Functional apoptosis profiling identifies MCL-1 and BCL-xL as prognostic markers and therapeutic targets in advanced thymomas and thymic carcinomas

Thymomas (TH) and thymic carcinomas (TC) are rare epithelial tumors of the thymus (TET). The World Health Organization (WHO) classifies TH into type A, AB, B1, B2, and B3 according to the morphology of the neoplastic epithelial cells and the proportion of immature lymphocytes [1]. TH show organotypic features, such as epithelial cells with cortical and medullary differentiation [2] and the capacity to promote the maturation of thymocytes. In contrast, TC have lost these features and are essentially indistinguishable from carcinomas in other organs.

Radical surgery is the only curative treatment, but advanced TH and TC often require a multimodal approach with radio-chemotherapy [3, 4]. Chemotherapy is the most common treatment for relapsed TET. Tyrosine kinase inhibitors such as sunitinib may be an option for relapsing TC [5, 6]. However, although the overall response rates are typically high [7, 8], most tumors eventually become refractory to these treatments.

The biology of TET is poorly understood [9‐13]. Although recent multi-omics molecular studies revealed a few recurrent mutations, such as alterations of GTF2I, HRAS, and NRAS in TH, and rare mutations of epigenetic modifier genes in TC, TET are among the adult cancers with the lowest tumor mutational burden (TMB) and are not driven by mutations in common oncogenes [10, 14]. A significant obstacle to a better understanding of TET biology is the lack of suitable in vivo or in vitro tumor models. Thus, functional profiling of primary tissue samples could be the key to identifying novel vulnerabilities in TET. BH3 profiling is a tool that allows interrogating dependencies in the mitochondrial apoptosis pathway of freshly isolated viable tumor cells [15]. This assay measures the readiness of the cell to undergo apoptosis and their dependence on pro-survival BCL-2 family proteins. In brief, living cells are exposed to synthetic BH3 peptides that mimic the activity of endogenous pro-apoptotic proteins. By titration of either activator or sensitizer peptides, the level of mitochondrial apoptotic priming towards apoptosis can be quantified. The clinical availability of so-called BH3 mimetic drugs [16, 17] makes upfront dynamic BH3 profiling ideal for targeted therapeutic approaches and helps select patients and minimize cytotoxicity. The key regulators of the mitochondrial apoptosis pathway, the BCL-2 family proteins BCL-2, BCL-xL, BCL-w, and MCL-1, bind to BAK or BAX and prevent the oligomerization and pore formation at the outer mitochondrial membrane [18]. The BCL-2 family interaction network reprogramming is closely related to cancer survival and therapeutic resistance [19, 20]. The inhibition of single anti-apoptotic factors like MCL-1, BCL-2, and BCL-xL was successfully used in leukemia [21‐23] and was proposed in several solid cancers [24‐26]. Apoptosis-related genes, including BIRC3, NOXA, MTCH2, and cFlip, have already been proposed to be involved in tumor progression and drug resistance, especially in B3 TH and in TC [27, 28]. In the healthy thymus of mice, MCL-1 is required for the survival of mature cortical and medullary thymic epithelial cells and the maintenance of the thymic architecture [29]. Previous studies described a copy number gain of the MCL-1 gene locus on chromosome 1q in about 51% of all subtypes with increased frequencies in B3 (70%) and TC (57%), as well as copy number gains of the BCL-2 gene locus (18q21.33) in up to 42% of all TC. High expression of BCL-2 also correlates with decreased survival and was suggested as a potential therapeutic target in advanced TH and TC [30].

In this study, we explored the relevance of pro-survival BCL-2 family proteins in TH and TC. Immunohistochemistry was used to measure the expression of TP53, BCL-2, MCL-1, BCL-xL, and NOXA in a set of 62 TH and TC with known clinical follow-up and to correlate the results with survival. BH3 profiling was used to investigate apoptotic priming and to directly measure the effect of inhibition of MCL-1 by AZD5991, BCL-xL by A-1331852, and BCL-2 by Venetoclax (ABT-199) in primary patient samples. We suggest that the specific inhibition of MCL-1 and BCL-xL alone or in combination may be a promising and realistic treatment option for selected TET patients.

The virtual absence of common driver mutations in TH and TC [10] points to alternative oncogenic mechanisms in the pathogenesis of TET, but the lack of suitable cellular or animal models has hampered the search for new treatment options in these tumors. Suppression of apoptosis, e.g., through upregulation of pro-survival proteins, is one of the most important hallmarks of cancer [19]. In this context, functional BH3 profiling is an extremely useful tool to identify tumor types that depend on targetable anti-apoptotic BCL2 family members like BCL-xL and MCL-1 for their survival. Importantly, BH3 profiling interrogates ex vivo isolated viable tumor cells for their readiness to undergo apoptosis or dependence on pro-survival BCL-2 family proteins and is, therefore, an ideal bridge for tumors such as TET where no cellular models are available. Previous studies have already shown altered expression of apoptosis-related factors such as BIRC3, NOXA, MTCH2, and cFlip in TH and TC [27, 28, 44] and that the inhibition of BCL-2 family proteins, including MCL-1 and BCL-xL, induces apoptosis in TET cell lines [30].

