Longitudinal Molecular Profiling of Circulating Tumor Cells in Metastatic Renal Cell Carcinoma

PURPOSE

Liquid biopsies in metastatic renal cell carcinoma (mRCC) provide a unique approach to understand the molecular basis of treatment response and resistance. This is particularly important in the context of immunotherapies, which target key immune-tumor interactions. Unlike metastatic tissue biopsies, serial real-time profiling of mRCC is feasible with our noninvasive circulating tumor cell (CTC) approach.

METHODS

We collected 457 longitudinal liquid biopsies from 104 patients with mRCC enrolled in one of two studies, either a prospective cohort or a phase II multicenter adaptive immunotherapy trial. Using a novel CTC capture and automated microscopy platform, we profiled CTC enumeration and expression of HLA I and programmed cell death-ligand 1 (PD-L1). Given their diametric immunological roles, we focused on the HLA I to PD-L1 ratio (HP ratio).

RESULTS

Patients with radiographic responses showed significantly lower CTC abundances throughout treatment. Furthermore, patients whose CTC enumeration trajectory was in the highest quartile (> 0.12 CTCs/mL annually) had shorter overall survival (median 17.0 months v 21.1 months, P < .001). Throughout treatment, the HP ratio decreased in patients receiving immunotherapy but not in patients receiving tyrosine kinase inhibitors. Patients with an HP ratio trajectory in the highest quartile (≥ 0.0012 annually) displayed significantly shorter overall survival (median 18.4 months v 21.2 months, P = .003).

CONCLUSION

In the first large longitudinal CTC study in mRCC to date to our knowledge, we identified the prognostic importance of CTC enumeration and the change over time of both CTC enumeration and the HP ratio. These insights into changes in both tumor burden and the molecular profile of tumor cells in response to different treatments provide potential biomarkers to predict and monitor response to immunotherapy in mRCC.

INTRODUCTION

Kidney cancer is a common cancer among both men and women with approximately 400,000 new cases and 175,000 deaths globally each year.¹ Renal cell carcinoma (RCC) encompasses a large heterogeneous group of cancers, of which clear cell RCC (ccRCC) is the most common, distinct in its immunogenicity and dependence on angiogenesis. Advances in our understanding of the pathogenesis of ccRCC have resulted in an expansion of treatment options for patients with advanced disease, which now includes immune checkpoint inhibitors and vascular endothelial growth factor (VEGF) targeting agents.²

CONTEXT

• Key Objective

• Although several US Food and Drug Administration–approved regimens exist for metastatic renal cell carcinoma (mRCC), our understanding of the molecular features associated with treatment response and resistance is limited. Using a circulating tumor cell (CTC) capture and automated immunofluorescence platform with sensitivity and specificity for mRCC CTCs, we provide the first insights into the molecular evolution of mRCC to treatment.

• Knowledge Generated

• By monitoring real-time shifts in CTC enumeration and HLA I and programmed cell death-ligand 1 protein expression, with a focus on the longitudinal trajectory of their ratio, we were able to capture new insights into renal cell carcinoma evolution to tyrosine kinase inhibitors and immune checkpoint blockers and demonstrate their prognostic importance.

• Relevance

• We demonstrate how liquid biomarkers can be used to monitor patients over time, rather than just as a single snapshot. This work forms the foundation for serial, blood-based molecular profiling in renal cell carcinoma and potentially has implications for immunotherapy in other tumor types.

Historically, patients with advanced disease received cytokine-based systemic therapy designed to provoke immune effectors. Only a subset of highly selected patients, however, experienced a response, but at the cost of substantial treatment-related toxicity.³ Subsequently, VEGF tyrosine kinase inhibitors (TKIs), designed to target angiogenesis, entered the clinic giving improved responses and progression-free survival for patients. Despite the initial efficacy, resistance proved to be inevitable and nearly all patients developed disease progression.⁴ The advent of immune checkpoint blockers (ICBs) targeting programmed cell death protein 1, programmed cell death-ligand 1 (PD-L1), and cytotoxic T-lymphocyte–associated protein-4 has dramatically changed the treatment paradigm for metastatic ccRCC. Immune checkpoint blockade in combination (ICB-ICB) or paired with a VEGF-TKI (ICB-VEGF) has now become the standard first-line therapy for most patients with metastatic RCC (mRCC). Combination therapy results in substantial responses in approximately 40%-60% of patients with 8%-10% of individuals experiencing a complete response^5-7; nevertheless, acquired resistance continues to pose a clinical challenge.⁸

