Nephrectomy in the Era of Targeted Therapy: Takeaways from the CARMENA Trial

Published Date: September 2018

A 62-year-old man presents with a one-week history of hematuria. Ultrasound and computed tomography identify a 7-cm exophytic anterior left renal tumor, adenopathy, and two
small lung nodules. No bone or central nervous system lesions are detected. His Eastern Cooperative Oncology Group (ECOG) performance-status (PS) and Memorial Sloan-Kettering Cancer Center (MSKCC) scores are 1. The patient asks whether to undergo cytoreductive nephrectomy. What do you tell him? 

This case highlights a common treatment dilemma. Along the clinical spectrum, cytoreductive nephrectomy remains appropriate for an otherwise healthy 41-year-old with renal cell carcinoma and oligometastases in the lung—while surgery does not make sense for an 80-year-old with competing risks and a high metastatic tumor burden. But most of our patients fall in the middle of these extremes. In our current era of effective targeted therapies for metastatic kidney cancer, how can we best manage decisions about cytoreductive surgery?

Two decades ago, no study had shown a definitive benefit for cytoreductive nephrectomy in patients with metastatic renal cell carcinoma. That changed in 2001, when two studies by the European Organization for Research and Treatment of Cancer (EORTC) and the Southwest Oncology Group (SWOG) demonstrated that surgery followed by interferon alfa-based immunotherapy significantly improved overall survival (OS) compared with interferon-alfa treatment alone. Median OS was 11.1 months in the nephrectomy-interferon arm versus 8.1 months in the interferon alone arm of the SWOG 8949 trial.¹ Similar results were seen in EORTC 30947, and these findings established the role of cytoreductive nephrectomy in the standard initial management of metastatic kidney cancer.²

Those findings and that clinical decision made sense at the time, particularly given the lack of effective systemic therapies. But in 2005, a sea change began when the FDA approved sorafenib (Nexavar), an orally available multikinase inhibitor of tumor cell proliferation and angiogenesis, as the first targeted treatment for kidney cancer (FIGURE).³ In the pivotal trial, sorafenib therapy roughly doubled progression-free survival (PFS) compared with placebo in patients with metastatic cytokine-refractory clear-cell disease.⁴ Shortly thereafter, the FDA approved sunitinib (Sutent), a vascular endothelial growth factor receptor tyrosine kinase inhibitor, based on promising objective response data.⁵ These results were confirmed in a phase III front-line study in which sunitinib significantly improved PFS over interferon-alfa treatment (hazard ratio, O.54; 95% confidence interval [CI], 0.45 to 0.64; P<.001) and trended toward improved OS (HR, 0.82; 95% CI, 0.67 to 1.001; P=.051).⁶

More pivotal trials and approvals followed over subsequent years, raising questions about the role of cytoreductive nephrectomy in this new era of targeted systemic therapies.⁷ Observational studies and big-data analyses sought to clarify this role,^8,9 most notably a large retrospective study by the International Metastatic Renal Cell Carcinoma Database Consortium (IMDC).¹⁰ In this study of patients with synchronous metastases from renal cell carcinoma, initial cytoreductive nephrectomy led to a 40% reduction in the hazard of death compared with systemic therapy (usually sunitinib) without surgery.¹⁰ However, uncontrolled, retrospective analyses are limited by selection bias. Furthermore, surgery only appeared to benefit patients who had three or fewer IMDC prognostic factors,¹⁰ underscoring the need to refine surgical selection criteria.

OVERVIEW OF CARMENA

CARMENA was a 79-center randomized clinical trial of 450 adults with metastatic clear-cell renal cell carcinoma and an ECOG-PS of 0 or 1.11 This was designed as a non-inferiority trial, which made sense—if sunitinib alone was just as effective as nephrectomy followed by sunitinib, then postponing nephrectomy could spare patients the perioperative risks of surgery and enable them to immediately begin systemic disease control.

Patients in CARMENA were naïve to systemic therapy, deemed amenable to cytoreductive nephrectomy by their surgeon, and eligible for treatment with sunitinib.¹¹ They had Memorial Sloan Kettering Cancer Center (MSKCC) intermediate-risk (one or two prognostic factors) or poor-risk disease (three or more prognostic factors).¹³ Additionally, they either had no brain metastases or had received surgery or radiotherapy for brain metastases without corticosteroids and without progression for 6 weeks. The primary endpoint was OS.

