Fumarate Hydratase Mutations and Alterations in Leiomyoma with Bizarre Nuclei

Abstract

Leiomyoma with bizarre nuclei (LM-BN), is a variant of uterine smooth muscle tumor with atypical histological features. Although some LM-BN share several significant genetic alterations with leiomyosarcoma (LMS), including p16 and p53, the underlying tumorigenesis of LM-BN remains largely unknown. As we previous reported, LM-BN can be divided into two subtypes, type I and type II, based on different nuclear features. Type I LM-BN have similar histologic features as uterine smooth muscle tumors with fumarate hydratase (FH) alterations. In this study, we examined FH expression and FH mutations in 77 LM-BN (40 type I cases and 37 type II cases). FH expression was examined by immunohistochemistry (IHC) using S-(2-succino)-cysteine antibodies (2SC, a protein modification associated with FH inactivation and subsequent fumarate accumulation) and FH antibodies (fumarate hydratase gene products). Seventy-two LM-BN tumors underwent Sanger sequencing to detect FH mutations. We found that 51% (39/77) of LM-BN showed FH alterations detected by immunohistochemistry with both 2SC and FH. Mutational analysis showed that 21% (15/72) of LM-BN harbored FH gene mutations. Further analysis revealed that 85% (34/40) of those with FH alterations were type I LM-BN while 19% (7/37) were type II LM-BN. Our findings suggest that over half of histologically-diagnosed LM-BN may be related to FH alterations or FH mutations and the majority of these have the characteristic histologic features of type I LM-BN.

INTRODUCTION

Uterine smooth muscle tumors consist of a diverse group of tumors based on histologic features and clinical behavior. Currently the 2014 WHO classifies them as benign (including usual-type leiomyoma, leiomyoma with bizarre nuclei, cellular leiomyoma, mitotically-active leiomyoma, and other rare variants), intermediate grade (atypical smooth muscle tumors) and malignant (leiomyosarcoma, LMS). Leiomyoma with bizarre nuclei (LM-BN) was changed from atypical leiomyoma due to its overall benign clinical course ^1–4. Thus, we use the terminology LM-BN in this report based on the WHO recommendation. The biologic nature of LM-BN and its histologic relationship to benign (usual-type leiomyoma, ULM), intermediate (atypical uterine smooth tumors) and malignant (LMS) uterine smooth muscle tumors remain unclear. Recently, several independent studies have investigated the molecular relationship among these variants and the results have shown that many cell cycle regulatory proteins including p16, p21, p27 and p53 are dysregulated in both LM-BN and LMS ^{2, 4–6}. Given the completely different clinical outcomes of these two entities, further investigation into LM-BN may assist in better understanding this tumor.

Fumarate hydratase (FH) mutations are most notably seen in hereditary leiomyomatosis and renal cell carcinoma syndrome (HLRCC), also known as Reed syndrome, and they can also be seen with uterine leiomyomatosis⁷. Somatic FH mutations have occasionally been identified in some patients treated for uterine leiomyomas⁸ with a higher incidence seen in younger patients⁹. Further histologic analysis has revealed that some ULM with FH mutations have different nuclear and histologic features and can be referred to as smooth tumors with features of fumarate hydratase alteration (SMT-FH)^{9, 10}. Historically, many sporadic SMT-FH were diagnosed as LM-BN due to the significant nuclear atypia^{11, 12}. LM-BN also present with a wide range of histologic features^{1, 3, 4} and in our set of tumors originally diagnosed as LM-BN, nearly half showed nuclear features similar to SMT-FH ^{9, 10} and we therefore defined them as type I LM-BN¹³. This suggests that many uterine smooth muscle tumors with atypical histology diagnosed as LM-BN may be related to FH alterations/mutations.