Although several US Food and Drug Administration–approved regimens exist for mRCC, our understanding of the molecular features associated with treatment response and resistance is limited. The expression of PD-L1 is the most investigated biomarker for ICB therapy. Multiple studies have demonstrated that PD-L1 expression is a negative prognostic factor in advanced RCC^9,10; however, the benefit of ICB-based therapy is independent of PD-L1 status.^5-7,11 HLA class I (HLA I) expression is less studied in mRCC but is a prognostic factor in TKI-treated patients, but not those treated with immunotherapies.¹² Given the opposing effects of HLA I, an antigen presentation molecule responsible for stimulating an immune response, and PD-L1, an immune checkpoint protein responsible for inhibition of an immune response, with respect to immune evasion by tumors, we hypothesized that the interplay of these two key immune modulators would provide an informative metric to characterize the dynamic changes of the immune repertoire in real time. To our knowledge, this relationship has never been studied in the context of RCC or ICB.

Capturing the molecular heterogeneity and complexity of the immune system and metastatic cancer burden through in vitro and in vivo model systems is challenging; thus, human studies are required. However, this presents its own challenges as tissue biopsies for molecular profiling of mRCC are invasive, logistically challenging to scale, and carry the risk of procedural complications. Therefore, many of the landmark studies in mRCC^13-17 have relied at least in part on archival nephrectomy specimens, which may not accurately reflect changes in the current tumor state because of time or treatment-induced alterations. Even when metastatic biopsies are available, the serial sampling required to understand real-time tumor evolution is not feasible and almost never performed. Liquid biopsies that assess circulating tumor cells (CTCs) represent a noninvasive alternative.

Past CTC studies in mRCC focused primarily on enumeration and demonstrated reduced overall survival (OS) with increased CTC abundance.^18,19 Herein, we profiled 457 longitudinal blood samples from 104 patients from two cohorts, a prospective institutional cohort and the phase II OMNIVORE (OM) immunotherapy trial.²⁰ We used an institutional CTC capture and immunofluorescence platform²¹ to assess enumeration and HLA I and PD-L1 protein expression, with a focus on the longitudinal trajectory of their ratio. By monitoring the real-time shifts in enumeration and the molecular profile of CTCs, we were able to capture new insights into RCC evolution to ICB therapy and demonstrate their prognostic importance.

METHODS

Two cohorts of patients with RCC were assessed in parallel: a phase II multicenter, adaptive immunotherapy trial (OM)²⁰ and a prospective institutional cohort from the University of Wisconsin (UW). CTCs were captured using an institutional platform as previously described.^21-23 Neither of the cohorts included patients with co-occurring malignancies. CTC enumeration and fluorescence were evaluated using a novel automated quantitative microscopy approach. Normalized CTC abundance was calculated as the number of CTCs divided by the volume of blood analyzed. The interplay between HLA I and PD-L1 has been investigated in other tumor types,^24-29 but not in the context of RCC or ICB. Given their diametric roles within the immune system, we sought to contextualize the expression of each by analyzing the log-ratio between the two, hereafter referred to as the HP ratio. To ensure that measures of expression were not biased by CTC abundance, we compared HLA I, PD-L1, and log₁₀(HLA I/PD-L1) expressions relative to normalized CTC abundance within samples (Data Supplement, online only). The controlled treatment of patients in the OM cohort allowed us to determine the value of several CTC-specific variables in predicting OS. The rate of change in abundance and the HP ratio were estimated for each patient. Treatment response was assessed by local radiology in OM. In UW, response was initially determined by the treating physician and then independently confirmed by a study investigator.