The trial ran until 2017, for a median follow-up time of 50.9 months (95% confidence interval [CI], 44.0 to 56.9 months; range, 0.0 to 86.6 months).¹¹ At this time, an interim analysis of the intention-to-treat population, stratified by MSKCC risk score, produced a 0.89 hazard ratio for OS (95% CI, 0.71 to 1.10), upholding the study hypothesis of non-inferiority. Median OS times were 18.4 months in the sunitinib-only arm and 13.9 months in the surgery-sunitinib arm. Findings were similar in MSKCC risk-score subgroups (intermediate-risk: HR, 0.92; 95% CI, 0.6 to 1.24; poor-risk: HR, 0.86; 95% CI, 0.62 to 1.17). Based on these findings and the trial’s slow accrual, the steering committee decided to close the study early.¹¹

SUNITINIB WITHOUT SURGERY: TRULY NON-INFERIOR?

But was sunitinib alone truly non-inferior to cytoreductive nephrectomy followed by sunitinib? Non-inferiority comparisons usually focus on the upper limit of the 95% confidence interval. The CARMENA investigators determined that sunitinib without surgery would be clinically acceptable if the upper bound of the 95% confidence interval for the OS hazard ratio did not exceed 1.20.¹¹ In the intention-to-treat analysis, the upper limit of the 95% confidence interval for OS was 1.10, with a hazard ratio of 0.89 favoring sunitinib alone. Thus, sunitinib without surgery was found clinically acceptable in this patient population.

However, a per-protocol analysis told a different story. This analysis included only those patients who were actually treated as assigned (sunitinib alone or sunitinib with surgery). Here, the upper limit of the confidence interval crossed the 1.20 threshold. Median OS times were 20.5 and 18.3 months, respectively, with a hazard ratio of 0.98 (95% CI, 0.77 to 1.25).¹¹ Based on this result, it is harder to definitively conclude the non-inferiority of sunitinib without surgery when this patient population is treated as planned.

The wider confidence intervals of the per-protocol analysis reflect the fact that many CARMENA patients were not treated as planned. In the surgery-sunitinib arm, 7% of patients did not receive nephrectomy, and an additional 18% never received sunitinib.¹¹ In the sunitinib-only arm, 5% of patients did not receive sunitinib, and an additional 17% subsequently underwent cytoreductive nephrectomy, although this was allowed on study.¹¹

These results show that none of us can predict with 100% accuracy which patients are fit enough to undergo cytoreductive nephrectomy and recover enough to receive systemic therapy. Likewise, some patients who are treated with sunitinib first might have such a robust, near-complete response in their metastases that a consolidative nephrectomy makes sense. This is why intention-to-treat analyses are so useful—they include all the unexpected outcomes of patients, from those who drop off a study after becoming too sick to those with extraordinary responses. This mirrors real-world practice.

Despite slight discrepancies between analyses, the results of CARMENA are practice-changing. They reflect a more contemporary practice pattern and the largest prospective study thus far in this setting. These findings support the practice of deferring nephrectomy in order to initiate systemic therapy in patients who are relatively poor-risk, with metastatic tumor burdens of at least 4 cm, even if their performance status is good. These patients were well represented in the intention-to-treat analysis of CARMENA, which showed non-inferiority with sunitinib alone.

RATIONALES FOR DEFERRING NEPHRECTOMY

Although relatively few large, controlled studies have evaluated deferred nephrectomy, their results largely reinforce this approach for carefully selected patients. For example, in the randomized multicenter SURTIME trial (NCT01099423) of 99 patients with synchronous, predominantly intermediate-risk metastatic renal cell carcinoma, three cycles of sunitinib prior to cytoreductive nephrectomy did not improve progression-free rate (PFR) at 28 weeks compared with upfront nephrectomy followed by sunitinib.¹⁴
The SURTIME trial was underpowered due to slow accrual, but the intention-to-treat analysis of OS showed a signal in favor of deferred nephrectomy. While median OS was 32.4 months in patients who first received sunitinib versus 15.1 months in patients who first received nephrectomy (HR, 0.57; 95% CI, 0.34 to 0.95; P = .032), sample size precluded definitive conclusions.14 Despite the small size of this study, SURTIME suggest that deferred nephrectomy is reasonable for some intermediate-risk patients with advanced kidney cancer in our current era of targeted therapy.

Secondary results from the phase III CheckMate 214 trial (NCT02231749) point the same way. The presence of a primary tumor did not influence the results of CheckMate214, in which ipilimumab-nivolumab showed a significant survival advantage over sunitinib among intermediate and poor-risk patients with treatment-naïve, advanced or metastatic renal cell carcinoma.¹⁵

The results of CheckMate 214 led to an FDA approval of ipilimumab-nivolumab for this patient population.¹⁶ Since CARMENA began, the FDA has approved several other first-line treatments for metastatic kidney cancer, and the most recent management guidelines from the National Comprehensive Cancer Center (NCCN) give both pazopanib and sunitinib category 1 (preferred) recommendations for the first-line treatment of metastatic clear-cell disease in IMDC favorable-risk patients.17 For intermediate and poor-risk patients, a phase II randomized controlled trial showed a significant PFS advantage for cabozantinib versus sunitinib.¹⁸ We lack head-to-head comparisons of these agents with upfront nephrectomy, but it is reasonable to conclude that they might perform at least as well as sunitinib, given an appropriate patient selection for surgical deferment. As even more efficacious systemic treatments for kidney cancer emerge, we will need to further refine our selection criteria for initial surgery.