In this study, we analyzed FH alterations/mutations in 77 cases of LM-BN. FH mutations were first examined by immunohistochemistry using S-(2-succino)-cysteine antibodies (2SC), a substrate of FH that accumulates following inactivation of FH, and then by FH antibody to identify the fumarate hydratase gene products. The FH gene was then sequenced for each tumor. Both IHC and molecular sequencing identified distinct patterns of FH alterations in the two LM-BN subtypes.

MATERIALS AND METHODS

Case Selection

This study was approved by the Northwestern University Institutional Review Board (IRB). A total of 77 cases of LM-BN were retrieved from the pathology databases at Northwestern Memorial Hospital (Northwestern University, Chicago, IL, USA) and Qilu Hospital (Shandong University, Shandong, China). This included 49 cases from Northwestern Memorial Hospital and 28 cases from Qilu Hospital. The slides from each case were reviewed by two pathologists to confirm the original diagnosis using the Bell criteria ¹⁴ and clinical information from each patient’s chart was also recorded. Detailed case information is summarized in Table 1.

Table 1.

General information for LM-BN cases and subtypes

Clinical parameters	Measurement	Total LM-BN	Type I	Type II	p value
No. Cases		77	40	37
Age (years)	Med (Range)	40 (21–65)	38 (21–57)	43 (28–65)	0.003
Hysterectomy	Percent (n)	54.5 (42/77)	42.5 (17/40)	67.6 (25/37)	0.027
No. Tumor slides	Med (Range)	7 (2–20)	7 (3–19)	6 (2–20)	0.063
Tumor size (cm)	Mean±sem	7.7±0.5	8.7±0.7	6.7±0.6	0.728

Histologic features of type I and II LM-BN

By reviewing all tumor slides, LM-BN were divided into two subtypes purely based on distinct nuclear features. Type I LM-BN were characterized by large round or oval nuclei, distinct smooth nuclear membranes, prominent nucleoli, and open chromatin (Figure 1). Type II LM-BN were characterized by elongated or spindled nuclei, irregular nuclear membranes, pinpoint or no nucleoli, and dark smudgy chromatin (Figure 1). For those cases with mixed type I and II nuclear features, we defined cases as either type I or II when over 70% of the tumor cells appeared to show nuclear features favoring one or another. The detailed description of type I and II characterization has been outlined in a previous study ¹³.

Figure 1.

Photomicrographs illustrating hematoxylin and eosin (H&E) stained slides in three randomly selected cases of type I (left) and type II (right) LM-BN. Type I LM-BN have large round to oval nuclei, prominent nucleoli and perinucleoli halos. They also have open chromatin and eosinophilic hyaline globules. Type II LM-BN have enlongated and spindled nuclei with dark, hyperchromatic, and smudgy chromatin. (Magnification 40X).

Immunohistochemical analysis

Antigen retrieval was performed by first microwaving in a citrate buffer at a pH 6.0 for 15 minutes then blocking for endogenous peroxidase. The specimen was then incubated with anti-2SC polyclonal antibody (1:5000) and washed in phosphate buffered saline (PBS). Myometrium was used as a negative control. The 2SC antibody ¹⁵ was provided by Dr. Norma Frizzell and used as previously described¹⁶. FH from Santa Cruz Biotechnology was used at a dilution of 1:200 (J-13 fumarate hydratase, Santa Cruz Biotechnology, Santa Cruz, CA, USA). The IHC stains were analyzed blindly by two pathologists and the percent staining and staining intensity (0=negative, 1+=weak, 2+=moderate, and 3+=strong) were recorded for each case. Strong and diffuse immunoreactivity for 2SC in the cytoplasm was considered positive. FH immunostaining was considered negative when there was no detectable signals in the tumor cells and FH staining was positive when the tumor cells showed strong cytoplasmic staining. Endothelial cells showed positive staining and were used as an internal control.

Semiquantitative scores for each tumor and matched myometrial controls were calculated using the staining intensity and percentage. We found that both 2SC and FH IHC had specific staining patterns and were easily identified by their positive or negative stain. 2SC positivity was defined as diffusely strong immunoreactivity for 2SC while weak or focal immunoreactivity for 2SC was considered negative. FH negativity was defined as complete absence of immunoreactivity for FH in tumor cells.

