Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2019 Jun 14;2(6):e195806. doi: 10.1001/jamanetworkopen.2019.5806

Comparative Survival Associated With Use of Targeted vs Nontargeted Therapy in Medicare Patients With Metastatic Renal Cell Carcinoma

Pengxiang Li 1,2, Jordan Jahnke 1, Amy R Pettit 3, Yu-Ning Wong 4,5, Jalpa A Doshi 1,2,3,✉
PMCID: PMC6575152  PMID: 31199450

Key Points

Question

Is the use of targeted therapy associated with survival advantages compared with older treatments in a real-world population of patients with metastatic renal cell carcinoma who are enrolled in Medicare?

Findings

This cohort study of 1015 patients used instrumental variable analysis to compare survival associated with targeted vs nontargeted therapy while controlling for measured and unmeasured confounders. Estimated overall survival improvements with targeted therapy were statistically significant—8% at 1 year, 7% at 2 years, and 5% at 3 years—despite the fact that the targeted therapy group exhibited more medical complexity.

Meaning

Targeted therapies may have enabled a broader range of patients with metastatic renal cell carcinoma to be treated and were associated with modest survival advantages over nontargeted therapies.

This cohort study compares overall and disease-specific survival associated with targeted vs nontargeted therapy in elderly patients with metastatic renal cell carcinoma who are enrolled in Medicare.

Abstract

Importance

Targeted therapies for advanced renal cell carcinoma (RCC) have shown increased tolerability and survival advantages over older treatments in clinical trials, but understanding of real-world survival improvements is still emerging.

Objective

To compare overall and RCC-specific survival associated with use of targeted vs nontargeted therapy for metastatic RCC.

Design, Setting, and Participants

This retrospective cohort study used Surveillance, Epidemiology, and End Results–Medicare data from 2000 to 2013 to examine patients with stage IV (distant) clear cell RCC at the time of diagnosis who received any targeted or nontargeted therapy. A 2-stage residual inclusion model was fitted to estimate the survival advantages of targeted treatments using an instrumental variable approach to account for both measured and unmeasured group differences. Data analyses were conducted from July 24, 2017, to April 4, 2019.

Exposures

Targeted therapy (study group) or nontargeted therapy (control group).

Main Outcomes and Measures

Overall survival and RCC-specific survival, defined as the interval between the date of first drug treatment and date of death or end of the observation period.

Results

The final sample included 1015 patients (mean [SD] age, 71.2 [8.1] years; 392 [39%] women); 374 (37%) received nontargeted therapy and 641 (63%) received targeted therapy. The targeted therapy group had a greater percentage of disabled patients (ie, those <65 years old who were eligible for Medicare because of disability) and older patients (ie, those ≥75 years old) and higher comorbidity index and disability scores compared with the nontargeted therapy group. Unadjusted Kaplan-Meier survival curves showed higher overall survival for targeted vs nontargeted therapy (log-rank test, χ21 = 5.79; P = .02); median survival was not statistically significantly different (8.7 months [95% CI, 7.3-10.2 months] vs 7.2 months [95% CI, 5.8-8.8 months]; P = .14). According to the instrumental variable analysis, the median overall survival advantage was 3.0 months (95% CI, 0.7-5.3 months), and overall survival improvements associated with targeted therapy vs nontargeted therapy were statistically significant: 8% at 1 year (44% [95% CI, 39%-50%] vs 36% [95% CI, 30%-42%]; P = .01), 7% at 2 years (25% [95% CI, 20%-30%] vs 18% [95% CI, 13%-23%]; P = .009), and 5% at 3 years (15% [95% CI, 11%-19%] vs 10% [95% CI, 6%-13%]; P = .01). Receipt of targeted therapy was associated with a lower hazard of death compared with nontargeted therapy (overall survival hazard ratio, 0.78 [95% CI, 0.65-0.94]; RCC-specific survival hazard ratio, 0.77 [95% CI, 0.62-0.96]).

Conclusions and Relevance

Targeted therapies were associated with modest survival advantages despite a treatment group with more medical complexity, likely reflecting appropriateness for an expanded population of patients. As advances in cancer treatment continue, rigorous methods that account for unobserved confounders will be needed to evaluate their real-world impact on outcomes.

