Fumarate Metabolic Signature for the Detection of Reed Syndrome in Humans

In the past two decades, loss-of-function mutations in genes that encode components of the citric acid cycle enzymes succinate dehydrogenase (SDH) and fumarate hydratase (FH) have been demonstrated to predispose to a range of benign and malignant tumors (2–6). The tumor risk for patients with SDHX mutations (SDHB, SDHD, SDHC, SDHA, and SDHAF2) varies according to the specific gene involved but the most frequent tumors overall are phaeochromocytomas, paragangliomas, head and neck paragangliomas, renal cell carcinomas (RCC), gastrointestinal stromal tumors, and rarely, pituitary tumors (7). The clinical phenotype associated with FH mutations is hereditary leiomyomas and renal cell carcinoma syndrome (HLRCC) or Reed syndrome and comprises cutaneous and uterine leiomyomas, type 2 papillary RCCs (5, 8), and occasionally phaeochromocytomas/paragangliomas (6).

Tumors from patients with inherited FH and SDHX mutations demonstrate biallelic inactivation of the relevant gene and this results in loss of SDH or FH activity, which in turn causes pathologic accumulation of the metabolites succinate and fumarate, respectively (1, 9). Accumulation of succinate and fumarate causes competitive inhibition of 2-oxoglutarate–dependent enzymes (e.g., prolyl hydroxylase and DNA and histone demethylase enzymes) that drive tumorigenesis through epigenetic and gene expression alterations (9–11).

Targeted molecular imaging has served as a key adjunct to morphologic cross-sectional imaging studies, for the diagnosis and management of cancer for decades. PET with the glucose analogue ¹⁸F-fluorodeoxyglucose in conjunction with CT (¹⁸F-FDG PET/CT) serves as a paradigm for metabolic imaging in clinical oncology by measuring glucose uptake and phosphorylation in tumor cells. Despite the great sensitivity and wide clinical applications for PET imaging, the method does not distinguish individual metabolites or their cellular compartmentalization and provides no direct information on glycolytic flux or mitochondrial oxidative metabolism. In contrast, although proton magnetic resonance spectroscopy (¹H-MRS) is many orders of magnitude less sensitive than PET, it can noninvasively distinguish endogenous metabolites in vivo without the use of ionizing radiation. This ability to characterize the metabolic phenotype or tumor signature and detect its genotype is particularly relevant for metabolically driven tumors, which often demonstrate high endogenous metabolite concentrations (12).

The ability to measure fumarate in vivo as a functional biomarker has a number of important potential clinical applications including the early identification of FH-deficient tumors, which can enable tailored patient surveillance and facilitate timely cascade family screening. In vivo detection of fumarate could also noninvasively verify the pathogenicity of genetic variants in the era of next-generation sequencing.

This study was performed in accordance with the Declaration of Helsinki. All participants gave written informed consent and the study was approved by South Birmingham Research Ethics Committee (reference number: 5175).

DNA was extracted from peripheral blood samples according to standard protocols. Next-generation sequencing of a clinical gene panel was performed for cases 1 and 2 at Cambridge University Hospital NHS Foundation Trust using the TrusightONE Sequencing Panels (Illumina Inc.) and included: SDHA, SDHB, SDHC, SDHD, SDHAF2, MAX, TMEM127, VHL, RET, and FH. An average coverage depth of >20-fold was achieved for 98% of the regions sequenced. Single-gene sequencing of all coding exons of the FH gene was undertaken for case 3, in West Midlands Regional Genetics Laboratory.

¹H-MRS studies were performed on a 3T MRI System (MR750, GE Healthcare), with body coil transmission and 32-channel reception coils. T₁- and T₂-weighted images were acquired and a single voxel was prescribed within the tumor in each case. Automated adjustment of transmitter frequency, power, and magnetic field homogeneity was performed on all voxels prior to acquisition. Spectra at an echo time of 144 ms were acquired from the tumor with and without chemical shift selective water suppression pulses. Acquisition parameters were: case 1: TR 1.5 seconds, 256 averages, 6:48 acquisition, voxel size 337 mL; case 2: TR 2 seconds, 256 averages, 9:04, voxel size 39 mL, case 3: variable TR (respiration triggered), 128 averages, 90.8 mL.