Molecular analysis

Mutational analysis for FH was performed on 72 of the LM-BN cases. Pure tumor tissue were sectioned at 10 μm and genomic DNA was extracted from formalin-fixed and paraffin-embedded (FFPE) tissue using the QIAamp DNA FFPE Tissue Kit (QIAGEN, 56404). For FH mutational analysis, 50ng DNA was amplified by PCR with HotStarTaq Master Mix (QIAGEN, 203446) at all ten FH exons. PCR products were then purified using the Gel Extraction and PCR Clean-Up Kit (Clontech) according to the user’s manual. DNA sequencing of the purified DNA products was performed using the ABI 3730 High-Throughput DNA Sequencer. The mutations and variations were analyzed using DNASTAR Lasergene 9 software. Detailed information regarding primers can be found in Supplementary Table 1.

Statistical analysis

SPSS Version 19.0 was used for statistical analysis. The mean differences between clinical parameters and histologic features were analyzed using a student T-test. Positivity rates for each gene were performed using Pearson’s chi-squared test and Fisher exact test. The association between 2SC and FH IHC expression was evaluated using the Pearson coefficient. P values less than 0.05 were considered statistically significant.

RESULTS

General case information

We reviewed and confirmed the diagnosis for 77 LM-BN cases by Bell’s criteria ¹⁴. The patient age ranged from 21 to 65 years with a median age of 40 years and the mean tumor size was 7.7 cm (Table 1). LM-BN were further divided into two categories, type I and type II, based on nuclear features ¹³. Type I LM-BN showed round-to-oval nuclei, distinct and smooth nuclear membranes, prominent nucleoli with perinucleolar halos, and open chromatin (Figure 1). Type II LM-BN showed large elongated or spindled nuclei, irregular nuclear membranes, pinpoint or absent nucleoli, and dark smudgy chromatin (Figure 1). In general, type I LM-BN patients were younger (mean age of 38 years) than that of type II (mean age of 43 years, p< 0.01) and the mean overall tumor size was larger in type I than in type II (8.7 cm vs. 6.7 cm, p< 0.05) (Table 1). Seventy-two of the 77 patients had more than one uterine nodule. The additional nodules were either additional LM-BN or usual-type leiomyomas (ULM). The general information is summarized in Table 1.

Immunohistochemical analysis of 2SC

S-(2-succino)-cysteine (2SC) is a downstream biomarker for biallelic FH inactivation and strong IHC reactivity has been shown to be a sensitive marker in detecting defective FH enzyme function in tumor cells (Figure 2). In the 77 LM-BN cases, diffuse positivity for 2SC was found in 53% (41/77) of cases while all 60 myometrial controls were completely negative (Figure 2B). Overall, patients with 2SC immunoreactivity were younger and had a larger tumor burden than the 2SC negative cases (p<0.05). Additionally, histologic features including prominent nucleoli, perinucleolar halos, hyaline globules, and staghorn vessels had high concordance with 2SC IHC positivity (Table 2).

Figure 2.

Side-by-side comparison of immunoreactivity for 2SC and FH by IHC in two randomly selected type I LM-BN (A) and control myometrium (B). Strong and diffuse immunoreactivity for 2SC can be appreciated in both the nucleus and cytoplasm. Loss of immunoreactivity for FH is evident in all tumor cells while positive staining for FH in endothelial cells is present as an internal positive control. H&E sections (40X) are presented on the left.

Table 2.

