Introduction

Upper tract urothelial cancer (UTUC) is a relatively rare manifestation of urothelial cancer (UC) that tends to be more aggressive [1]. The standard treatment for high-grade disease involves surgery in the form of radical nephroureterectomy (NU). There is also level 1 evidence on the use of adjuvant chemotherapy based on the POUT trial, which has showed improvement in disease-free survival (DFS) [2]. Additionally, a meta-analysis revealed better evidence for adjuvant chemotherapy than there is for neoadjuvant chemotherapy (NAC) at this time [3]. Although the use of cisplatin-based NAC is considered nephrotoxic and there is currently no published level 1 evidence supporting the use of NAC in the management of high-grade UTUC (HG UTUC), a neoadjuvant approach may still be beneficial prior to definitive surgery [4]. Therefore, NAC followed by surgery is often used in clinical practice [5]. Nevertheless, much of our understanding of NAC benefit in UTUC is extrapolated from bladder-UC data and from retrospective or small prospective UTUC cohort studies.

Of interest, several molecular biomarkers, such as transcriptome-based molecular subtypes, have been evaluated for use in urothelial bladder cancer. At the highest level, bladder tumors can be classified among luminal and non-luminal subtypes and non-luminal tumors have been suggested to benefit most from NAC [6]. In the present study, we assessed molecular subtyping classifications of UTUC patients who underwent NU with or without NAC, by using Decipher Bladder^®, a clinical-grade, transcriptome-wide assay that has been validated in the peri-operative setting for bladder cancer [7]. The validated signatures include classes of Luminal, Infiltrated, Basal, Claudin-Low [8, 9] and Neuroendocrine-like (NE-like) [10, 11].

We conducted a small, single institutional, retrospective pilot study to obtain information regarding genomic profiling and molecular subtypes of four consecutive patients with a final diagnosis of UTUC who underwent NU and platinum-based chemotherapy [12]. A supplement to an existing IRB retrospective protocol was submitted to evaluate molecular subtypes in these patients. Four FFPE archival tissue samples were assayed through the commercially available Decipher Bladder^® test (Veracyte Inc, San Diego, CA). The tissue block with the highest grade and volume of disease was selected, and a region of interest for the assay was selected after pathological review of the hematoxylin & eosin tissue slide. The Foundation One^® next generation sequencing test was available for two of the four patients (Foundation Medicine, Cambridge MA). Further evaluation included clinical phenotypes of response, demographics, clinical staging, imaging results, final pathologic grade, and tumor stage. Overall survival and progression-free survival were evaluated though no formal statistical analyses were conducted given the small sample size of the study.

Case report

Four patient samples that were sent for transcriptome-wide analysis are reported herein. Table 1 summarizes clinical characteristics, and Figure 1 depicts selected marker expression from the Bladder Decipher^® Grid from these four cases.

Display full size

Figure 1. 

Bladder Decipher^® GRID. A. Pertains to clinical case 1; B. pertains to clinical case 2; C. pertains to clinical case 3; D. pertains to clinical case 4. Selected marker expression from the Bladder Decipher^® Grid which contains RNA expression values with the list of 48 genes grouped according to biological modules of luminal, basal, immune, epithelial-mesenchymal transition, stromal, and neuronal. The population rank refers to the percentage of tumor RNA profiles in the GRID Reference population. The bottom panel shows the genomic atlas or gene expression signatures that defines tumor biology

Clinical case 1

Case 1 was a 74-year-old male with coronary artery disease presented with a left upper tract mass with positive urine cytology showing high-grade UC (HGUC). He underwent four cycles of gemcitabine and cisplatin, with imaging findings showing stable left renal pelvis mass. A radical NU was performed, and the final pathology showed ypT3pNxMx HG UTUC with glandular and focal squamous differentiation (Figure 2). While stable findings were seen without evidence of distant disease, eventual death ensued from cardiovascular disease.

Display full size

Figure 2. 

