EIF1AX mutation in thyroid tumors: a retrospective analysis of cytology, histopathology and co-mutation profiles

To our knowledge, this is the largest cohort of thyroid FNA samples with indeterminate cytology (Bethesda III and IV) harboring EIF1AX mutations with surgical follow-up.

Frequency of EIF1AX mutation in thyroid nodules

Amongst a total of 621 thyroid FNA specimens of indeterminate cytology with molecular testing, the frequency of the EIF1AX mutation was 4.99% which is similar to previously reported rates of 4.5% (14/904) and 4.2% (27/647) [2, 11]. As with previous studies, the most common mutation in the EIF1AX gene was the A113_splice site. In addition to previously described exon 2 mutations, we identified new exon 2 mutations not previously mentioned in the literature. One new mutation identified, the p.N17_K23dup, was associated with malignant histopathology (minimally invasive HCC).

Association of EIF1AX mutation with malignancy

In our study, of the 31 Bethesda III and IV nodules with EIF1AX mutation, 17 were benign, 2 were borderline, and 12 were malignant on surgical follow-up, conferring an overall risk of malignancy (ROM) of 38.7% and a ROM or borderline of 45.2%.

Our results align with previous findings that EIF1AX mutations occur in both benign and malignant thyroid nodules, and that they also coexist with other driver mutations including NRAS, HRAS, TP53, TERT, and PIK3CA in a significant proportion of cases.

Though 8/12 (67%) of malignant nodules harbored the A113_splice mutation, it was not significantly associated with malignancy when compared to non-A113_splice mutations (p = 0.106). Moreover, all aggressive malignancies had the A113_splice mutation (n = 4), but the presence of the A113_splice EIF1AX mutation was not statistically associated with more aggressive malignancies, such as PDTC or ATC. Singular A113_splice mutations (without co-mutations), were not statistically associated with malignancy but did show a strong correlation with malignancy (p = 0.059). We highlight that out of the 12 singular non-A113_splice mutation tumors, 11 were of benign pathology. The statistical significance may not have been achieved due to small number of samples.

Role of EIF1AX and co-mutations in malignancy risk stratification

When considering only cases with an EIF1AX mutation in isolation (i.e., no co-mutation), the ROM was 25% (6/24) and ROM or borderline was 33.3% (8/24). In the literature, isolated EIF1AX mutation is estimated to carry ~ 20–36.4% of ROM. In their recent study of 26 cases, Gargano et al. found a ROM of 36.4% (4/11) and a ROM or NIFTP of 54.5% (6/11) for cases with EIF1AX mutation alone and no other coexisting mutations [2, 11]. In their study, Karunamurthy et al. found EIF1AX mutations in 27/647 (4.2%) of indeterminate cytology samples, of which only 5 had surgical follow-up information (1 Encapsulated follicular variant PTC, 1 hyperplastic nodule, and 3 FA) yielding an estimated ROM of 20%. Therefore, the clinical value of finding an EIF1AX mutation in isolation in a thyroid nodule with indeterminate cytology (Bethesda III/IV) is limited as it does not change the ROM conferred by FNA significantly (~ 10–40% for Bethesda III/IV).

As the EIF1AX mutation alone is limited in bettering the malignancy risk stratification, co-mutation profile with EIF1AX plays a more integral role in the risk evaluation. In our study, the coexistence of EIF1AX mutations with other driver pathogenic mutations such as RAS, TERT and TP53 conferred an 86% ROM (6 out of 7 cases). Similarly, Gargano et al. [11] found that EIF1AX mutation with RAS mutation (seen in 26.9% of their cases) conferred a 71.4% ROM (85.7% with NIFTP), while EIF1AX with any other mutation (eg., TP53, TERT, RAS + TERT or TP53, BRAF fusion, seen in 30.6% of their cases) conferred a 100% ROM. Similarly, Karunamurthy et al. found co-mutations in 3/11 cases (2 cases with only NRAS and one with NRAS, TP53 and TERT), all of which were malignant neoplasms [2]. Therefore, surgical intervention with at least lobectomy should be considered for patients with such co-mutation profile.

Given the high allelic frequency of EIF1AX mutations in comparison to that of co-existing mutations, it has been suggested that EIF1AX mutations represent an early event, at least in some cases, that promotes initiation of the thyroid tumors and malignant transformation [11]. In other words, EIF1AX mutations alone are not sufficient for full transformation, but requires other mutations, particularly RAS, for progression to overt malignancy. Therefore, the clinical significance of finding an isolated EIF1AX mutation in a thyroid nodule, even though the immediate outcome is likely to be benign, is uncertain. Even nodules that histologically look like hyperplastic/adenomatoid nodules and were classified as such (8 cases in our series) can be clonal tumors.