Ethics Approval and Consent to Participate

Patients consented under the institutional review board–approved protocol from either Dana-Farber Cancer Institute (2017-1567) or the UW-Madison Carbone Cancer Center (2014-2014); all patients provided written informed consent before study enrollment, and all methodologies conformed to the standards of the Declaration of Helsinki.

Statistical Methods

Plotting and statistical tests were performed in R v4.1.0. Statistical tests performed were all two-sided. Cox regression was used for survival analyses. A Wilcoxon signed-rank test was used to compare continuous variables between groups and the paired pre/postinduction nivolumab samples. Boxplots represent quartiles, and whiskers are equal to 1.5× interquartile range. All analyses were also reproduced in only the patients with ccRCC, excluding the small number of non-ccRCC patients (Data Supplement). Additional details on these cohorts, such as sample collection, processing, and analysis, can be found in the Data Supplement.

RESULTS

CTC Enumeration Is Associated With Clinical Outcomes

We collected a total of 457 blood samples for CTC analysis from 104 patients. The OM cohort contained 305 samples from 60 patients while the UW cohort contained 152 samples from 44 patients. Baseline characteristics including timing/frequency of blood collection for the two cohorts are shown in the Data Supplement. CTC enumeration at baseline is prognostic for OS in patients with mRCC treated with TKI therapy.^18,19 We sought to evaluate the importance of CTC enumeration in the setting of the modern ICB-containing treatment paradigm and determine if its utility could extend to longitudinal monitoring. Normalized CTC abundance ranged from 0 to 718.4 CTCs/mL in the OM cohort and 0.7 to 91.3 CTCs/mL in the UW cohort. In total, only two samples from two patients in the OM cohort had no detectable CTCs, and both were at baseline.

Patients with confirmed progressive disease (PD) to induction nivolumab displayed a significantly higher abundance of CTCs while on treatment than at baseline (Wilcoxon P = .009, Fig 1A); these patients also displayed significantly higher on-treatment CTC abundances than patients with partial response (PR) who were on treatment (P = .035). Furthermore, patients with a PR exhibited generally lower baseline CTC abundances, never exceeding 28.3 CTCs/mL, whereas several patients with stable disease (SD; n = 7, 17%) or PD (n = 3, 7%) exceeded this value. When extended to the entire follow-up period, we observed a similar pattern, with PR patients never exceeding 89 CTCs/mL throughout the course of their treatment, while 12 patients with SD (7%) and 9 patients with PD (8%) surpassed this value. We next sought to confirm these findings in the UW cohort, where response specifically at the time of CTC collection was available. Given the smaller numbers within each response category, samples were compared between those associated with (complete response/PR/SD) and without (mixed response/PD/untreated) a clinical benefit. Significantly higher CTC abundance (Wilcoxon P = .047, Fig 1B) was observed in the UW collections associated with a negative clinical benefit at the time of collection when compared with collections associated with a positive clinical benefit, consistent with our findings from OM.

FIG 1.

CTCs per milliliter of blood, grouped by treatment response. (A) All OM collections, grouped by their collection timing (baseline in blue and on-treatment in red) and confirmed response during nivolumab induction (PR, SD, PD). Dashed lines represent the maximum CTCs per milliliter exhibited by partial responders at baseline (28.3 CTCs/mL, blue) and during treatment (89 CTCs/mL, red). (B) All UW collections, grouped by the clinical benefit associated with that collection. CTC, circulating tumor cell; OM, OMNIVORE; PD, progressive disease; PR, partial response; SD, stable disease; UW, University of Wisconsin.

Next, we sought to investigate trends in CTC enumeration over time, as we hypothesized that changes in CTC number could be prognostic for OS. We first performed patient-specific linear regressions, across all OM CTC collections, to quantify the slope of change in their CTC numbers over time. Patients whose annual rate of change in CTC abundance was in the top quartile for the cohort (> 0.12 CTCs/mL annually) had significantly shorter OS relative to patients with a rate of change in the bottom three quartiles (≤ 0.12 CTCs/mL; median 17.2 months v 21.1 months; HR = 5.9; 95% CI, 1.9 to 18.7; P < .001; Fig 2). These results confirm the importance of CTC enumeration and newly demonstrate that the change in CTC enumeration over time is an important prognostic variable in mRCC.