Introduction

Targeted therapies have become first-line pharmacological treatments for advanced renal cell carcinoma (RCC),1 and the US Food and Drug Administration has approved 12 new agents for advanced RCC since December 2005. These treatments have been shown to extend survival in clinical trials2,3,4,5,6; for example, median overall survival has been shown to be 24 to 30 months among patients receiving targeted therapies7 compared with less than 1 year in the era before targeted therapy.8 Initial progress with targeted therapies has since been surpassed by newer immunotherapy agents.9 These advances have been welcome in light of the fact that older treatments for advanced RCC (cytokines such as interleukin 2 and recombinant interferon alfa) were characterized by low response rates, minimal clinical benefit, and considerable toxicity.10

Although increased tolerability itself is a worthwhile advance, particularly when it enables treatment for elderly patients or those with more medical complexity who may not have been eligible for older therapies, the degree to which targeted therapies have offered real-world survival advantages is not entirely clear. Most of the available studies examining real-world data have examined targeted treatments alone, sequentially, or in relation to one another.11,12,13 The limited research comparing outcomes for targeted therapies to those achieved with older treatments has shown advantages for targeted therapies, but interpretation of the evidence is complicated by unmeasured confounders and selection bias14 that make it difficult to discern the degree to which observed survival advantages are associated with the treatments themselves, rather than additional unobserved differences between groups (eg, severity of illness not captured by available data).

To address this issue, we examined real-world outcomes for patients with advanced RCC using instrumental variable analysis, a statistical approach that serves as an alternative to random assignment to treatment and addresses confounding due to both measured and unmeasured variables, thereby allowing for causal inferences.15 Because one-half of RCC cases in the United States are diagnosed in individuals aged 65 years or older16 and because older patients are underrepresented in clinical trials,17,18 we focused our analysis on the population enrolled in Medicare. Furthermore, because up to one-third of patients present with metastatic disease,19,20 we focused on patients whose initial diagnosis was metastatic RCC (mRCC).

Methods

Study Design

The institutional review board at the University of Pennsylvania approved the study protocol with a waiver of written informed consent, and data analyses were conducted from July 24, 2017, to April 4, 2019. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.21 This retrospective study examined differences in survival outcomes associated with the use of targeted vs nontargeted therapy in mRCC, using an instrumental variable approach to address bias due to unobserved factors.22 Although traditional risk-adjusted survival models address the difference between treatment and control groups attributable to observed factors, instrumental variable analysis goes a step beyond to balance all potentially relevant variables, measured or unmeasured, between treatment groups.

Data Source

The primary data source for the study was the 2000 to 2011 Surveillance, Epidemiology, and End Results Program (SEER) cancer registry,23 linked with Medicare enrollment data and Medicare claims data (Parts A, B, and D) from 2000 to 2013. The SEER registries collect demographic, diagnostic, clinical, and cause-of-death information for patients with cancer and include representative data from 20 sites, covering approximately 28% of the US population. Medicare claims provide information for covered health care services for all patients enrolled in fee-for-service Medicare. The SEER-Medicare files used in our study included all patients included in the SEER registry from 2000 through 2011 who had fee-for-service Medicare-linked data available, or more than 90% of persons 65 years and older included in the SEER files.24

Sample Selection

The initial sample included patients with RCC as the primary cancer diagnosis (International Classification of Diseases for Oncology, Third Edition codes 8310-8319) between 2000 and 2011. Patients without complete diagnosis date information and those for whom a diagnosis was made at autopsy were excluded (eFigure 1 in the Supplement). For inclusion in the final study sample, patients were required to have stage IV (distant) clear cell RCC (International Classification of Diseases for Oncology, Third Edition code 8310 or 8312) at the time of initial diagnosis and to have received targeted or nontargeted therapy for mRCC after their diagnosis date. They were also required to have fee-for-service Medicare coverage (prior to 2006, before Medicare Part D was in effect) and/or fee-for-service Medicare and Medicare Part D prescription drug plan coverage (2006 and after) from the time of diagnosis to the time of their first drug treatment claim. The study group included patients with mRCC receiving a targeted therapy as the first drug treatment, whereas the control group included patients receiving a nontargeted therapy (interferon and aldesleukin) as the first drug treatment. A list of therapies is provided in Table 1, with further details on targeted therapies available in eTable 1 in the Supplement.