Correlation between 2SC status and histologic features

	Immunoreactivity for 2SC
	Positive	Negative	p value
No. cases	53.2% (41/77)	46.8% (36/77)
Age (years)	38	43	0.045
Size (cm)	8.6	6.7	0.024
Hyaline globules	96.8%	30.7%	<0.001
Prominent nucleoli	80.6%	17.9%	<0.001
Perinucleolar halo	83.3%	16.7%	<0.001
Staghorn vessels	91.7%	33.3%	<0.001

Next, immunoreactivity for 2SC was compared between type I and type II LM-BN. Of the 40 type I LM-BN cases, 85% (34/40) were 2SC positive (Figure 2), while 19% (7/37) of type II were positive for 2SC (p<0.01) (Table 3). These findings suggest that the majority of type I LM-BN were related to FH alterations and only a small fraction of type I LM-BN (7/40) were completely negative for 2SC by IHC (Figure 3). As shown in Figure 3, tumors with characteristic nuclear features of type I LM-BN that were completely negative for 2SC were re-reviewed. To confirm that there were no misdiagnoses of uterine PEComa, immunohistochemical stains for Melanin A and HMB45 were performed and were negative for all cases. Of the 37 type II LM-BN, 81% (30/37) were negative for 2SC (Figure 4) and only 19% were positive for 2SC by immunohistochemistry (Figure 5). While the nuclear features of type I LM-BN strongly correlated with 2SC immunopositivity, this alone was not always an accurate predictor of 2SC positivity (Table 3). The positive rates of 2SC in LM-BN between the two different institutions were also similar (55% vs 51%, respectively).

Table 3.

FH IHC expression and mutational analysis in LM-BN

	No. cases	2SC +ve % (n)	FH −ve % (n)	FH mutation % (n)	2SC +ve FH −ve*	2SC +ve mutation*
Total LM-BN	77	53.2 (41/77)	50.6 (39/77)	20.8 (15/72)	−0.939	0.428
Type I	40	85.0 (34/40)	82.5 (33/40)	26.3 (10/38)	−0.791	0.224
Type II	37	18.9 (7/37)	16.2 (6/37)	14.7 (5/34)	−1.000	0.808
2SC Positive	41		95.1 (39/41)	36.6% (15/41)
2SC Negative	36		0 (0/36)	0 (0/33)

^*: Pearson correlation

Figure 3.

Histologic and immunohistochemical analysis of 2SC in two randomly selected type I LM-BN. H&E slained slides (60X) reveal the large tumor cells with round to oval nuclei, prominent nucleoli, perinucleoli halos, open chromatin, and distinct nuclear membranes. Occasional eosinophilic hyaline globules are seen. Stains for Melanin A and HMB45 were performed (not shown) and were negative, excluding the differential diagnosis of PEComa. The tumor cells showed weak or negative immunostaining for 2SC.

Figure 4.

Side-by-side comparison of immunoreactivity for 2SC and FH in two randomly selected type II LM-BN (A) and control myometrium (B). Strong and diffuse immunoreacitvity for FH can be appreciated in the cytoplasm of both the tumor cells and the endothelial cells. Immunostaining for 2SC was negative in both cases. H&E sections (40X) are presented on the left.

Figure 5.

Photomicrographs illustrating examples of diffuse immunoreactivity for 2SC in two cases with type II LM-BN nuclear features.

Immunohistochemical analysis of FH

2SC immunoreactivity detects the byproduct of accumulated fumarate in tumors with defective FH. While it appears to be sensitive, its specificity for FH gene mutations in uterine smooth muscle tumors is still unclear^{8, 10, 17}. Thus, to test this concept further to see whether 2SC-positive tumors are indeed caused by FH genetic alterations, all cases were also stained with FH antibodies. FH IHC should be negative in tumors with loss of FH and is quite specific ¹⁷. As illustrated in Figure 2, complete absence of the immunostain in tumor cells for FH was identified and compared to myometrial tissue and internal endothelial cells. Among the 41 cases of 2SC-positive LM-BN, 95% (39/41) were negative for FH by IHC (Figure 2). A Pearson rate correlating the 2SC and FH immunohistochemical staining was −0.939 (Table 4). Only two cases were positive for both markers and FH mutational analysis revealed that neither of those cases had FH mutations (see below). All 36 2SC-negative cases were also FH-positive (Figure 4). Overall, IHC for 2SC and FH were complimentary to each other in almost all cases and can be reliably used to detect FH alterations in LM-BN (Figure 2 and 4, Table 3).