FIG 2.

Kaplan-Meier curve showing association between the rate of change in CTC abundance and patient OS in the OMNIVORE cohort. Patients are grouped by their annual rate of change in CTC abundance, determined from patient-specific linear models fit to sample CTC abundances throughout treatment, with patients in the red group belonging to the top quartile (> 0.12 CTCs/mL annually) versus bottom three quartiles in black group (≤ 0.12 CTCs/mL annually). CTC, circulating tumor cell; HR, hazard ratio; OS, overall survival.

HP Ratio Decreases Over Time With ICB

As no trend was observed between HLA I or PD-L1 expression and CTC abundance (Data Supplement), we next examined the log-ratio of HLA I to PD-L1 protein expression (HP ratio). We hypothesized that a ratio capturing both PD-L1 and HLA I expression would potentially be an informative metric to characterize the dynamic changes in the immune repertoire in real time. Specifically, we were interested in how this HP ratio changed over time (using the slope of a global linear model across patients) under the selective pressure of ICB, and if this carried any clinical importance.

The global linear model of the HP ratio over time revealed a significant decrease in the expression of HLA I relative to PD-L1 within the OM cohort (P < .001, slope = –0.2339, Fig 3A). Intriguingly, although this negative relationship was present in UW samples collected from patients receiving ICB therapy (slope = –0.10548, P = .0671, Fig 3B), it was absent in the samples derived from patients receiving TKI therapy (slope = 0.012, P = .821, Fig 3C). It is important to note that while the rate of change in the HP ratio may not appear large, the log-ratio is base 10, and thus, the difference of one unit indicates a 10-fold change.

FIG 3.

Linear regressions (red), weighted by circulating tumor cells per mL (point size), showing a decrease in measured log₁₀(HLA I/PD-L1) throughout treatment when on ICB therapy in both the (A) OM and (B) UW cohorts, but not (C) TKI therapy (UW only). Vertical dashed line in (A) indicates when the arm assignment is complete. ICB, immune checkpoint blocker; OM, OMNIVORE; PD-L1, programmed cell death-ligand 1; TKI, tyrosine kinase inhibitor; UW, University of Wisconsin.

Further analysis showed that this relationship was also present on an individual patient level; a significant decrease in the HP ratio was detected between paired preinduction and postinduction collections within the OM cohort (paired Wilcoxon P = .013, Fig 4), which were separated by a median of 113 days. This shift in the HP ratio over time, in response to ICB therapy, was also prognostic in OM. Patients with an HP ratio trajectory in the top quartile (≥ 0.0012 annually) displayed significantly shorter OS than patients whose trajectory was in the bottom three quartiles (< 0.0012 annually; median 18.35 months v 21.22 months; HR, 4.8; 95% CI, 1.5 to 15.3; P = .003, Fig 5).

FIG 4.

Boxplots showing the distribution of the measured log₁₀(HLA I/PD-L1) before nivolumab induction and again after a mean of 113 days on treatment (28-day cycles) for 30 patients with paired collections in the OMNIVORE cohort. Paired measures are connected by dotted lines. PD-L1, programmed cell death-ligand 1.

FIG 5.

Kaplan-Meier curve showing association between the rate of change in log₁₀(HLA I/PD-L1) and patient OS in the OMNIVORE cohort. Patients are grouped by annual rate of change in log₁₀(HLA I/PD-L1), determined from patient-specific linear models fit to sample abundances across treatment, with patients in the red group belonging to the top quartile. HR, hazard ratio; OS, overall survival; PD-L1, programmed cell death-ligand 1.