Finally, there are at least two biological rationales for prioritizing initial systemic therapy over cytoreductive surgery. The first is that the primary tumor can be a rich source of neoantigens,¹⁹ and treatments that stimulate even a modest or short-lived response in this tumor might prime the immune system for a stronger response to immuno-oncologic therapy. This is a key rationale for the perioperative design of the ECOG-ACRIN cooperative group’s PROSPER RCC study (NCT03055013), which is evaluating the efficacy of neoadjuvant and adjuvant nivolumab in patients with localized kidney cancer undergoing nephrectomy.²⁰

The second biological rationale is that delaying nephrectomy might avoid or slow metastasis. Studies of patients with breast cancer have identified a sharp peak in the risk of metastatic recurrence approximately 12 to 18 months after surgery.²¹ In preclinical studies of mice, T-cells were found to keep breast cancer tumor cells in check.²¹ Surgery and subsequent wound healing disrupted this balance, leading to distant metastasis.²¹ Confirmatory studies are needed; an intriguing hypothesis is that under certain yet-to-be-defined clinical circumstances, surgery might induce an inflammatory response that could potentially heighten the risk of metastases. Taken together with the results of CARMENA, these observations support a thoughtful and multidisciplinary approach to the timing of surgery in patients with metastatic kidney cancer.

BENEFITS OF UPFRONT NEPHRECTOMY

Conversely, several findings from CARMENA do support initial nephrectomy in certain patients with advanced kidney cancer.

First, patients undergoing cytoreductive nephrectomy had fewer related complications, particularly urinary tract infections and hematuria. Surgery also was fairly well tolerated; the rate of postoperative mortality at 1 month was only 2%, and although 39% of patients experienced postoperative morbidity, only 16% developed Clavien grade III or higher surgical complications.^11,22 These data suggest that cytoreduction in a well-selected, randomized setting is better tolerated than previously reported.

In contrast, patients in the sunitinib-only arm of CARMENA received an average of 2 months more sunitinib (8.5 vs. 6.7 months with nephrectomy-sunitinib; P=.04) and were more likely to develop grade 3-4 adverse events (43% vs. 38% for nephrectomy-sunitinib; P=.04). The most common grade 3-4 adverse events among patients who received sunitinib included asthenia, handfoot syndrome, anemia, and neutropenia, all of which are documented side effects of sunitinib. Furthermore, nine patients in the sunitinib-only arm developed grade 3 renal or urinary disorders, compared with only one patient in the nephrectomy-sunitinib arm (P=.05).

Thus, CARMENA showed that initial nephrectomy in carefully selected patients can potentially shorten overall treatment duration, yielding fewer complications with a non-inferior outcome. The primary risk of this approach appeared to be undertreatment. Almost 48% of patients in the sunitinib-only arm achieved disease control beyond 12 weeks, compared with only 37% of patients assigned to nephrectomy-sunitinib (P=.02). Importantly, nearly 23% of patients who underwent cytoreductive nephrectomy never recovered to an adequate degree to receive sunitinib.¹¹

STUDY LIMITATIONS

Several limitations of CARMENA merit discussion. Firstly, the risk criteria used for enrollment may have biased the study population toward poor-risk patients. The investigators used the MSKCC model, which is used perhaps most often to risk-stratify patients with advanced kidney cancer. The original MSKCC model included five independent predictors of poor survival, one of which was lack of nephrectomy.¹³ Because all CARMENA patients were considered candidates for nephrectomy, the 43% classified as MSKCC poor-risk had to have had at least three other negative prognostic factors. When the MSKCC model was developed, having just one poor prognostic factor reduced 1-year survival by almost 50%, and having three or more poor prognostic factors was nearly universally fatal at 1 year.¹³

Thus, CARMENA patients had additional negative prognostic indicators of poor outcomes even though they were amenable to nephrectomy.