Table 1. Patient Characteristics.

Characteristic Treatment Group, No. (%) P Value
Nontargeted Therapy (n = 374) Targeted Therapy (n = 641)
Age, ya
18-64 35 (9) 98 (15) .01
65-74 223 (60) 317 (49)
≥75 116 (31) 226 (35)
Women 128 (34) 264 (41) .03
Race
White 333 (89) 534 (83) .002
Black or otherb 41 (11) 107 (17)
Married 257 (69) 340 (53) <.001
Census region
Northeast 74 (20) 96 (15) <.001
Midwest 50 (13) 71 (11)
South 137 (37) 181 (28)
West 113 (30) 293 (46)
Urban residence 313 (84) 525 (82) .47
County-level characteristics
Per capita income, mean (SD), $1000s 31.4 (9.9) 39.9 (10.7) <.001
Unemployment rate per 1000 residents, mean (SD) 27.0 (10.5) 39.4 (16.1) <.001
Physicians per 1000 residents, mean (SD), No.c 2.3 (1.3) 2.3 (1.3) .95
Renal cell carcinoma grade
1 or 2d 60 (16) 103 (16) .71
3 71 (19) 118 (18)
4 25 (7) 59 (9)
Not determined 218 (58) 361 (56)
Selected metastatic sitese
Lung 233 (62) 360 (56) .06
Liver 65 (17) 114 (18) .87
Bone 128 (34) 279 (44) .004
Brain 40 (11) 87 (14) .18
Received radiation therapy 99 (26) 182 (28) .51
Underwent nephrectomy 149 (40) 225 (35) .13
National Cancer Institute–Charlson comorbidity score, mean (SD) 0.9 (1.1) 1.4 (1.5) <.001
Claims-based disability score, mean (SD)f 0.05 (0.06) 0.08 (0.10) <.001
Fatigue 106 (28) 248 (39) .001
Time to first drug treatment, mean (SD), mo 4.1 (5.1) 5.8 (10.0) .002
Year of diagnosis
2004 and before 344 (92) 19 (3) <.001
2005 and after 30 (8) 622 (97)
First agent receivedg
Interferon alfa-2b 208 (56)
Aldesleukin 116 (31)
Interferon alfa-2a 50 (13)
Sunitinib 345 (54)
Temsirolimus 101 (16)
Sorafenib 98 (15)
Bevacizumab 51 (8)
Pazopanib 34 (5)
Everolimus or axitinib 12 (2)
Open in a new tab
a

Age at diagnosis. Some patients were eligible for Medicare because of disability rather than age.

b

Other race category includes American Indian/Alaska Native and Asian/Pacific Islander. Categories are combined because of a cell size smaller than 11, per the terms of the data use agreement.

c

Includes both primary care physicians and specialists.

d

Categories were grouped together because of small cell size, per the terms of the data use agreement.

e

Categories are not mutually exclusive; patients may have had metastasis in more than 1 site.

f

Claims-based prediction for disability status, a proxy measure for the Eastern Cooperative Oncology Group performance status scale.

g

All agents approved by the US Food and Drug Administration for metastatic renal cell carcinoma at the time of the study were eligible for inclusion. Nontargeted agents included interferon alfa-2a, interferon alfa-2b, interferon alfa-n3, aldesleukin, and peginterferon alfa-2b. Targeted therapies (in order of Food and Drug Administration approval) included sorafenib, sunitinib, temsirolimus, everolimus, bevacizumab, pazopanib, and axitinib.

Outcomes

Primary outcomes were overall survival and RCC-specific survival. In the main analysis, survival times were defined as the interval from the date of the first drug treatment until the date of death or December 31, 2013 (ie, the end of the observation period).

Statistical Analysis

We used t tests and χ2 tests to compare continuous and categorical patient characteristics, respectively, among patients receiving targeted and nontargeted therapy. We created Kaplan-Meier survival curves for overall and RCC-specific survival and compared the study and control groups using log-rank tests.