Radiation Therapy

Radiation therapy has been hypothesized to potentially synergize with ICB^30,31 and lead to systemic responses from local radiation treatment.³² In an exploratory analysis, we sought to examine how CTC abundance and the HP ratio fluctuated after palliative radiotherapy. Although radiotherapy was not allowed in the OM protocol, 12 patients from the UW cohort did receive palliative radiation during their treatment course. For these patients, we calculated the pairwise differences in CTC abundance and HP ratio between an individual's sample collected directly before radiotherapy and all other collections. We then plotted these differences relative to their time from receiving radiotherapy (Fig 6). Interestingly, when we fit polynomial spline curves to the abundance and ratio, we found that although patients initially displayed a trend of increasing CTC abundance and HP ratio before radiotherapy (a negative prognostic sign), this trend reversed for approximately 6 months upon receiving radiotherapy with both the abundance and ratio decreasing, both of which were found to associate with a positive prognosis above. Ultimately, however, this change did not last and both trends continued to increase, which is not surprising as palliative radiation is typically only given to patients with a poor response. Nevertheless, our data suggest that radiotherapy is having an impact on the expression of immunomodulatory molecules on mRCC tumor cells, through which a positive response, albeit short-term and variable, is elicited. Immunological changes in response to radiation have been previously observed,^31,33 which warrants further study. The fact that we do observe transient biomarker changes associated with better prognoses may suggest the potential for combining radiotherapy and immunotherapy, a concept that is being tested in prospective clinical trials.

FIG 6.

Modeled influence of radiotherapy on CTC measures for 12 patients in University of Wisconsin cohort. Polynomial spline models fit to the difference between measures from a patient's CTC collection taken directly before radiotherapy (time = 0) and CTC collections taken throughout treatment (time = years between collections). Differences are calculated for (A) log₁₀(CTCs/mL) and (B) log₁₀(HLA I/PD-L1). CTC, circulating tumor cell; PD-L1, programmed cell death-ligand 1.

DISCUSSION

Herein, we present the largest CTC study to our knowledge in mRCC, profiling 457 blood samples from 104 patients from a prospective institutional cohort and the phase II OM immunotherapy trial. Using an advanced microfluidic platform that has shown high sensitivity and specificity for ccRCC CTC capture,²¹ we obtained longitudinal collections which allowed us to monitor changes in CTC enumeration and the log-HP ratio in response to treatment. We identified that the change in CTC abundance over time and the change in the HP ratio were both prognostic. These results provide insights into potential biomarkers for predicting patient response to ICB and assisting in prognostication for mRCC.

CTC enumeration has been the focus of prior CTC studies in mRCC and is prognostic for OS in patients with mRCC treated with TKI therapy in one study,¹⁸ whereas a second study showed a nonsignificant trend for cancer-specific survival, also in TKI-treated patients.¹⁹ In the first examination in the immunotherapy era, we again found that CTC enumeration is prognostic. Specifically, we demonstrated that the direction that CTC enumeration changes during immunotherapy treatment is strongly associated with OS and may complement imaging in monitoring disease and predicting the trajectory of the disease.

In addition, we used an automated immunofluorescence platform to examine PD-L1 and HLA I protein levels in CTCs. To our knowledge, this is the first such study in mRCC. Downregulation of HLA I is a well-known mechanism of immune evasion in cancer, as is PD-L1 upregulation; however, neither has been individually demonstrated as a clear predictor of ICB response in mRCC.^5-7,11,12 As the effects of HLA I and PD-L1 are opposite on immune evasion by tumors, we sought to contextualize their coexpression by examining the log-HP ratio. We found that this ratio decreased over time in patients treated with ICB in both the UW and OM cohorts, but not those treated with single-agent TKI. Tumor cells with high HLA I and low PD-L1 are likely preferentially cleared by immunotherapy, and thus over time, the population average would reflect only the survivors, leading to a decreasing ratio over time. This is further supported by our finding that an increasing HP ratio over time portends worse outcomes. If a patient's HP ratio does not drop with ICB, it may reflect the poor efficacy of ICB, which is not affecting the HP ratio of the overall population of tumor cells. The importance of this relationship on tumor stage, prognosis, and T-cell infiltration has been described in non–small-cell lung carcinoma,^24,25 intrahepatic cholangiocarcinoma,²⁶ hepatocellular carcinoma,²⁷ head and neck squamous cell carcinoma,²⁸ and bladder cancer.²⁹ However, none of these prior studies in other cancers were in the context of RCC or ICB. Our data suggest that the changes in the HP ratio on mRCC CTCs complement enumeration and reflect shifts in the population of tumor cells toward resistant subclones in response to immunotherapy.

