Introduction

Renal medullary carcinoma (RMC) and fumarate hydratase-deficient renal cell carcinoma (FH-RCC) are rare but highly aggressive malignancies that are often refractory to the therapies used for the more common RCCs. FH-deficient RCC most often occurs in individuals with hereditary leiomyomatosis and renal cell carcinoma syndrome (HLRCC) and is characterized by loss of FH activity, whereas RMC is characterized by loss of the tumor suppressor SMARCB1. The combination of bevacizumab (B) with erlotinib (E) is a preferred first-line therapy for FH-deficient RCC and has shown efficacy in RMC although the contribution of each drug is unknown for either malignancy. We developed patient-derived xenograft (PDX) models of RMC and FH-deficient RCC which allowed us to test the activity of B and E alone or in combination in each model. 

Methods

Resected tumor tissue was immediately implanted into NSG (NOD-scid) mice. After tumor engraftment, animals were euthanized, and the tumor tissue was implanted into another set of NSG mice. PDX tissues were confirmed to be genetically identical to the original patient tissue by short tandem repeat (STR) analysis. Histology was confirmed to be identical between original patient tumor and PDX by a trained clinical pathologist. PDX models were then treated with either control (IP injection of saline once a week with oral gavage of PBS five days a week); dual therapy (25 mg/kg of E once a day five days on and two days off with 5 mg/kg of B once a week); 25 mg/kg of E once a day five days on and two days off; or 5 mg/kg of B once a week.  Mice were treated for 21 to 24 days prior to euthanasia and tumor size was measured by calipers twice a week. 

Results

We successfully generated PDX models for RMC and FH-deficient RCC. The RMC PDX model was generated from a 28-year-old male with sickle cell trait, who had previously received cisplatin plus paclitaxel followed by carboplatin plus paclitaxel. His lesion was characterized as ypT3apN1pMx and histology was consistent with RMC. The FH-deficient RCC model was generated from a 24-year-old female with HLRCC, who had been previously treated with E plus B. Her lesion was characterized as pT3apN1pM1 and histology consistent with FH-deficient RCC. Germline testing confirmed an established R233H pathogenic mutation in the FH gene. Our PDX models were confirmed to be genetically identical to the original patient tissue by STR analysis. Histology between original patient tumor and PDX were confirmed to be similar by a trained clinical pathologist.

Each of the PDX models was treated with control, E, B, or E + B. We observed similar statistically significant reduction (P < 0.05) in tumor growth with either E or E + B in our RMC PDX model, whereas the B monotherapy cohort behaved similarly to vehicle control. In the FH-deficient RCC model, each of the E and B monotherapies reduced tumor growth compared to vehicle control (P < 0.05). Furthermore, significant tumor growth reduction was noted when E and B are combined in the FH-deficient RCC PDX model compared to vehicle control (P < 0.05) but not compared to either E or B monotherapies. 

Conclusion

We were able to successfully create a PDX model of treatment-experienced RMC and FH-deficient RCC. Whereas E showed activity in both our RMC and FH-deficient RCC animal models, there was no activity of B in our RMC animal model either as monotherapy or in combination with E. Conversely, in the FH-deficient RCC model, B reduced tumor growth either as monotherapy or in combination with E. This suggests that angiogenesis targeted by B is a viable therapeutic target in FH-deficient RCC but not RMC, consistent with the established clinical observations that anti-angiogenic tyrosine kinase inhibitors are ineffective against RMC but show activity against FH-deficient RCC. Conversely, EGFR inhibition by E is effective in both preclinical models. We are in the process of mechanistically interrogating and validating these observations in additional RMC and FH-deficient RCC cell lines and animal models.