The type of EIF1AX mutation also appears to render different outcomes with co-mutation (Table 3). Notably, the co-occurrence A113_splice mutation with RAS correlated with malignancy and aggressive tumor behavior in our study. All 4 samples with RAS co-mutations in our study were malignant. Moreover, one tumor with A113_splice and RAS mutation displayed PDTC characteristics. Therefore, we also confirm the existence of RAS mutation with EIF1AX and its association with more poorly differentiated malignancies [1, 2, 7, 12]. In contrast, exon 2 mutations without RAS mutation were only seen in benign, borderline and PTCs [2, 13].

Table 3 Compilation of malignancy risk stratification according to the EIF1AX mutation type and co-mutation profile from current study, Gargano et al. [11], and Karunamurthy et al. [2]Histopathology of nodules harboring EIF1AX mutation

In the literature, the frequency of EIF1AX mutations in all PTCs is reported to be 1–2.5% [2, 4,5,6]. The majority of the studies found these tumors to be predominantly encapsulated follicular variant [1, 2, 11]. In our study, we had a similar proportion of encapsulated FVPTC and infiltrative FVPTC (10% each). In addition to FVPTCs, most other cases with EIF1AX mutation, both benign and malignant, were follicular phenotype (i.e., with a follicular growth pattern) including hyperplastic/adenomatoid nodules, FA, NIFTP, WDFT-UMP, and Hürthle cell neoplasms. Similarly, Gargano et al. found most of their EIF1AX mutated tumors to be follicular phenotype with or without papillary nuclear features and/or oncocytic features (FA, FTC, NIFTP, FVPTC, and Hürthle cell) [11]. These findings are to be expected since these neoplasms are the most common histologic follow-up for thyroid nodules classified as Bethesda III and IV on cytology. In addition, NIFTP, encapsulated and invasive FVPTC cannot be distinguished on cytology, as their distinction requires the histologic evaluation of the whole tumor.

In our study, a significant proportion (38.7%) of both benign and malignant nodules were oncocytic or had oncocytic features, including 8/12 malignancies and the four cases (13% of nodules) with a PDTC component. EIF1AX mutation was previously reported with a high incidence in HCC, highlighting the importance of this gene for this particular histopathology. In a comprehensive genomic characterization of 56 primary HCC tumors, Ganly et al. [14] identified EIF1AX mutations in 11% of cases, a frequency similarly seen in PDTC and ATC. The co-occurrence of EIF1AX and RAS mutations was not observed in their study. In a recent study where molecular testing was performed in 85 thyroid FNA specimens that were Bethesda IV-suspicious for a Hürthle cell neoplasm, EIF1AX mutations were found in 5 cases and, of those, three had surgical follow-up, two of which were HCC (including one with a co-TERT mutation) and one showed multinodular goiter on histologic evaluation [15]. Five other cases of HCC with EIF1AX mutation were reported previously. All were of A113_splice mutations, two of which had a co-mutation of TP53 [11, 16, 17]. In contrast, Karunamurthy et al. did not find EIF1AX mutations in their study of 53 FTCs, including 22 HCCs.

Study limitations

There are many limitations to our retrospective study. Firstly, the study was limited to Bethesda III/IV nodules with surgical follow-up. As a result, we were not able to report the frequency of EIF1AX mutations in specific types of thyroid neoplasms including conventional PTC or ATC that usually correlate with Bethesda V or VI categories. Secondly, the estimated ROM calculated in studies such as this one may overestimate the actual ROM, due to the impact of selection bias. Nodules that undergo surgical resection are more likely to have suspicious pre-operative clinico-radiological findings, increasing the likelihood of malignancy regardless of the FNA diagnosis and molecular result. We note, however, that while 55% of thyroid nodules within this study were benign at the time of surgery, their natural history, including the possible progression to a malignancy and/or the acquisition of a second mutation (eg., TP53 or TERT) if they were not removed, is unknown. Indeed, EIF1AX mutations may represent an early event that promotes initiation of the thyroid tumors and malignant transformation in a subset of cases. In similar regards, we did not have long-term follow-up of the patients with malignancy. This is of particular importance as, EIF1AX-mutated PDTCs showed significantly shorter survival and were present in larger tumors when compared to wild type EIF1AX PDTCs [7]. Finally, although our study had the largest collection of EIF1AX mutations with histopathology correlation, the sample size remains small and necessitate further meta-analyses.

留言 (0)

沒有登入
gif