Mutation profile and immunoscore signature in thymic carcinomas: An exploratory study and review of the literature

Introduction

Thymic epithelial tumors (TETs) are still considered rare tumors, with an incidence of 0.15 cases out of 100 000 per year which account for only for 0.2%–1.5% of all malignancies.¹ Thymic carcinomas (TCs) are more rare, representing approximately 15% of all TETs.² The rarity of these tumors has precluded the development of large phase II and III clinical trials for many years, thus delaying the investigation of new drug therapies. Moreover, the etiology of TETs is unknown. Over the past few years, significant efforts have been made to investigate the underlying molecular pathways involved in thymic carcinogenesis. Consequently, molecular analysis has facilitated the identification of important oncogenes (EGFR, HER2, IGF1R, KIT, KRAS, and BCL2), tumor suppressor genes (TP53, CDKN2A), chromosomal aberrations (LOH3p,6p,6q,7p,8p), angiogenic factors (VEGF), and tumor invasion factors (matrix metalloproteinases and tissue inhibitor of metalloproteinases) as major contributors to disease development.³ Subsequent studies reported, by using different platforms, the association of distinct molecular clusters with different histological subtypes.^4-8 In particular, TCs have been found to display more chromosomal losses and gains and occasionally harbor somatic mutations in KIT.⁵ Despite the limited number of samples analyzed in these studies, unique molecular changes in TETs have been found, such as dysregulation of antiapoptotic genes and mutations in genes involved in histone modification, DNA methylation, and chromatin remodeling.⁹ More recently, Radovich and colleagues performed a multiplatform, comprehensive analysis of TETs as part of The Cancer Genome Atlas (TCGA) project to uncover the integrated genomic landscape of these rare tumors.¹⁰ They found that thymomas have the lowest mutational burden among adult cancers. However, the number of TCs analyzed was limited (only 10 cases) and these genetic findings do not significantly affect the clinical practice. Therefore, improving our understanding of the molecular biology of these rare tumors remains a key challenge.

The aim of this study was to evaluate the molecular profile and the immunoscore of a series of 15 TCs, and to correlate the identified molecular profiles with clinical characteristics and patient survival data.

Methods

Results

Discussion

The results of our study are consistent with the literature data confirming a low rate of mutations in TC.⁵ In fact, one third of our patients did not show any mutation in the investigated genes, whereas one patient was a carrier of three variants, four patients were carriers of two variants, and the remaining five patients only showed one variant each. Although the panel of genes used was limited (only 50), the sensitivity of our test was adequate, considering that the overall mean coverage was more than 1500X. Certainly, the analysis of a greater number of tumors and/or genes may have highlighted other significantly mutated genes or additional clinically, or biologically relevant mutations in other genes outside our panel.

Histology of TC has been historically separated into low- and high-grade lesions.²³ Even if the clinical impact of this grading has been questioned, the results of the present study suggest that it may be indicative of tumor biology. Indeed, low-grade lesions were characterized by the absence of mutations; high-grade lesions without mutations showed the highest immunoscore level; finally, among high-grade lesion with mutations, those with FGFR3 mutations showed a proliferative fraction (as evaluated by Ki67) lower than the remaining cases, as opposed to those with CDKN2A, which had a higher proliferative fraction.

In our 15 tumor samples, the most recurrently mutated genes were FGFR3 and CDKN2A and the only recurrent mutation was p.Ser249Cys in FGFR3. Despite reports in the literature suggesting that mutations of TP53 are frequent in TCs (approximately 25%), we did not observe point mutations in this gene. Interestingly, null mutations in APC (frameshift) and CDKN2A (two nonsense) showed a percentage of reads corresponding to the mutated allele higher than expected for a heterozygous somatic mutation (Table 3), suggesting the possible combination, in trans, with a gene deletion. The same higher-than-expected percentages of mutated allele were found for CDKN2A by Wang et al.⁹ and Enkner et al.²⁴; moreover, this is compatible with the identification of deletions involving the CDKN2A genomic regions in TCs.^{7, 10}

Figure 1 summarizes our data and that in the literature evaluating gene mutations in wide gene sets (at least 50) in different series of TCs. All the series of TCs explored were small (range 10–47 cases)^{7, 9, 10, 24} and showed a low rate of mutations (percentage of tumors with at least one mutation ranging from 20% to 67%). Out of a total of 41 genes found mutated in TCs, the most recurrently mutated were TP53 (39% of all mutated tumors; 21% of all analyzed TCs) and CDKN2A (17% and 9%), confirming that TCs are a molecularly heterogeneous group of tumors driven by a limited number of genomic events. Of the 66 mutated tumors, 34 showed just one mutation, and the others were characterized by the presence of two to four mutations; one sample showed an exceptional number of 12 mutations, accordingly with the presence of two deleterious variants in the ATM gene. We are aware that this mutational landscape derives from heterogeneous studies, which analyzed different – although overlapping – target regions.

