Background
Clear-cell renal cell carcinoma (ccRCC) is the most common type of RCC and poses a major health threat worldwide [
1]. Whilst ~ 70% of cases are localized or locally advanced that can be treated with extirpative surgery with or without adjuvant tyrosine kinase inhibitors (TKIs) that entails prolonged survival [
2‐
4], ~ 30% of cases are diagnosed with initial metastatic disease with a median overall survival of 43.2 months, 22.5 months and 7.8 months corresponding to favorable to poor risk groups by International Metastatic RCC Database (IMDC) criteria [
5]. TKIs target tyrosine kinases which are enzymes responsible for the activation of a variety of proteins by signal transduction cascades [
6]. The proteins are activated by phosphorylation, a step that TKIs inhibit [
6]. Most TKIs are typically used as anticancer drugs acting both in cancer-intrinsic and/or micro-environmental setting [
7]. The IMDC risk group category reflects comprehensive survival data worldwide in the era of TKIs for metastatic renal cell carcinoma (mRCC) and in spite of newly emerging immune checkpoint inhibitors (ICIs) taking up frontlines of mRCC treatment, TKIs still play critical roles as TKI + ICI combo now stands at the standard of care for mRCC as revealed by latest randomized trials [
8‐
11].
Copy number alteration (CNA) plays critical role in carcinogenesis. Unlike point mutations that requires persistent selection stress, DNA rearrangements can dynamically alter more swiftly throughout cancer evolution. Thus far, the 2 massive deep sequencing studies, namely the Cancer Genome Atlas (TCGA) and TRACERx (TRAcking Cancer Evolution through therapy (Rx)) projects have revealed instantaneous and evolutionary alteration of CNA in ccRCC, showing truncal events of 3p loss and 5q gain [
12‐
15]. Other CNAs that occur at a lesser frequency also play critical roles as the genotypes often represent distinct phenotypes. Chromosome 14q loss is reported to define a molecular subtype of clear-cell renal cell carcinoma associated with poor prognosis [
16]. Deletion of 18q has also been reported to confer worsened prognosis [
17]. Likewise, 9p loss has been validated in multiple studies to be associated with poor outcome [
18‐
21]. However, there are only a handful of studies that link CNAs with treatment response in ccRCC.
In the current study, we have investigated associations between sensitivity to commonly used tyrosine kinase inhibitors (TKIs) and CNAs in ccRCC using an in silico exploration with in vitro validation.
Discussion
In the current study, we identified several CNAs that were associated with cross sensitivity to TKIs in ccRCC. Although TKIs in ccRCC mainly target angiogenesis which was a major feature of the cancer, those multi-target agents also provided functional blockade to cancer-intrinsic pro-tumorigenic pathways. Those targets could overlap between TKIs and cross-resistance was common [
32]. In the current study, we sought to analyze association between CNAs and TKI sensitivity and we surprisingly found that cross-sensitivity associated with common CNAs in ccRCC. As we aimed only at minor CNAs besides 3p and 5q in our study, those CNAs could very likely be clonal due to treatment stress.
We first found +20q featuring overexpression of RBL1 and enriched FoxO signaling is associated with resistance to both Sunitinib and Cabozantinib, despite that in silico exploration shows resistance to all 4 TKIs. Common cancer-intrinsic targets between Sunitinib and Cabozantinib converge to Akt/PI3K signaling which has extensive crosstalk with FoxO signaling. We speculate that RBL1 participates in cell cycle regulation downstream of FoxO signaling and thus by pass the inhibition by Sunitinib and Cabozantinib, both of which were shown to mediate cell cycle arrest as the cancer-intrinsic mechanism [
33,
34]. On chromosomal level, there was a report that Chr20q amplification accounted for only 3.8% of 79 ccRCC cases who received whole genome profiling of chromosomal aberrations [
35], while another research identified amplification on 20q with a proportion of 20% based on 90 patients [
36]. Beroukhim et al. [
36] implemented statistical method GISTIC to identify potential copy number alterations as driver events of tumorigenesis, and screened out several peak regions of gain or loss. They suggested 20q be regarded as tumor suppressor gene (TSG) target due to more than 20 significantly overexpressed genes located on this chromosome arm. These genes include RGS19, TPD52L2, TNFRSF6B, C20orf11 PRPF6, BIRC7, RTEL1 and SOX18. However, it is probable that TSG targets of these expression-altered genes or copy-number changes were replaced by passenger events occurring nearby them. Moore et al. indicated that 25.5% of 412 ccRCC cases are detected 20q gain, which is significantly associated with stage rather than grade. Particularly, both grade and stage are associated with gain of 20q13, a region harboring TP53NK and SALL4 genes. Besides, Chr20 gain appears frequently in late-stage tumors and is observed only in high-grade disease. Interestingly, males more often harbor 20q gain [
37,
38].