This helps explain why median OS in CARMENA (18.4 months in the sunitinib-only group and 13.9 months in the nephrectomy-sunitinib group) was shorter than in other recent studies of metastatic kidney cancer. In CheckMate 214, median OS with sunitinib was 26.0 months although 38% of patients were classified as poor-risk based on IMDC criteria.¹⁵ In the randomized phase II CABOSUN trial, median OS among sunitinib-treated patients was 21.8 months; 19% of patients were classified as IMDC poor-risk.¹⁸
In addition to poor-risk features, patients in CARMENA also had substantial metastatic tumor burdens. The median size of primary tumors exceeded 8 cm in greatest dimension and median overall tumor burden was at least 14 cm.¹¹ Thus, metastatic disease comprised at least 40% of overall tumor burden for most patients. Although the typical aim of surgical cytoreduction is to remove the vast majority of tumor burden, this would not have been possible for many CARMENA patients.

CARMENA excluded patients with low metastatic burden at the investigator’s discretion;¹¹ patients with low metastatic burdens probably were not enrolled due to the prevailing belief that they would benefit from upfront removal of the primary tumor mass. Excluding these patients may have biased the results of this trial against surgery followed by sunitinib in those most likely to benefit from that strategy. Additionally, more patients in the nephrectomy-sunitinib arm had locally advanced stage T3 or T4 tumors (70%) than in the sunitinib-only arm (51%), which could have affected operative outcomes.

Finally, we note that CARMENA fell far behind in accrual. This may have been due to many unrelated factors. However, among many patients and physicians, surgery is reserved for those who are most motivated and subjectively believed to be most likely to benefit. This reflects not only a lack of truly coordinated multidisciplinary care for patients with advanced kidney cancer, but also, perhaps, a hope that surgery in some patients may be associated with disease regression or stabilization that might delay or, rarely, avoid the need for systemic therapy. It remains crucial to close this practice gap given our rapidly evolving treatment landscape.

SUMMARY AND TAKEAWAYS

The randomized phase III CARMENA trial compared sunitinib alone with nephrectomy followed by sunitinib in ECOG-PS 0 or 1 patients with intermediate or poor-risk metastatic clear-cell renal cell carcinoma. In the intention-to-treat analysis, sunitinib without surgery was found to be non-inferior to initial nephrectomy followed by sunitinib. Overall, the results supported the use of sunitinib alone in lieu of nephrectomy, especially in poor-risk patients and patients with a high metastatic tumor burden. However, the trial suffered from slow accrual and excluded patients with metastatic favorable-risk disease, which somewhat limits the generalizability of the findings.

Despite its shortcomings, CARMENA provides the best data we are ever likely to have on postponing nephrectomy in the era of targeted therapy. This trial was approximate twice the size of the original SWOG study, and its survival findings reflect what we would expect for poor-risk patients with relatively high-volume disease treated between 2009 and 2017. The per-protocol analysis did not support the non-inferiority hypothesis, but as clinicians, we manage patients by intention to treat, and the intention-to-treat analysis demonstrated that for most patients with metastatic renal cancer, starting with sunitinib was just as effective as upfront nephrectomy for this patient population.

The results of CARMENA highlight the importance of identifying the correct therapies and sequence on a case-by-case basis. Treatment remains multimodal, and tradeoffs and patient preferences must be considered. For patients with high risk stage III renal cell carcinoma, options include adjuvant clinical trials and, in selected cases, adjuvant sunitinib therapy, which demonstrated a 24% reduction in risk of disease recurrence in the recent S-TRAC trial (NCT00375674).²³ While not all patients will choose this approach, appropriate patients should at least discuss it with a medical oncologist. For patients with newly diagnosed metastatic disease, we should consider prognostic risk scoring and metastatic tumor burden. Systemic therapy increasingly is the first choice, but palliative surgery should be an option for well-selected, good-risk patients or those who are symptomatic.

We lack prospective trials comparing nephrectomy with systemic targeted therapies in patients with very limited metastatic disease. Although many of these patients and their surgeons probably prefer surgery to remove a large primary tumor, data from CARMENA offer at least some support for initial sunitinib, followed by consolidative surgery to remove the primary tumor source if metastatic lesions show excellent partial responses or prolonged stable disease. As systemic therapies continue to improve, this approach may apply to other patient populations. For example, patients with substantial metastatic tumor burdens might receive initial systemic therapy and proceed to consolidative surgery, depending on their response.

Despite substantial recent progress in treating metastatic kidney cancer, most patients cannot be cured. The results of CARMENA indicate that cytoreductive nephrectomy continues to make sense for select patients. Rather than discarding nephrectomy, we will need to continue to refine patient selection as new data on systemic treatments emerge.
Such complexities demand a multidisciplinary approach for all patients with synchronous metastatic renal cell carcinoma and for any patient at significant risk of developing metastatic disease. We recommend multidisciplinary tumor boards over ad hoc consultations. Formal tumor boards help physician-specialists think more systematically and overcome our inherent biases. In doing so, we can assess cases more objectively and ultimately improve patient outcomes.