We here confirm previous observations by Petrini et al. [30] that many TH and TC show amplification of the MCL-1 gene locus on chromosome 1q. Our findings extend these data by showing that TET belong to the growing family of solid tumors that depend critically on MCL-1 and BCL-xL for survival and chemotherapy resistance. We believe that these observations could and should be exploited for therapeutic purposes with BH3 mimetics, but the heterogeneity of these tumors will make upfront BH3 profiling mandatory for patient selection. Furthermore, our findings indicate that any clinical trial using BH3 mimetics in TET should ideally address both MCL-1 and BCL-xL.

MCL-1 is necessary for the maintenance of cortical and medullary thymic epithelial cells [10, 29], suggesting that it may also be relevant for TET development. The earliest study by Chen et al. showed frequent co-expression of MCL-1 and BCL-2 in many aggressive TET [45]. Our IHC analysis of 62 TET samples with known clinical follow-up identified MCL-1 and BCL-xL as prognostic markers. Notwithstanding the inherent limitations of a retrospective study based on relatively small case numbers, we found that the combination of strong MCL-1 and BCL-xL expression together with low levels of NOXA was particularly informative to identify patients with poor OS. In line with a previous report by Petrini et al. [30], overexpression of MCL-1 correlated with increased copy numbers of the MCL-1 gene. As pointed out previously [30], the MCL-1 copy number alteration is probably the result of gains of the whole 1q arm rather than focal MCL-1 gene amplification. The gain of 1q has long been recognized as a cytogenetic hallmark of aggressive TET [46].

Treatment with single agents like Venetoclax has been successfully used in hematopoietic cancers [32, 37, 40] but was less effective in solid tumors, including breast cancer, SCCL, and melanoma cell lines [38, 47, 48]. However, the situation in melanoma resembles TET since MCL-1 and BCL-xL are considered the main pro-survival factors [38]. Single inhibition of either factor in melanoma cells did not have a significant effect, and only the combined inhibition resulted in the induction of apoptosis. In the same line, we found strong priming for MCL-1 in the chemo-resistant TC cell line 1889c using dynamic BH3 profiling. Only the combination of the MCL-1 inhibitor AZD5991 with either A-1155463 or ABT-263 but not a single treatment induced a potent caspase-dependent apoptosis and almost entirely eliminated 1889c in a nanomolar range. The siRNA knockdown of MCL-1 and/or BCL-xL supports the model that a certain threshold of pro-survival proteins must be inhibited to induce apoptosis and that other BCL-2 family members sequester pro-apoptotic proteins [49]. It has been observed that the single inhibition of BCL-2/BCL-xL leads to treatment resistance by increased MCL-1 or BCL-xL levels and redistribution of BIM [36, 40]. Our results show that BIM was sequestered away from MCL-1 and shifted to BCL-xL as an escape mechanism in 1889c treated with AZD5991. These findings indicate that in most cases the inhibition of MCL-1 alone will not be sufficient to induce satisfactory clinical effects in a therapeutic setting due to a dynamic synergy with BCL-xL as proposed before [50].

Thrombocytopenia or liver damage as a specific side effect of BCL-xL inhibitors [17, 49] makes combinational treatment difficult. However, He et al. recently described an interesting approach to circumvent BCL-xL inhibition toxicity by a BCL-xL degrader with a better anti-tumor activity [51].

Importantly, BH3 profiling of primary human tissue samples revealed strong priming for MCL-1 and/or BCL-xL in malignant type B2 and B3 TH, but not in SCCL. In line with these measurements, only TH but not SCCL samples showed cytochrome c release after treatment with the inhibitors ABT-199, A1331852, and AZD5991.

In summary, these findings indicate that TET are exquisitely dependent on pro-survival factors such as MCL-1 and BCL-xL and that suppression of apoptosis is a major functional hallmark of these non-oncogenic driven tumors. This observation makes TET ideal candidates for clinical trials with BH3 mimetics that ideally address both MCL-1 and BCL-xL. Given the observed marked heterogeneity of TET concerning their dependence on pro-survival BCL-2 family proteins, upfront BH3 profiling should be used to select those patients with the most significant benefit and to tailor the optimal therapy with the least possible toxicity in a targeted therapeutic approach.