Table 4.

FH alterations and mutations in sporadic uterine leiomyomas, hereditary uterine leiomyoma and atypical leiomyoma

	2SC +ve % (n)	FH −ve % (n)	FH mutation % (n)	References
ULM	N/A	1.3 (31/2449)	1.3 (5/413)	Five studies 8, 9, 19, 20
HLRCC			85 (89/105)	Kiuru et al 18
SMT-FH	100% (9/9)	N/A	66.7 (2/3)	Reyes; Wheeler et al 10, 11
ALM	N/A	37.3 (68/182)	N/A	Miettinen et al 24
LM-BN	53.2 (41/77)	50.6 (39/77)	20.8 (15/72)	Current study

ULM: usual leiomyoma; SMT-FH: smooth muscle tumor with features of fumarate hydratase alterations; ALM: atypical leiomyoma; LM-BN: atypical leiomyoma/leiomyoma with bizarre nuclei

Molecular analysis

The immunohistochemical analysis of 2SC and FH appears to be both sensitive and specific in identifying FH alterations in LM-BN cells. To test whether the alterations of FH expression by IHC were due to FH gene mutations, FH gene mutation analysis was conducted in 72 LM-BN cases using Sanger sequencing, specifically looking at the ten exons of the FH gene. Among the 72 LM-BN cases, 21% (15/72) harbored FH mutations. In our current study, 36.6% (15/41) of 2SC-positive LM-BN harbored FH gene mutations, mainly comprised of missense mutations (Figure 5). FH mutations were found in 38.5% (15/39) cases that were negative for FH by IHC. The FH mutant tumors were exclusively identified in tumors with positive 2SC and negative FH by IHC (Table 3, FHFigure 6). Yet, there was still a large number of 2SC-positive and FH-negative cases by IHC with no detectable mutations by sequencing analysis.

Figure 6.

Sequencing analysis of FH mutations in uterine smooth muscle tumors. A) Sequencing plots with red arrows designating the mutation sites, B) Mutation distribution of fumarate hydratase (red dots) in each of the ten FH exons.

DISCUSSION

FH alterations are seen in HLRCC (Reed Syndrome) and FH germline mutations are present in 85% of these cases ¹⁸. In contrast, the incidence of FH mutations in sporadic leiomyoma is very low. Previous studies have shown only 1% of unselected leiomyomas will have FH-deficiency and most are somatically-acquired FH mutations (Table 4) ^{8, 9, 19, 20}. In uterine leiomyomas from patients younger than 40 years of age, FH mutations have been reported to be only slightly higher, accounting for 2.57% of these cases ⁹. Additionally, FH mutations in LMS have occasionally been reported ^{21, 22}. In those leiomyomas with FH gene mutations, some have characteristic histologic features, defined as SMT-FH ^{10, 11} (Table 4). These tumors are typically cellular with eosinophilic macronucleoli, perinucleolar halos, and eosinophilic cytoplasmic globules. They also frequently show branching staghorn vessels. Miettinen et al. examined FH IHC staining in different types of uterine smooth muscle tumors and found 37% of LM-BN were negative for FH by IHC (Table 4). S-(2-succino)-cysteine (2SC) is another protein that may be used as a sensitive and specific marker for fumarate in the setting of defective FH and can be used as a surrogate marker for FH by immunohistochemistry ¹⁰. Absent immunohistochemical staining for FH has also been shown to have high specificity when evaluating FH alterations ^{8, 10, 23}. Using the two IHC stains in combination, along with FH mutational analysis, appears to be an extremely reliable way to confirm FH alterations and mutations in LM-BN.

