterozygous variant. Parents of FH deficiency children have variable phenotypes reported, and a strong predisposition to MCUL varies (Alam et al., 2003). The location of FH mutations associated with FH deficiency differs significantly from those associated with leiomyomatosis (Tomlinson et al., 2002; Toro et al., 2003). In terms of variant types, while missense variants are primarily associated with FH deficiency, truncating variants account for more than half the mutations associated with HLRCC. Classification of inframe indel variants is less straightforward as functional studies are needed to assess whether these changes sufficiently alter protein activity to predispose to cancer.

The FH c.1431_1433dupAAA variant in NM_000143.3 (GRCh37:Chr1: 241661228–241661230) results in the duplication of a single lysine codon at amino acid position 477 (p.Lys477dup). This variant has been described in combination with another FH mutation in multiple families with autosomal recessive FH deficiency and is the most common variant found in this disorder (Coughlin et al., 1998). The FH c.1431_1433dupAAA variant has been reported as a heterozygous variant in patients with cutaneous leiomyomas (Ezgu, Krejci, & Wilcox, 2013), ovarian mucinous cystadenoma, uterine myomas (Ylisaukko-oja et al., 2006), and leiomyomas (Martinek et al., 2015). A recent study identified this variant in three individuals with multiple primary tumors, so the presumption of pathogenicity is stronger. However, none of whom had typical hereditary leiomyoma or renal cell carcinoma tumors (Whitworth et al., 2018). Although it is a well-established pathogenic variant for autosomal recessive FH deficiency, its association with renal cell carcinoma is less well defined. The goal of this study was to determine the association of this variant with cancer in the heterozygous state.

The FH c.1431_1433dupAAA variant is located in the last exon (exon 10) of the FH gene and occurs in the C-terminus of the fumarase hydratase protein. In this series, 24/7571 (0.32%) patients were identified carrying the FH c.1431_1433dupAAA variant including prostate (n = 6), pancreatic (n = 2), colorectal (n = 3), bladder (n = 2), ovarian (n = 2), bile duct cholangiocarcinoma (n = 2), endometrial (n = 1), breast (n = 2), lung (n = 1), nonseminomatous germ cell tumor (n = 1), astrocytoma (n = 1), and melanoma (n = 1) cancers (Table 1). None of the 24 carriers have a personal or family history of renal cell carcinoma. Carriers (20/24) of this variant in our cohort were diagnosed with cancer at age ≥50. Of 24 patients, 11 were females with two having a prior history of uterine leiomyomata and diagnosed with urothelial cancer and endometrial cancer at ages 50 and 64, respectively. Of 24, 4 carriers had pathogenic variants in other genes and three matched their phenotypes.

We examined whether the tumors from these individuals showed loss of heterozygosity (LOH) or harbored a somatic second mutation in FH. LOH analysis was performed using the FACETS algorithm. LOH was not observed in 21 out of 23 tumors from these patients (one patient's tumor was not available for sequencing). LOH was observed in two tumor specimens, one from a patient who was diagnosed with recurrent melanoma at age 50, and one from a patient with prostate cancer who had an AJ founder mutation, BRCA2 c.5946delT. None of the 23 tumors had a second hit in FH (Table 1).

We then studied the frequency of the FH c.1431_1433dupAAA variant in patients with renal cancer tested within the same period. A total of 372 patients with renal cancer (Table S1 for subtypes) were included in this study and none of them carried the FH c.1431_1433dupAAA variant, while 11 patients carried other pathogenic FH variants that are associated with HLRCC (Table 2).

Evaluating the gnomAD database across populations for the frequency of the FH c.1431_1433dupAAA showed that this variant is present in 0.0015 (188/127, 286 alleles with 1 homozygote) in European (non-Finnish) population (http://gnomad.broadinstitute.org/ variant/1-241661227-A-ATTT). The allele frequency in our cohort is 0.0015 (24/(7571 × 2)), which is very close to that in the European (non-Finnish) population. Our data indicate that there was no association of this variant with cancer in our cohort of patients (n = 7,571; p = .74; odds ratio [OR] = 1.07; Table S2).