As observed in previous studies, patients carrying CDKN2A mutations (three out of 15 patients, 20%) showed a statistically significantly worse survival.^{25, 26} The CDKN2A gene is located on chromosome 9p21.3 and encodes p16 (INK4a) and p14_ARF. The p16 protein is a negative regulator of cell cycle progression which blocks cyclin-dependent kinase 4 (CDK4) and 6 (CDK6), whereas the p14_ARF activates the TP53 tumor suppressor. A CDKN2A inactivating mutation may lead to activation of cyclin dependent kinases, thus permitting an inappropriate progression through the cell cycle and promoting uncontrolled tumor cell proliferation. Currently, several inhibitors for these kinases are being investigated in clinical trials for various malignancies and might constitute a therapeutic option. In two phase II trials, it has been reported that milciclib (PHA-848125AC), a potent inhibitor of CDKs, is safe, well-tolerated, and achieved disease stabilization in a majority of patients with TC and B3 thymoma.²⁷

The patients in our study carrying FGFR3 mutations (4/15 patients, 26.6%) showed a statistically significantly better survival. Enkner et al. observed an FGFR3 missense mutation in two out of 35 TCs,²⁴ while no FGFR3 mutations were reported in 42 TCs screened with a 197-gene panel including FGFR3.⁹ FGFR genes are deregulated in solid tumors by amplification, translocation or mutation and mutations and fusions in FGFR2/3 can lead to constitutive FGFR signaling that may contribute to carcinogenesis.²² FGFR3 mutations have been found to be particularly frequent in bladder cancer (20%), where the results showed an association with early stage, low grade, and better survival.²⁸ Several FGFR inhibitors are currently under investigation in clinical trials. In a phase II trial, erdafitinib, a potent tyrosine kinase inhibitor of FGFR 1–4, was found to be associated with an objective tumor response in 40% of previously treated patients who had locally advanced and unresectable or metastatic urothelial carcinoma with at least one FGFR3 mutation or FGFR2/3 fusion.²⁹ In a recent phase II trial, lenvatinib, a potent inhibitor of receptor tyrosine kinases, targeting vascular endothelial growth factor receptors (VEGFR1-3), FGFR1-4, KIT, and RET, showed clinical efficacy in 42 patients with advanced or metastatic TC, with an overall response rate of 38%, disease control rate of 95.2%, and median duration of response of 11.6 months.³⁰ Thus, inhibition of FGFR3 might represent an intriguing novel target in a subset of TCs. Moreover, patients with urothelial carcinoma carrying FGFR alterations may be less likely to have a response to immunotherapy than are those without such alterations.^{29, 31} Considering that pembrolizumab showed an interesting activity in patients with TC,^{32, 33} it may represent a potential predictive biomarker.

In our series, a KIT putative splicing variant (GnomAD allelic frequency 3.98 × 10⁻⁶) was detected in one patient. KIT mutations are a therapeutic target for kinase inhibitors such as imatinib, and represent the only known druggable targets in TCs, as evidenced by few but encouraging case reports.²⁴ However, the data in the literature confirm a low rate of KIT mutations in TCs.^{7, 9, 24}

Only one of the patients in our study harbored a NRAS mutation. At present, the RAS oncogenes are still not druggable targets.

The immunoscore was >2 in five patients (33%), whereas the PD-L1 expression by immunohistochemistry was positive in eight patients (53%), which was consistent with previous data in the literature.^{24, 32, 33} No statistically significant correlation between immunoscore and PD-L1 expression, mutations, stage of disease, or survival outcomes was observed. An absence of survival difference between positive and negative PD-L1 expression in patients with TC has also been described in other studies.^{24, 34} On the other hand, considering the correlation of PD-L1 expression by tumor cells with the likelihood of response to anti-PD-1/PD-L1 therapy, immune checkpoint inhibitors might be an active treatment option for unresectable or relapsed TCs. In fact, recently, two phase II trials showed promising activity of pembrolizumab as monotherapy in TCs and this activity correlated with PD-L1 expression.^{32, 33}

The major limitations of our study are: (i) the use of archive FFPE samples, for which DNA extraction may suffer from the degradation of nucleic acids by formalin fixation as well as formalin-induced sequence artifacts; however, we carefully quality-checked the extracted DNA with the deep sequencing step (given that the artifacts are random, they become less apparent as sequencing depth increases); (ii) the small number of analyzed cases; and (iii) the limited number of genes analyzed, precluding therefore larger and deeper analyses to discover other significantly mutated genes or additional clinically or biologically relevant mutations. However, our study confirms the low rate of mutations in TC compared with other solid tumors and suggests FGFR3 and CDKN2A mutations as intriguing potential therapeutic targets. A better understanding of the molecular architecture of thymic neoplasms may potentially impact on disease classification, targeted therapeutic decision-making, and the design of future clinical trials.

Disclosure

No authors report any conflict of interest