Case Report

A 57-year-old Caucasian woman with an unremarkable past medical history presented with lumbar pain. Physical examination revealed a mass on the right flank. An FDG-PET scan revealed a renal mass highly suggestive of malignancy, signs of liver spread, and pelvic and lumbar spinal cord bone metastases (Figure 1A). Tumor staging at initial diagnosis was pT1bN0M1 (stage IV). In April 2007, an open right radical nephrectomy was performed. The histopathologic report revealed Fuhrman grade 4 advanced ccRCC. According to the Memorial Sloan Kettering Cancer Center (MSKCC) prognostic model,¹⁴ the patient was classified as being in the intermediate-risk group based on 2 risk factors for the prognostic score: Karnofsky performance status of 70% and a serum lactic dehydrogenase level >1.5 times the upper limit of normal. She received palliative radiotherapy

View Full Size

(A) Initial tumor dissemination detected by PET/CT (arrows). (B) MRI showing skull metastasis (arrow), which was surgically removed in 2015.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 15, 11; 10.6004/jnccn.2017.7018

We performed WES in 3 different regions of the primary tumor obtained before treatment, in the skull metastasis resected 7 years after temsirolimus treatment initiation, and in the germline DNA obtained from the patient's blood. The mean depth of coverage was >80x for the 3 regions of the primary tumors, 136x for the metastasis, and 94x for the blood (see supplemental eAppendix 1 for details regarding methodology, available online with this article at JNCCN.org). The number of tumor mutations (single nucleotide variants and indels) leading to nonsynonymous coding or loss-of-function (LOF) variants in the primary tumor was 160. Tumor mutations refer to those detected in the tumor but absent in the blood. In total, 116, 76, and 65 mutations were detected in each of the 3 tumor regions and from these, 41 were shared (Figure 2; supplemental eTable 1). Among the shared tumor mutations, 39 were missense variants and 2 were LOF variants caused by frameshifts. The variants affecting genes frequently mutated in ccRCC were detected in all tumor samples analyzed: VHL (c.263G>T; p.W88L), BAP1 (c.1359dup; p.E454Rfs*15), and MTOR (c.5920T>C; p.Y1974H), and were validated by Sanger sequencing. In ccRCC the primary driver event is VHL inactivation, and thus VHL mutations are early alterations present in all tumor cells.¹⁵ The MTOR mutation frequencies detected in the different tumor samples sequenced were similar to those found for VHL, indicating that the MTOR mutation was also a truncal event in this patient (average ratio of 1.1; supplemental eTable 1).

View Full Size

Mutations revealed by whole-exome sequencing (WES). (A) Venn diagram indicating the number of tumor (absent in blood) loss-of-function and nonsynonymous coding variants identified through WES in 3 regions (R1–R3) of the primary tumor. Shared variants affecting relevant clear cell renal cell carcinoma genes are indicated below the arrow. (B) Representative genome images from the Integrative Genomics Viewer (Broad Institute), along with the number of reads for the reference (ref) and variant allele of the MTOR Y1974H mutation, and Sanger sequencing chromatogram.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 15, 11; 10.6004/jnccn.2017.7018

View Full Size

Functional assessment of MTOR p.Y1974H variant. Constructs expressing wild-type (WT) mTOR, mTOR Y1974H, and mTOR S2215Y (a positive control for mTOR activation) were transfected in HEK293 cells and the effect of phosphorylation of the downstream target S6K1 (pS6K1) was tested with Western blot. (A) Western blot analysis 24 hours after transfection cells were serum-starved for 24 hours (Starving) and then stimulated with 1% or 10% fetal bovine serum for 2 hours (Stimulation). In the Starving assay, lanes 1–2, 3–4, and 5–6 are duplicates from different experiments. (B) Cells transfected with mTOR constructs were treated with 2.5, 12.5, and 100 nM of rapamycin (rap) for 6 hours.

Abbreviation: GAPDH, glyceraldehyde 3-phosphate dehydrogenase.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 15, 11; 10.6004/jnccn.2017.7018

To assess the effect of the MTOR novel variant p.Y1974H, HEK293 cells were transfected with constructs expressing the wild-type mTOR protein or the Y1974H variant protein. The mTOR-activating mutation S2215Y was used as control for the experiments. The Thr389 phospho-S6K1 (pS6K1) levels, which represent TORC1 activity, were augmented for the Y1974H variant compared with mTOR wild-type, and similar to those obtained for S2215Y (Figure 3A), indicating that mTOR Y1974H activates mTORC1. In addition, we observed that mTOR Y1974H was sensitive to rapamycin (Figure 3B). mTOR Y1974H also led to an increased number of cells in S phase with a concomitant decrease in G0/G1, and an alteration in forward scatter (data related to an alteration in cell size) compared with mTOR wild-type and similar to mTOR S2215Y (supplemental eFigure 1).¹⁶ Immunohistologic analysis of the tumor samples revealed activation of the mTOR pathway through a positive immunostaining of phospho-S6, a downstream target of mTOR (Figure 4A–C), with the metastasis staining being more intense than that in the primary tumor. The staining of p-ERK (Figure4 D–F) was weak in the primary tumor, whereas the metastasis exhibited a nuclear and cytoplasmatic intense and extensive staining.