Because 6 out of 24 patients (predominantly AJ) were affected with prostate cancer, we investigated whether this variant is associated with increased risk of prostate cancer. We generated genotypes for a hospital-based ascertainment of AJ prostate cancer (Sullivan et al., 2015; Vijai et al., 2011) patients (5/886) versus noncancer diagnosed AJ controls (Carmi et al., 2014; 3/557) and found no evidence of association for this variant with prostate cancer in AJ population (p = .87; OR = 1.12; Table S3).

To determine the FH activity in such carriers, we measured the fumarase activity from skin fibroblasts harvested from one carrier, who had a history of uterine leiomyomas. The enzymatic activity in the patient (22.3 nmol·min⁻¹·mg⁻¹ protein) is significantly reduced (24%) relative to controls (91.5 ± 22.8 nmol·min⁻¹·mg⁻¹ protein, n = 43; Figure S1). However, it is significantly higher than the residual activity in FH deficiency cases which lies between less than 1% to 10% (Coughlin et al., 1998; Deschauer et al., 2006; Rustin et al., 1997; Tregoning et al., 2013).

This study highlights the need to extensively sequence ethnically diverse populations to accurately interpret genetic variations and their clinical significance. FH c.1431_1433dupAAA is a relatively common variant in the general population with minor allele frequencies of 0.5% and 0.15% in AJ and European (non-Finnish) populations in gnomAD cohort, respectively, including two homozygous individuals. All but 1 of 24 carriers in our cohort self-identified as White or AJ. The allele frequency of the FH c.1431_1433dupAAA variant in our cohort is consistent with that of European (non-Finnish) population in the gnomAD database, suggesting this variant was detected in our cohort simply because it is relatively common in the general population and that it may not be related to HLRCC.

None of the 372 patients with RCC in our cohort were carriers of this variant while 11 had other FH pathogenic variants. In addition, none of the 24 FH c.1431_1433dupAAA carriers have a personal or family history of renal cell cancer. Two patients reported a history of uterine fibroids but not cutaneous leiomyomas. Uterine fibroids are the most common benign tumors in women worldwide (Merrill, 2008; Zimmermann, Bernuit, Gerlinger, Schaefers, & Geppert, 2012), and in isolation are not indicative of HLRCC by themselves (Menko et al., 2014; Pithukpakorn & Toro 1993–2018–2018). None of the 24 FH c.1431_1433dupAAA carriers in our cohort meet either the major or minor criteria for HLRCC diagnosis, supporting our observations for a possible nonpathogenic role for this variant in the heterozygous state. Of 24, 3 carriers had pathogenic variants in other genes that matched their phenotypes. As noted, 2 of the 24 heterozygous germline carriers in our cohort showed LOH of the wild-type allele in the tumor and the remaining tumors do not have another somatic mutation in the tumor. Because these two patients were affected with cutaneous melanoma and prostate cancer, which are not associated with HLRCC, we do not speculate that LOH of FH contributes to the tumorigenesis in these two cases. Of note, the FH c.1431_1433dupAAA was detected in one case of HLRCC, but the patient also had a second germline variant in FH, c.683T>A (p.I228N), which is predicted to affect the protein function using in silico prediction tools, hence the role of the c.1431_1433dupAAA, in this case, remains unclear (Chen et al., 2014).

While FH c.1431_1433dupAAA has been reported as one of the most common mutations associated with autosomal recessive FH deficiency, none of the published FH-deficient cases carry homozygous c.1431_1433dupAAA (Coughlin et al., 1998; Deschauer et al., 2006; Rustin et al., 1997; Tregoning et al., 2013). The presence of two homozygous individuals in the gnomAD database suggests this variant is only pathogenic in conjunction with a different loss of function variant.

This report emphasizes the value of a large and diverse dataset in clinical interpretation. One caveat is that most of the patient's family history is self-reported and there is a selection bias in our cohort as AJ are overrepresented and our patients with cancer are predominantly White. Nevertheless, the large cohort of patients in this study indicates the FH c.1431_1433dupAAA is not associated with cancer including renal cell cancer.

ACKNOWLEDGMENT

This study was funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748.