The full width at half maximum height of the water peak in Hz was measured as an additional data quality metric. The chemical shift of the peak assigned as choline was 3.22 ppm and for fumarate was 6.54 ppm. The SAGE (GE Healthcare) spectroscopy analysis program was used to reconstruct, analyze, and display spectra. The concentrations of choline and fumarate were calculated on the basis of an assumed water concentration of 35 mmol/L and the following relaxation constants for pelvic tumor at 3 T (13): T1 water 1.6 seconds, T1 choline 1.1 seconds, T2 water 109 ms, and T2 choline 220 ms (Table 1). Because relaxation measurements have never been performed in vivo for fumarate, they were assumed to be the same as for choline. For the purpose of this pilot study, MRS was regarded as a technical failure if choline was not detected because it was assumed that choline should be detectable in a metabolically active tumor.

A rabbit polyclonal anti-2SC antibody (Cambridge Research Biochemicals LTD; dilution 1:400) was used to detect 2-succinyl cysteine (2SC), a product of succination secondary to excess fumarate accumulation. A second rabbit polyclonal antibody directed against the FH protein (Abcam, ab95947; dilution 1:4,000) was also employed. 2SC and FH IHC was performed on 4-μm sections of formalin-fixed, paraffin-embedded tissue, after appropriate selection of tissue blocks by an experienced pathologist.

Paraffin-embedded tissue from the retroperitoneal mass excised from case 2 was analyzed. A xylene/ethanol method was used to remove the paraffin wax from the tissue sample. Once the wax was removed, the remaining tissue was homogenized in water to a concentration of 33.3 mg/mL. Homogenate (10 μL) was taken for extraction by protein precipitation. The sample was then analyzed using LC/MS. The calibration range for fumarate was 18.8–6,006 nmol/g.

A female presented at the age of 25 years with a skin lesion on her back, subsequently diagnosed as a leiomyoma following a skin biopsy. Her family history revealed that multiple family members including father, paternal grandfather, and three siblings had similar skin lesions. At the age of 38 years, the patient re-presented to the dermatology clinic with two new painful skin lesions, both of which were confirmed histologically as leiomyomas. During this episode, the patient was also investigated for menorrhagia, and diagnosed with renal cysts and a 10-cm uterine fibroid (Fig. 1A). Analysis of the uterine fibroid using ¹H-MRS demonstrated an elevated fumarate peak at 6.54 ppm (Fig. 1B; Table 1).

A 45-year-old female, presented to the dermatology clinic with multiple cutaneous leiomyomas. Past medical history was significant for a previous hysterectomy for uterine fibroids and a family history of uterine fibroids. CT imaging of the abdomen and pelvis was performed to further investigate abdominal pain and incidentally demonstrated a large 7.5 × 5.7 cm right-sided heterogeneous retroperitoneal mass. A review of historic imaging studies, determined that this mass was present in 2008, when it measured 4.1 × 3.3 cm. At that time a biopsy of the lesion suggested a diagnosis of reactive lymphadenopathy. As the retroperitoneal mass had almost doubled in size over a 10-year period, the differential diagnosis included a sarcoma, paraganglioma, or hematologic malignancy. Plasma metanephrines were normal, thereby making a secretory paraganglioma unlikely [normetanephrine 440 pmol/l, normal range (NR) < 1,000 pmol/l; metanephrine 323 pmol/l, NR < 900 pmol/l; and 3-methoxytyramine levels < 75 pmol/l, NR < 180 pmol/l]. An ¹⁸F-FDG PET/CT demonstrated heterogeneous FDG uptake in the mass (maximum standardized uptake value, SUV_max 4.7; Fig. 2C) with no evidence of metastatic or distant disease. ¹H-MRS was performed on the retroperitoneal mass and despite the excellent spectral acquisition, there was no evidence of a fumarate peak (Table 1; Fig. 2B and C). Following a multi-disciplinary review, surgical excision of the retroperitoneal mass was recommended.

A 43-year-old male patient first presented to a dermatology clinic with painful widespread biopsy-proven cutaneous leiomyomas, requiring laser treatment, and nifedipine for pain control. This man was subsequently diagnosed with a left-sided renal tumor (Fig. 3A) on surveillance imaging and a biopsy confirmed a low-grade renal cell carcinoma. Germline genetic testing identified an FH variant of uncertain clinical significance (c.1370C>A, p.Thr457Lys). Analysis of parental samples confirmed maternal inheritance. The patient had no family history of renal cell carcinoma or uterine fibroids, but reported that a maternal uncle had similar skin lesions; as this relative was not living in the United Kingdom, clinical and molecular confirmation was not possible. Insufficient tissue was available to perform IHC for FH or 2SC on the renal biopsy or skin samples. ¹H-MRS was performed on the left-sided renal mass to gain further functional evidence to support the pathogenicity of the identified FH variant given the suspicious phenotype. Unfortunately, the quality of the spectra was affected by the presence of adjacent renal calcification (Fig. 3B) and the study was classified as a technical failure as neither choline nor fumarate was detected in the acquired tumor spectra (Fig. 3C). The renal tumor is currently under close surveillance and has not yet been excised.