Cross-resistance to Sunitinib and Sorafenib was also indicated in −14q yet in vitro assays only validated Sunintinb. Based on our findings that hypoxia pathway was enriched in only cases without −14q, we speculate that established target gene HIF1A on 14q plays a critical role upon its presence or deletion. We found that −14q cases show enriched immune modulation pathway, indicating that ccRCC without HIF1A may be more sensitive to ICI rather than angiogenesis-targeting TKIs. On the other hand, we found Wnt pathway genes were marginally enriched in -14q cases. Wnt signaling has been reported to play a role in Sunitinib resistance [
39,
40]. We primarily identified KLHL33 as potential target gene. Nonetheless, KLHL33 silencing was not associated with resistance to Sorafenib. We therefore consider −14q a relatively distinct genotype. This corresponds to previous reports. Basically, 14q has been considered a key attribution to poor clinical outcome and recurrence after surgery. Apart from the loss of 3p in 99% of samples, it was detected that loss of 14q accounted for 55% of samples and ranked the second most frequent copy number variations. It was also found that 14q loss is associated with significantly decreased overall survival and may be used as one of recurrence biomarkers for organ confined tumors [
16]. Notably, the evidence showed the association between 14q loss and diminished mRNA and protein expression of HIF1α, which is located on chromosome 14q23.2 and was often identified lost in high-grade ccRCC [
41]. Meanwhile, the stabilization of hypoxia inducible factor(HIF) is also influenced by the trunk event 3p loss and VHL gene inactivation in ccRCC. Moore [
38] et al. took array comparative genomic hybridization to recognize copy number alterations in ccRCC and found 14q loss took place in 46.8% of patients and had a strong association with tumor grade (p < 0.0001) and a lower extent with stage (p = 0.006). Furthermore, this type of chromosome aberration was found to be correlated with progression and metastases of disease for clear-cell RCC and may define a subtype of RCC [
42].
Interestingly, we found that loss of ARHGAP28 located on 18p is associated with cancer-intrinsic sensitivity to Cabozantinib and Sorafenib. ARHGAP28 plays an important role in Rho-mediated cell adhesion establishment. Whereas Sorafenib has been shown to mediate adhesion, there is so far no reports on Cabozantinib [
43]. However, given the cancer-intrinsic targets of Cabozantinib, both MET and AXL have been shown to participate in cell adhesion [
44,
45]. Unlike 3p loss and 5q gain, deep loss of 18p is much less frequent in clear cell renal cell carcinoma. A large study [
38] used array comparative genomic hybridization to identify clinicopathological characteristics attributed to CNA in ccRCC and reported that loss of 18p11.2-12 was associated with both tumor grade and stage. Similarly, another study [
35] of 80 cases showed evidence for close correlation between 18p loss and tumor stage (p = 0.038). In addition, the telomere lengths of tumor cells were measured and came to a conclusion that tumor samples with loss of heterozygosity (LOH) at chromosome 14q and 18p had shorter telomeres statistically to the borderline significance. Then we could hypothesize that 18p loss imply the malignancy of tumors. Indeed, Chr8 loss rarely occurred in early stage tumors and was observed appeared as tumor proceed and advance in stage, finally peaking at the last stage. This pattern may be caused by genomic material instability along with process of tumor, which denies the hypothesis that loss of Chr8 may act as driver event for tumor proliferation, invasion or metastasis. Furthermore, Chr18 loss was reported detective only when tumor developed into high-grade [
37]. Nevertheless, we can’t exclude the possibility that 18p loss is involved in tumorigenesis and progression.
Our study has limitations. First, the in vitro validation in our study is still premature. Substantially more functional analysis should be performed to consolidate the findings. Second, due to lack of ccRCC cell lines with specific CNAs queried in our study, we used RNA interference of potential target gene to simulate effect of chromosomal re-arrangement. Further studies include tissue sample studies that can better validate CNA and treatment response, and biological assays in patient-derived cells with CNA of interest.
In all, we postulate that +20q mediates cross-resistance via FoxO signaling. We also found that −14q was associated with resistance to Sunitinb and the enriched immune modulation could imply an increased response to ICI. Both alterations together with the frequency of CNAs in part elucidate the initial response rate of ccRCC patients to upfront TKIs. Finally, we reported that −18p could be sensitive to Cabozantinib via Rho-mediated cell adhesion regulation. All those findings warrant further mechanistic validation which is now under investigation.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.