We used SEER data to code sociodemographic characteristics (age, sex, race, marital status, census region, and urban vs rural residence), county-level characteristics (income, unemployment rate, and number of physicians per 1000 residents), and clinical characteristics (RCC grade, selected metastatic sites [lung, liver, bone, and brain], radiation status [yes or no], and nephrectomy status [yes or no]) for each patient. We captured 3 additional variables: the National Cancer Institute (NCI)–adapted Charlson comorbidity score,25,26 a claims-based marker of disability based on a measure of functional status (specifically developed for cancer treatment and outcomes studies),27,28 and an indicator of fatigue29 (International Classification of Diseases, Ninth Revision, Clinical Modification codes 300.5, 780.7, 799.3, and 780.72), according to Medicare claims in the 6 months before the first drug treatment. In addition, we included a covariate capturing time from mRCC diagnosis to first drug treatment.

Instrumental Variable Analysis

Valid instrumental variables must meet 2 criteria: (1) they must be strongly associated with the treatment being examined (ie, targeted therapy or nontargeted therapy) and (2) they cannot be associated with the relevant outcome (ie, survival), except through their association with the treatment. These requirements are known as the instrumental variable strength assumption and exogeneity assumption, respectively. An instrumental variable meeting these criteria can be used to generate pseudorandomization and an unbiased estimate of treatment effect.

Because targeted treatments were approved beginning in December 2005, we chose the year of mRCC diagnosis as the instrumental variable and confirmed that it met both validity requirements (eMethods, eFigure 2, eTable 2, eTable 3, and eTable 4 in the Supplement). We used a Weibull 2-stage residual inclusion estimation framework for the instrumental variable analysis.30,31,32 We also calculated the model-derived median survival and probability of 1-, 2-, and 3- year survival for patients receiving targeted and nontargeted therapy. We conducted subgroup analyses to examine targeted therapy survival benefits among subgroups of patients by age (<65 years or ≥65 years) and nephrectomy status (yes or no).

To test the robustness of our findings, we also conducted sensitivity analyses. First, we modified our selection criteria in 3 separate analyses: (1) excluding patients receiving treatment in 2005 and 2006, when targeted therapy had just become available; (2) excluding patients who also had other cancers; and (3) excluding patients with the brain as the metastatic site. Next, we modified our outcome measure to calculate survival from the date of mRCC diagnosis instead of from the date of the first drug treatment, in recognition of the fact that watch-and-wait periods may vary. Finally, we modified our statistical approach and used Cox survival models in the instrumental variable analysis, as an alternative to the Weibull model.

All reported P values are 2-sided and are considered statistically significant if they are less than .05. Data analyses were conducted using SAS statistical software version 9.4 (SAS Institute) and Stata statistical software version 15.0 (StataCorp).

Results

The final sample included 1015 patients with mRCC, with a mean (SD) age of 71.2 (8.1) years; 392 (39%) of them were women (data not shown). Most (641 patients [63%]) received targeted therapy and 374 patients (37%) received nontargeted therapy (Table 1). Compared with the nontargeted therapy group, the targeted therapy group had greater percentages of patients in the youngest age group (<65 years [ie, eligible for Medicare because of disability]) and in the oldest age group (≥75 years), a slightly smaller percentage of white patients, and a higher percentage of patients who were not married. Differences in region of residence were also observed. In addition, the targeted therapy group had a significantly higher county-level mean per capita income but also a significantly higher county-level unemployment rate.

Clinical differences were also observed between the 2 groups. Metastatic disease to the lung was most common in both groups, but a greater percentage of patients receiving targeted therapy had metastatic disease to the bone. The targeted therapy group had a higher mean National Cancer Institute–Charlson comorbidity score and a higher mean claims-based disability score, and a higher percentage had evidence of fatigue. Although the groups did not have statistically significant differences in the rate of nonpharmacologic treatments (ie, nephrectomy and radiation), on average, the nontargeted therapy users received their first drug treatment more quickly after their initial mRCC diagnosis date (4.1 months vs 5.8 months; difference, −1.7 months; 95% CI, −2.8 to −0.6 months; P = .002).