Skin incision biopsy. An ill-defined collection of elongated, eosinophilic, spindled cells, arranged in intersecting fascicles in the dermis was observed. Features were in keeping with a pilar leiomyoma (Fig. 2G). IHC demonstrated loss of FH protein expression (Fig. 2H) and positive 2SC expression (Fig. 2I).

Retroperitoneal mass biopsy. Well circumscribed lymphoid tissue with focal peripheral sinus but no paracortical sinuses was identified (Fig. 2D). The morphology and IHC profile of this lesion was diagnostic of hyaline vascular Castleman disease. IHC demonstrated preserved FH protein expression (Fig. 2E) and absent 2SC expression (Fig. 2F).

Skin incision biopsy. Two lesions were sampled and both show features of benign pilar leiomyomas.

Left-sided renal tumor biopsy. The core renal biopsy demonstrated nests and tubules of eosinophilic epitheliod cells with mild variation in nuclear size and occasional nucleoli. No necrosis was evident. The IHC and morphology was nonspecific, so a diagnosis of low-grade renal cell carcinoma was made but further subclassification was not possible on the biopsy specimen.

A pathogenic missense variant in exon 8 of the FH gene (c.1189G>A, p.Gly397Arg) was identified in patient 1, a likely pathogenic missense variant in exon 7 of the FH gene (c.956A>G, p.Asp319Gly) was present in case 2, and a variant of uncertain clinical significance in exon 9 of the FH gene (c.1370C>A, p.Thr457Lys) was identified in case 3.

No fumarate was detected in the retroperitoneal mass from case 2 ex vivo, correlating with the in vivo findings and IHC results, suggesting an intact FH enzyme and therefore ruling out a causative role of the germline FH variant in the pathogenesis of the retroperitoneal tumor in case 2.

Reed syndrome, the re-occurrence of uterine and skin leiomyomas, was first described in 1973 (8). However, three decades later the clinical observation that renal cell carcinoma cosegregated with cases of uterine and cutaneous leiomyomas prompted a renaming of the syndrome to HLRCC (14). The morbidity and mortality associated with this syndrome is accounted for by the predisposition to an aggressive form of renal carcinoma: HLRCC-associated renal tumors predominately exhibit a type 2 papillary morphology, typically present during the fourth decade, and possess metastatic potential, which is independent of tumor size (14–16). A clinical diagnosis of HLRCC is based on the presence of cutaneous smooth muscle tumors, but it is notable that almost 30%–50% of patients with FH gene mutations may not show evidence of cutaneous leiomyomas (15, 17). Therefore, HLRCC may be under-reported because of the bias for patients presenting with cutaneous leiomyomas, highlighting the need for confirmatory genetic testing or sensitive biomarkers to make a definitive diagnosis of HLRCC (16).

In the cases we describe here, cutaneous and uterine leiomyomas raised the possibility of a diagnosis of HLRCC. Early diagnosis of HLRCC in patients presenting with common benign tumors such as uterine and skin leiomyomas enables targeted screening for RCC in probands and prompts family genetic screening and surveillance of other mutation carriers in the family, therefore helping to reduce the morbidity associated with this syndrome. However, not infrequently, rare FH missense substitutions in patients with possible HLRCC may be difficult to categorize as pathogenic or benign. Although IHC for protein succinylation is a helpful adjunct for assessing variant pathogenicity (18), this can only be performed after surgery or biopsy. Given the prevalence of uterine leiomyomas and the rarity of HLRCC syndrome, the availability of a noninvasive diagnostic tool to assess the likelihood that a lesion is HLRCC related is an important advance and could be considered in clinical practice for those women presenting with uterine leiomyomas at a young age and or a family or personal history of cutaneous leiomyoma, RCC, or uterine leiomyomas.

Here, we have demonstrated a proof of principle that ¹H-MRS can provide a noninvasive diagnosis of HLRCC through detection of pathologic fumarate accumulation in vivo. Furthermore, ¹H-MRS can also assist in noninvasively determining whether an intraabdominal mass in a patient with HLRCC is likely to be a component of the syndrome or coincidental: this is an important consideration given the high incidence of incidental findings on surveillance imaging. In case 2, this was particularly helpful in demonstrating that the retroperitoneal mass was not a nonfunctional paraganglioma. Given that ¹H-MRS can also be used to detect succinate, as well as fumarate within a paraganglioma, both SDH and FH deficiency could be assessed simultaneously.