WES of the metastasis that arose 7 years after starting temsirolimus treatment revealed 80 tumor nonsynonymous coding or LOF variants (71 missense, 2 in-frame deletions, and 7 LOF) present in the tumor and absent in the blood. Of these, 51 were shared with the primary tumor and 29 were exclusively present in the metastasis (supplemental eTable 1). Among the genes exclusively mutated in the metastasis, only 3 (CRTC3, KAT6B, PBRM1) were in the Cancer Gene Census (CGC) but none were directly related to the MTOR pathway, and the variants have not been previously described in tumors, according to the COSMIC database.

The full region of the MTOR FKBP-rapamycin binding domain (FRB), including intronic regions, was sequenced by Sanger. However, no metastasis-specific variants were detected.

Discussion

Rapalogs are drugs approved for metastatic ccRCC in first-line treatment¹ and pretreated patients.² Although recent studies have shown less efficacy of rapalogs compared with nivolumab³ and cabozantinib,⁴ combination of everolimus and lenvatinib has become a second-line standard,^5,6 and ongoing trials are assessing combinations in first-line treatment. Interestingly, monotherapy with mTOR inhibitors is very useful in a considerable subgroup of patients, with long-term responses, regardless of the line of treatment.^10,12 Mutations affecting mTOR pathway genes have been explored as possible biomarkers of activity for these drugs, and despite initial descriptions of highly sensitive patients with mutations in MTOR, TSC1, and TSC2,^7–9 more recent studies show only

View Full Size

Immunohistochemical (IHC) study of TSC2 and phospho-S6. Representative pictures of phospho-S6 IHC staining (original magnification x20) in (A) nontumoral kidney, (B) primary tumor, and (C) bone metastasis from the patient. Phospho-ERK staining (original magnification x20) in (D) non-tumoral kidney, (E) primary tumor, and (F) bone metastasis.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 15, 11; 10.6004/jnccn.2017.7018

This report presents a patient with metastatic ccRCC who experienced an unusual prolonged response to temsirolimus. Multiregion sequencing of the tumor, functional studies on a candidate variant, and immunohistologic characterization of the tumors allowed the identification of an MTOR mutation (p.Y1974H) as causative for the response. This mutation is located in the FAT (FRAP-ATM-TTRAP) domain of mTOR, a region that has a prominent cluster of hyperactivating mutations in RCC that lead to an increase in mTORC1 and mTORC2 activities.^17,18 These mutations may destabilize the structure of the FAT domain, directly de-regulating mTOR kinase activity and affecting the binding of mTOR complex proteins.¹⁸ As expected for ccRCC, a VHL mutation (p.W88L) was found in all regions of the primary tumor and in the metastasis, in agreement with an early event.^15,19 Additional alterations in genes frequently mutated in clear cell histology were present in BAP1 and MTOR.²⁰ The MTOR mutation (p.Y1974H) was detected in the 3 regions of the primary tumor and in the metastasis with similar frequency to VHL mutation, the major ccRCC driver event (supplemental eTable 1), and functional in vitro assays confirmed it was an activating mutation. ccRCC intratumor heterogeneity has been shown to affect the mTOR pathway,¹⁵ and the moment during clonal evolution in which the mutations occur has been suggested to impact drug response.^12,13 Our results are consistent with a truncal MTOR mutation, which would render the tumor ubiquitously addicted to mTORC1 hyperactivity. In addition, the BAP1 mutation found in the patient suggested a poor outcome, because these mutations are associated with poor RCC prognosis.^21,22 The remarkable response obtained with third-line temsirolimus highlights the high sensitivity to this drug.

Interestingly, the patient developed a bone metastasis while receiving temsirolimus. To date, there is one report identifying a mechanism for acquired mTOR inhibitor resistance, which consisted of a secondary mutation (p.F2108L) in mTOR FRB domain.⁹ In the present patient, the only MTOR mutation found in the metastasis was p.Y1974H. Sequencing of the intronic region of MTOR FRB domain ruled out mutations that may lead to alternative splicing events altering this region. Regarding other genes, WES revealed mutations exclusive of the metastasis; however, none of the mutated genes was directly related to the mTOR pathway. IHC of the tumors revealed more intense staining for phospho-S6 in the metastasis than in the primary tumor, suggesting that the metastasis had acquired an additional alternative mechanism hyperactivating mTORC1. Phospho-ERK–strong IHC staining in the metastasis (Figure 4F) suggested the mechanism could be connected with the Ras-ERK pathway; however, none of the metastasis mutations were directly linked with this pathway. At any rate, the resistance mechanism may involve alterations not detectable by DNA sequencing (eg, epigenetic changes).

Conclusions

Although patients with extraordinary responses to mTOR inhibitors have been reported, the mechanisms underlying these responses are still not well understood. In this study, multiregional WES directly linked a truncal novel activating MTOR mutation with the exceptional response to temsirolimus in a patient with ccRCC. Our results support the sequencing of multiple tumor areas to identify early mutations affecting the mTOR pathway, to ultimately reveal predictive markers able to personalize RCC treatments.