The median (interquartile range) follow-up period after the first treatment was 8 (3-20) months (data not shown). During follow-up, 369 patients who received nontargeted therapy (99%) and 562 patients who received targeted therapy (88%) died from any cause. The RCC-specific mortality rate was 76% (283 patients) for the nontargeted therapy group and 62% (395 patients) for the targeted therapy group (data not shown). As shown in Figure 1, unadjusted Kaplan-Meier survival curves indicated that patients receiving targeted therapy had higher overall and RCC-specific survival rates than did patients receiving nontargeted therapy (log-rank test, χ21 = 5.79 and P = .02 for overall survival; χ21 = 10.86 and P = .001 for RCC-specific survival). As shown in Table 2, median unadjusted overall survival was 8.7 months (95% CI, 7.3-10.2 months) vs 7.2 months (95% CI, 5.8-8.8 months) (P = .14), and RCC-specific survival was 13.2 months (95% CI, 11.3-15.1 months) vs 10.5 months (95% CI, 8.1-13.4 months) for the targeted therapy vs the nontargeted therapy group, respectively. When examining those in the top tenth percentile of the survival curve, we found an overall survival difference of 11.5 months; 10% of patients who received nontargeted therapy survived longer than 36.1 months, whereas 10% of patients receiving targeted therapy survived longer than 47.6 months (data not shown).

Figure 1. Unadjusted Kaplan-Meier Estimates of Overall and Renal Cell Carcinoma–Specific Survival for Patients Receiving Targeted vs Nontargeted Therapy.

Figure 1.

Open in a new tab

Log-rank test comparing curves, P = .02 for overall survival and P = .001 for renal cell carcinoma–specific survival. n* indicates a cell size of 10 or less, which is not reported per the Centers for Medicare & Medicaid Services data use agreement; NTT, nontargeted therapy; and TT, targeted therapy.

Table 2. Median Survival Duration After Receipt of Nontargeted or Targeted Therapya.

Survival Nontargeted Therapy Targeted Therapy Difference (95% CI) P Value
Overall survival
No. of deaths/No. of patients (%) 369/374 (99) 562/641 (88)
Survival duration, median (95% CI), mo
Unadjusted nonparametric estimate 7.2 (5.8-8.8) 8.7 (7.3-10.2) 1.6 (−0.5 to 2.1) .14
Instrumental variable analysis estimate 8.8 (7.3-10.4) 11.8 (10.2-13.5) 3.0 (0.7-5.3) .01
Renal cell carcinoma–specific survival
No. of deaths/No. of patients (%) 283/374 (76) 395/641 (62)
Survival duration, median (95% CI), mo
Unadjusted nonparametric estimate 10.5 (8.1-13.4) 13.2 (11.3-15.1) 2.7 (−0.4 to 5.8) .09
Instrumental variable analysis estimate 13.3 (10.3-16.4) 18.2 (14.9-21.5) 4.9 (0.4-9.4) .02
Open in a new tab
a

Unadjusted nonparametric estimates are based on Kaplan-Meier curves. Instrumental variable analysis estimates controlled for all covariates listed in Table 1 (except specific agent received) in the first and second stages of the 2-stage residual inclusion approach. Median survival was based on the Weibull survival model with 2-stage residual inclusion. Differences in median survival between the targeted and nontargeted therapy groups were tested using a z statistic.

Our instrumental variable analysis showed that targeted therapy use was associated with an increase in median overall survival compared with nontargeted therapy use. Median overall survival increased from a nonsignificant difference of 1.6 months (95% CI, −0.5 to 2.1 months) without any adjustment to a statistically significant 3.0 months (95% CI, 0.7-5.3 months) after adjusting for both observed and unobserved covariates (Table 2). The RCC-specific survival findings were similar, with gains in median survival associated with targeted therapy increasing from an unadjusted 2.7 months (95% CI, −0.4 to 5.8 months) to an adjusted 4.9 months (95% CI, 0.4-9.4 months). According to the instrumental variable analysis, overall survival improvements associated with targeted therapy vs nontargeted therapy were statistically significant: 8% at 1 year (44% [95% CI, 39%-50%] vs 36% [95% CI, 30%-42%]; P = .01), 7% at 2 years (25% [95% CI, 20%-30%] vs 18% [95% CI, 13%-23%]; P = .009), and 5% at 3 years (15% [95% CI, 11%-19%] vs 10% [95% CI, 6%-13%]; P = .01) (Figure 2A). The RCC-specific survival rate was also significantly higher for targeted therapy compared with nontargeted therapy at 1 year (55% [95% CI, 49%-61%] vs 47% [95% CI, 40%-53%]), at 2 years (36% [95% CI, 30%-42%] vs 28% [95% CI, 21%-34%]), and at 3 years (25% [95% CI, 20%-31%] vs 18% [95% CI, 12%-24%]) (Figure 2B).