Pathological findings on H&E for clinical case 1. A. Clinical case 1 pathological staging ypT3 high grade urothelial carcinoma with glandular and focal squamous differentiation involving the renal pelvis (40× H&E); B. high grade urothelial carcinoma extending into peri pelvic adipose tissue (100× H&E)

Clinical case 2

Case 2 was a 58-year-old male who underwent biopsy of a left kidney tumor, which revealed HG UTUC. He then underwent left NU, which showed pT3NxMx UC with squamous differentiation of 40%. A re-staging computed tomography (CT) scan showed multiple lung nodules, a liver mass, and retroperitoneal (RP) adenopathy. A subsequent CT-guided liver biopsy showed benign liver tissue, but a follow-up re-staging scan identified a new left hilar mass (Figure 3) and enlarging lung nodules. An endobronchial ultrasound-guided fine-needle aspiration biopsy of the hilar mass confirmed metastatic HGUC. He was started on four cycles of gemcitabine and carboplatin, with a bronchial stent placement required due to worsening lung lesions. The patient transitioned to comfort care given lack of response to treatment.

Display full size

Figure 3. 

Computed tomography (CT) scan findings of clinical case 2 with enlarging hilar mass at 30 mm × 37 mm size (arrows)

Clinical case 3

Case 3 involved an 80-year-old female who presented with a left renal pelvis mass and underwent left renal pelvis washing and biopsy, which showed HGUC. Further staging CT scan showed a small, non-specific 3 mm lung nodule. She then proceeded with left NU, which showed pT3pNxMx UC (Figure 4) with carcinoma in situ. She opted not to undergo adjuvant chemotherapy, but then subsequent re-staging CT showed prominent RP adenopathy, prompting the initiation of chemotherapy with gemcitabine and carboplatin for three cycles. However, due to worsening pain, another scan was performed which revealed an interval increase in RP and left periaortic adenopathy. Given these results, she was switched to pembrolizumab but only received two cycles before developing new liver metastases (Figure 5). This new development led to the patient’s eventual transition to hospice care.

Display full size

Figure 4. 

Pathologic findings on H&E for clinical case 3. A. Clinical case 3 pathological staging pT3 high grade papillary urothelial carcinoma of renal pelvis (40× H&E); B. high grade papillary urothelial carcinoma invades renal parenchyma (100× H&E)

Display full size

Figure 5. 

Computed tomography (CT) scan of clinical case 3 showing liver metastases (red arrows) and largest size measuring 38 mm × 32.4 mm

Discussion

UTUC makes up only a fraction of cases of UC but tends to be more aggressive in its course [13]. Systemic therapy follows closely the treatment used in bladder cancers [1]. Decipher Bladder^® assay is a genomic subtyping tool that has been shown to predict response to NAC in muscle-invasive bladder cancers (MIBC) [6]. In addition, the standard treatment for UTUC involves surgery with NU in early stages for high-grade disease though NAC followed by surgery is frequently used in clinical practice [1]. Patients in this series all received chemotherapy, but only case 1 underwent NAC followed by NU. While all three other cases underwent NU upfront, all eventually got diagnosed with metastatic disease and received further platinum-based chemotherapy for metastatic disease but all cases quickly progressed thereafter.

The four cases presented in our series (Table 1) had mixed stages and clinical course. Only case 1 remained locally advanced stage (stage III) and underwent NAC with residual ypT3 disease after NU. Unfortunately, this patient succumbed to cardiovascular morbidity, resulting in a non-cancer-related death. The remaining three cases, none of whom have prior NAC exposure, quickly progressed to metastatic disease, and showed poor overall response to systemic therapy.