Although our study was large and diverse, there were limitations within each cohort. Our institutional cohort was heterogeneous in treatments, as all institutional cohorts tend to be. As our study only had a small number of patients treated with ICB plus TKI, as well as those with sarcomatoid/rhabdoid features, future studies focused on these populations will be required to extrapolate our results. Nevertheless, this study did allow for a comparison of the HP ratio between ICB and TKI therapy and provided extensive longitudinal data. The OM ICB trial had much more uniform inclusion criteria and treatment, allowing for survival analyses to be performed. However, it was designed as a therapeutic trial, and patients who progressed or had treatment-limiting toxicity went off trial, which meant that liquid biopsies became sparser with longer follow-up. Although each cohort has caveats, the weaknesses of each were balanced by the strengths of the other, allowing for a more complete picture of CTCs in RCC.

In conclusion, we demonstrate the potential clinical utility of monitoring both CTC enumeration and the HP ratio on CTCs. In particular, our study leverages the serial nature of liquid biopsies to demonstrate how biomarkers can be used to monitor patients over time, rather than just as a single snapshot. The trajectory of both CTC enumeration and HLA I/PD-L1 could provide guidance on a range of important treatment decisions such as when to initiate or change systemic therapy, especially in situations where imaging is ambiguous. This work forms the foundation for blood-based molecular profiling in RCC. Further validation of the clinical utility of these biomarkers requires testing in multiple large prospective clinical trials. Therefore, this assay is being implemented as a correlative end point in a large multi-institutional prospective observational mRCC trial, the recently approved randomized NRG SAMURAI trial³⁴ (ICB with or without radiation), and the randomized Alliance RadiCal trial (TKI with or without Radium-223; ClinicalTrials.gov identifier: NCT04071223).

DATA SHARING STATEMENT

A data sharing statement provided by the authors is available with this article at DOI https://doi.org/10.1200/JCO.22.00219.

AUTHOR CONTRIBUTIONS

Conception and design: Matthew Bootsma, Rana R. McKay, Hamid Emamekhoo, Jamie Sperger, Sabina Signoretti, Toni K. Choueiri, Joshua M. Lang, Shuang G. Zhao

Financial support: Joshua M. Lang, Shuang G. Zhao

Administrative support: Joshua M. Lang, Shuang G. Zhao

Provision of study materials or patients: Hamid Emamekhoo, Sabina Signoretti, Christos E. Kyriakopoulos, John Floberg, Toni K. Choueiri, Joshua M. Lang, Shuang G. Zhao

Collection and assembly of data: Matthew Bootsma, Rana R. McKay, Hamid Emamekhoo, Rory M. Bade, Jennifer L. Schehr, Matthew C. Mannino, Anupama Singh, Serena K. Wolfe, Zachery D. Schultz, Jamie Sperger, Wanling Xie, Christos E. Kyriakopoulos, John Floberg, Nan Sethakorn, Toni K. Choueiri, Joshua M. Lang, Shuang G. Zhao

Data analysis and interpretation: Matthew Bootsma, Rana R. McKay, Hamid Emamekhoo, Zachery D. Schultz, Christos E. Kyriakopoulos, David Kosoff, Edwin J. Abel, Kyle T. Helzer, Nicholas Rydzewski, Hamza Bakhtiar, Yue Shi, Grace Blitzer, Michael Bassetti, Menggang Yu, Marina Sharifi, Paul M. Harari, Joshua M. Lang, Shuang G. Zhao

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Longitudinal Molecular Profiling of Circulating Tumor Cells in Metastatic Renal Cell Carcinoma

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).