In our study, the 77 cases with an LM-BM diagnosis were divided into two subtypes based on cytologic features¹³, with type I LM-BN showing features similar to SMT-FH (Figure 1) ¹³. In total, there were 40 cases showing type I features and 37 with type II features. We found that 85% of type I LM-BN were strongly and diffusely immunoreactive for 2SC (Table 3, Table 3, FHFigure 2) and all but two of those cases also showed loss of immunoreactivity for FH. These findings suggest that most, if not all, type I LM-BN are related to FH alterations. We found that roughly 20% of all LM-BN had mutations and 37% (15/41) of those with 2SC positivity by IHC harbored FH mutations (Table 3). The underlying molecular mechanisms for FH alterations in the remaining cases needs to be further investigated. Our findings suggest that for those sporadic tumors originally diagnosed as LM-BN, careful evaluation of their nuclear features may be helpful in the potential differential diagnosis of an FH-related tumor. All type I LM-BN tumors and tumors with SMT-FH features should be tested for FH alterations by 2SC/FH IHC or by mutational analysis. Recently Joseph et al ⁹ proposed an algorithm for screening SMT-FH in young patients and we believe their approach should also be applied to all LM-BN cases. Since nearly 20% of the type II LM-BN in our study presented with FH alterations by both 2SC and FH IHC and none had recognizable features of SMT-FH, immunohistochemistry may be the only method for detection of FH alterations in this tumor subset.

As summarized in Table 4, the rate of FH mutations in HLRCC is quite high¹⁸ while the rate in sporadic LM-BN is largely unknown. The results from Miettinen et al ²⁴ concur with our current study, We have shown that FH alterations can be detected by immunohistochemistry for 2SC and that FH mutations appear to be disproportionately higher in LM-BN than in other types of sporadic uterine smooth muscle tumors including usual-type leiomyomas, although additional FH gene mutation analysis is still needed. Our current study reveals that over 20% (15/72) of LM-BN harbor FH mutations and over one-third with FH IHC alterations also have corresponding FH gene mutations (Figure 6). The subset that shows FH alterations by IHC without an associated genetic cause requires further study. In a recent study, Liegl-Atzwanger et al. reported that FH genomic deletions can be found in over 30 % LM-BN ²⁵. Preliminary data from our cohort also showed that in 4 cases of LM-BN with extracted DNA failed to amplify in any of the 10 FH exons, suggestive of FH genomic deletions.

2SC is a sensitive and specific immunohistochemical marker for detection of FH deficiency, whereas negative staining for FH by IHC is more specific. We found a high concordance rate between these two markers in FH-altered tumors, with 39 of 41 2SC-positive cases also showing loss of FH by IHC (Table 4). For those two cases with immunoreactivity for both 2SC and FH, we postulate that there are likely other molecular defects resulting in succinate accumulation in this small subset of LM-BN (~5%). Yet FH by IHC alone is sensitive enough to detect over 95% of FH alterations and it can be used alone for routine screening.

The correlation of 2SC and FH by immunohistochemistry and their concordance with FH gene mutations in uterine smooth muscle tumors is of great clinical interest. The most common type of mutations are missense mutations (57%), followed by frameshift and nonsense mutations (27%), with the remainder consisting of a diverse array of deletions, insertions and duplications ²⁶. In our current study, 36.6% (15/41) of 2SC positive LM-BN harbored FH gene mutations and they were mainly missense mutations (Figure 6). Since the FH gene is located in chromosome 1q and this region has been found to be deleted frequently in uterine smooth muscle tumors ^27–29, analysis of additional chromosomal alterations deserves further investigation.

This study examined FH alterations and mutations by immunohistochemistry and gene mutation analysis in 77 LM-BN. More than 50% of cases originally diagnosed as LM-BN were related to FH alterations and over 20% of them harbored FH gene mutations. FH alterations and mutations were more frequently identified in type I LM-BN, suggesting a different mode of tumorigenesis in this histologic tumor subtype and which may have important clinical implications in this subset of patients.