Figure 2. Estimated Survival Rates at 1, 2, and 3 Years After Receiving Targeted Therapy or Nontargeted Therapy.

Figure 2.

Open in a new tab

Survival rates were estimated from a 2-stage residual inclusion model adjusting for patients’ sociodemographic characteristics (age, sex, race, marital status, census region, and urban vs rural residence), county-level characteristics (income, unemployment rate, and physicians per 1000 residents), clinical characteristics (renal cell carcinoma grade, selected metastatic sites [lung, liver, bone, and brain], radiation status [yes or no], nephrectomy status [yes or no], National Cancer Institute–Charlson comorbidity score, claims-based disability score, and fatigue), and time to first drug treatment. P values for statistical comparisons were estimated by assessing the predicted marginal difference in survival between treatment and control groups at each time point. 95% CIs are shown as error bars. NTT indicates nontargeted therapy; TT, targeted therapy.

In the main sample, receipt of targeted therapy was associated with a statistically significant lower hazard ratio of death compared with nontargeted therapy (overall survival hazard ratio, 0.78 [95% CI, 0.65-0.94]; RCC-specific survival hazard ratio, 0.77 [95% CI, 0.62-0.96]) (Table 3). Subgroup analyses of patients aged 65 years and older showed similar and statistically significant results. Subgroup analyses of patients who had undergone nephrectomy indicated even lower hazard ratios that were not statistically significant, perhaps because of the small sample size (Table 3). Differences by treatment group were much smaller and not significant in other subgroups, such as patients younger than 65 years and those who had not undergone nephrectomy.

Table 3. Subgroup and Sensitivity Analyses of Estimated Treatment Effect of Targeted vs Nontargeted Therapy.

Model Overall Survival Renal Cell Carcinoma–Specific Survival
Hazard Ratio (95% CI) P Value Hazard Ratio (95% CI) P Value
Instrumental variable models
Main sample (n = 1015) 0.78 (0.65-0.94) .01 0.77 (0.62-0.96) .02
Subgroup analyses
Age, y
≥65 (n = 882) 0.77 (0.63-0.95) .01 0.76 (0.60-0.97) .03
<65 (n = 133) 0.95 (0.49-1.85) .88 0.96 (0.45-2.03) .92
Underwent nephrectomy
Yes (n = 374) 0.73 (0.53-1.01) .06 0.74 (0.51-1.08) .11
No (n = 641) 0.84 (0.66-1.07) .15 0.82 (0.62-1.08) .15
Sensitivity analyses
Excluding patients receiving treatment in 2005 and 2006 (n = 939) 0.78 (0.64-0.95) .01 0.77 (0.62-0.97) .03
Excluding patients with other cancers (n = 881) 0.81 (0.66-1.00) .05 0.78 (0.63-0.98) .03
Excluding brain as metastatic site (n = 888) 0.82 (0.67-1.00) .05 0.80 (0.63-1.01) .07
Measuring survival from diagnosis date (n = 1015) 0.76 (0.63-0.92) .01 0.75 (0.61-0.94) .01
Cox survival model (n = 1015) 0.78 (0.65-0.94) .01 0.78 (0.63-0.97) .03
Open in a new tab

Sensitivity analyses in our main sample confirmed the robustness of our findings. A change in the sample selection criteria, measuring survival from mRCC diagnosis date (rather than first drug treatment date), and using the Cox survival model (rather than the Weibull model) in the instrumental variable analysis all led to consistent results (Table 3).

Discussion

To our knowledge, this is the first observational comparative effectiveness study of targeted vs nontargeted therapy for mRCC using an instrumental variable approach to control for both observed and unobserved confounders. In keeping with clinical trial results, we found that the use of targeted therapy was associated with longer overall survival and RCC-specific survival compared with the use of nontargeted therapy among Medicare beneficiaries who received a diagnosis of mRCC. We observed these advantages despite the fact that a larger percentage of the targeted therapy group had baseline characteristics that placed them at higher risk of worse survival outcomes (eg, not married, a claim for fatigue, or a higher mean National Cancer Institute comorbidity score).33,34 Our instrumental variable analyses revealed larger survival improvements associated with targeted therapy compared with statistical models that adjusted only for observed covariates, suggesting that observational studies using alternate methods may underestimate true survival advantages.