All four cases had basal molecular markers, although two patients had additional mixed claudin-low molecular subtype with increased stromal markers on the Decipher Bladder^® GRID (Figure 1). This led to their classification among the stroma-rich subtype by the consensus subtyping model [14]. Of interest, both case 1 and 4 had basal and basal/claudin-low molecular subtype but their clinical course and phenotype differed. Case 1 manifested residual ypT3 HGUC after NAC but remained non-metastatic, whereas case 4 showed a sarcomatoid component on immunohistochemistry. These factors likely portended poor outcomes regardless of chemotherapy treatment.

The co-occurrence of non-luminal subtype and sarcomatoid variant histology has been reported in other series where basal MIBC have been enriched with both squamous and sarcomatoid features, which can be associated with metastatic disease or advanced stage at presentation [15, 16]. Furthermore, low expression of keratin 5/6 (KRT 5/6) and KRT14 is also consistent with sarcomatoid transition with loss of epithelial markers [17]. While several mutations were notable in case 2, the rapid progression of the disease did not allow for any further switch in treatment.

The genomic characteristics of patients who undergo NU and their association with response to systemic chemotherapy in UTUC have not been studied extensively. In contrast, molecular markers in bladder cancer have been evaluated and suggested to predict both pathological upstaging at radical cystectomy and response to NAC. Non-luminal tumors have shown higher rates of pathological upstaging in the cystectomy-only setting without NAC [9] and were suggested to benefit most from NAC in a separate multi-institutional MIBC study [18]. Notably, all four cases within the present UTUC profiling study had a non-luminal subtype, with only one out of four cases being treated with NAC. Although none of these cases showed a convincing response, the aggressive behavior of non-luminal tumors was validated in the present study as reflected by high pathologic tumor stage. Moreover, whereas previous efforts on transcriptomic subtyping revealed many luminal-type tumors among UTUC [19], the present study solely included non-luminal tumors.

In addition to observing a luminal subtype in UTUC, Robinson et al. [19] described UTUC having a T-cell depleted contexture. In our study, two out of four non-luminal cases harbored substantial signal for the claudin-low subtype, a subtype previously described to have increased levels of immune signatures and suggested to benefit from systemic neoadjuvant pembrolizumab [20]. Lastly, although activated FGFR3 signaling has been observed in UTUC [19], it was not evident in the present study when evaluating an FGFR3 activity signature (Table 1) [21]. Furthermore, Foundation DNA analysis results did not show any FGFR-related DNA alterations. As such, both the non-luminal transcriptomic subtypes and lack of FGFR-related DNA alterations in the present case-series may represent UTUC cases with more aggressive biology, reflected by their clinical course associated with poor outcomes.

The initial goal of this small pilot project was to determine feasibility and pave the way for a future prospective trial to predict the response to NAC in UTUC. However, since the initiation of this project, additional studies, including the POUT trial, have demonstrated the benefits of adjuvant chemotherapy, revealing an improvement in three-year DFS in favor of adjuvant chemotherapy [2]. In addition, the CheckMate 274 trial showed that adjuvant nivolumab improved DFS, although the subgroup of UTUC patients who were capped at 20% enrollment had seemingly suboptimal responses with subgroup analyses hazard ratios for disease recurrence or death of 1.23 and 1.56 for renal pelvis and ureter, respectively [22].

Understanding the biologic aspects may help prioritize UTUC patients to certain types of systemic therapy in a prospective manner. It is important to note that this case series has several limitations, including the small retrospective nature, the wide heterogeneity of the population with mixed stages, and inconsistent availability of genomic sequencing platform other than the molecular subtyping.

In summary, our study evaluated the feasibility of obtaining molecular subtyping information from UTUC specimens and provided potential feasibility and increased understanding of genomics in UTUC. While a prospective trial may assess whether molecular biomarkers could predict response to NAC, it is yet unclear whether these studies could facilitate future sparing of completion NU in those who achieve complete or partial response. In addition, the real-world application of next generation sequencing as a tool to be used for clinical judgement in UTUC patients requires further clarification in a larger cohort study. Furthermore, the future role of biomarkers including ctDNA, may help further refine the molecular landscape of the UTUC patient population.