Type 2 papillary RCC associated with germline FH mutations, is generally highly aggressive and has a poor prognosis (17). Advances in our understanding of the molecular mechanisms of tumorigenesis in HLRCC, such as activation of hypoxic gene response pathways, DNA hypermethylation, defective DNA repair, and increased sensitivity to PARP inhibitors suggests a range of personalized experimental therapeutics that could be applied in patients with metastatic disease (19). Functional imaging of FH-deficient RCC by ¹H-MRS could be used as a sensitive tool to determine tumor response to therapy as we have previously described in SDH-deficient tumors, in which reductions in tumor succinate levels occurred before changes in tumor morphology (12).

However, important limitations of in vivo metabolomic analysis using ¹H-MRS have been demonstrated in this study and earlier studies (12); for example, spectral quality was poor in case 3 due to the relatively small tumor size and close proximity of renal calcification. We also speculate that spectroscopic detection of metabolites may have been affected by the earlier tumor biopsy, suggesting that the timing of the ¹H-MRS in relation to a diagnostic biopsy is an important consideration due to the risk of hemorrhage, necrosis, or subsequent calcification affecting spectral quality.

In an earlier study investigating succinate accumulation in vivo, our group has reported a technical failure rate of 26%, which is similar to the failure rate reported in earlier reports using ¹H-MRS (12). Therefore, further studies are required to determine the most appropriate selection criteria when considering in vivo fumarate analysis using ¹H-MRS in a clinical setting. On the basis of the findings of this pilot study, we recommend that ¹H-MRS is performed on tumors greater than 1.5 cm in size, or greater than 2.5 cm if the tumor location is deep because coil sensitivity decreases with distance, and, when possible, ¹H-MRS should be performed prior to tumor biopsy or alternatively several weeks later if a surgical excision is not being considered. In addition to investigating appropriate patient selection criteria, future validation studies should also include repeatability analysis as either repeated analysis of the same tumor or comparison between primary and metastatic lesions to confirm how robust this method is for the in vivo detection of fumarate.

In conclusion, this study demonstrates that in vivo detection of fumarate using ¹H-MRS could be employed as a functional biomarker of metabolic derangement in patients suspected of having HLRCC. In the future it could be used as a treatment response biomarker for targeted therapies.

R.T. Casey has received speakers bureau honoraria from IPSEN. B.G. Challis is a director physician (paid consultant) at AstraZeneca. F.A. Gallagher reports receiving a commercial research grant from GSK and GE Healthcare. E.R. Maher has received speakers bureau honoraria from Alexion Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

The views expressed are those of the authors and not necessarily those of the NHS or Department of Health.

Conception and design: R.T. Casey, M.A. McLean, F.A. Gallagher, E.R. Maher

Development of methodology: R.T. Casey, M.A. McLean, A.Y. Warren, L. Mendil, F.A. Gallagher

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.T. Casey, M.A. McLean, B.G. Challis, T.P. McVeigh, A.Y. Warren, L. Mendil, R. Houghton, V. Kosmoliaptsis, R.N. Sandford

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.T. Casey, M.A. McLean, B.G. Challis, R. Houghton, F.A. Gallagher

Writing, review, and/or revision of the manuscript: R.T. Casey, M.A. McLean, B.G. Challis, T.P. McVeigh, A.Y. Warren, V. Kosmoliaptsis, F.A. Gallagher, E.R. Maher

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Kosmoliaptsis

Study supervision: R.T. Casey, S. De Sanctis, F.A. Gallagher, E.R. Maher

The authors would like to thank Stephen Provencher for providing the simulated basis set used in spectral fitting, the radiographers and staff of the MRIS Unit at Addenbrooke's Hospital, the staff of the Tissue Bank at Addenbrooke's Hospital for assistance, and the patients who participated in this study. We thank the following funding organisations: GIST Support UK (to R.T. Casey), NIHR (to R.T. Casey), Cambridge Experimental Cancer Medicine Centre, Addenbrooke's Charitable Trust, National Institute for Health Research (NIHR) Cambridge Biomedical Research Centre, Cancer Research UK CRUK (to F.A. Gallagher), CRUK Cambridge Centre (to M.A. McLean and F.A. Gallagher), the University of Cambridge, and Hutchison Whampoa Ltd (to M.A. McLean), NIHR Senior Investigator Award (to E.R. Maher), European Research Council Advanced Researcher Award (to E.R. Maher), and CRUK and Engineering and Physical Sciences Research Council Imaging Centre in Cambridge and Manchester (to F.A. Gallagher). The University of Cambridge has received salary support in respect of E.R. Maher from the NHS in the East of England through the Clinical Academic Reserve.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.