At the same time, we observed substantially shorter median survival times among these real-world patients enrolled in Medicare than have been reported for individuals participating in randomized clinical trials of targeted therapies.2,3,4,35 One potential explanation is that clinical trials in advanced RCC typically include both individuals who are experiencing progression of disease initially diagnosed at an earlier stage and those with late-stage diagnoses, whereas our sample was identified through SEER staging data and included only patients with stage IV disease at diagnosis. These individuals may have a different clinical course than those who receive a diagnosis at an earlier stage. In addition, the Medicare population is, by definition, older and/or disabled, and clinical trials typically enroll individuals who are younger and healthier than the general population of patients with cancer. Even older clinical trial participants may be healthier than their same-age peers in routine clinical practice; for example, older clinical trial participants have been found to have a rate of nephrectomy that was relatively similar to that of younger trial participants (71% vs 82%-100%),2,3,4,35 whereas only 35% of patients in our targeted therapy group had undergone nephrectomy. In keeping with prior findings that patients who have undergone nephrectomy tend to be healthier and have better survival outcomes than those who have not,36 our subgroup analysis of patients who had undergone nephrectomy showed longer median overall survival times compared with those who had not (14 months in the targeted therapy group for those who had undergone nephrectomy vs 6 months among those who had not undergone nephrectomy; data not shown). Median overall survival in that subgroup was still lower than what has been observed in some clinical trials, however. Therefore, our outcomes likely reflect the fact that elderly patients and those with more medical complexity may face special treatment concerns.37,38

Our findings have several key implications for clinical practice. First, the demographic and clinical characteristics of the patients observed in our real-world sample suggest that targeted therapies likely offered new options for individuals who might have foregone treatment in the era before targeted therapy because of toxicity concerns,10 and our results underscore that survival benefits may vary when treatments are administered to a broader population of patients. This finding highlights the need for patient-centered conversations that focus on potential benefits and risks in the context of the diverse personal circumstances, values, and goals of individuals treated in real-world settings. For example, individual patients may emphasize specific quality-of-life considerations (eg, willingness to forego some progression-free survival in exchange for less-severe fatigue39) and may vary in terms of tolerance for risk when a treatment has uncertain outcomes.38,40,41,42,43 For example, 1 study found that individuals with advanced cancers endorsed treatments that provide a chance of tail-of-the-curve survival with greater frequency than physicians did44; in our study, unadjusted median overall survival for the targeted therapy group was 9 months, yet 10% of patients who received targeted therapy survived longer than 47 months (compared with 36 months in the nontargeted therapy group).

Second, it is difficult to capture the full impact of expanded options for patients and families, even when survival advantages are modest or variable. Several additional targeted therapies have been approved since our study period, and a recent analysis found greater survival improvements in the late targeted therapy era (2010-2012) compared with the earlier years we examined.45 Newer immunotherapy agents are also showing promise.9 Thus, even a treatment with limited benefits may still offer patients real option value, or the ability to survive long enough to take advantage of future breakthroughs.46,47

Third, although our study did not examine treatment adherence, it is important to note that many targeted therapies are oral medications, and real-world adherence patterns are often less consistent than those found under carefully monitored clinical trial conditions. For example, a prospective study conducted in an outpatient clinic of a major medical center found that one-third of patients with mRCC did not achieve a target of 90% adherence to their oral targeted therapy during the 12-week study period.48 Financial and other barriers also have been shown to be associated with adherence issues in oncology, particularly in the Medicare population.49,50 Furthermore, it is more difficult to detect suboptimal adherence to oral medications; missed visits for infused therapies are visible to the care team, but patients may not report inconsistent adherence to oral therapies. Proactive, ongoing conversations concerning the importance of adherence, as well as troubleshooting of adherence barriers, are especially important to ensure that individuals receive optimal treatment benefits in routine care.

Limitations

Our study has several limitations. First, because we focused on the Medicare population (including those who were eligible for Medicare because of age or disability status), our results may not be generalizable to younger or healthier populations. Second, we did not have access to fine-grained details of patient treatment regimens; thus, our findings do not offer insight into whether the dosing or duration of treatment received by the individuals in our study was consistent with clinical trial regimens. Oncologists may make different decisions regarding the intensity of treatment regimens with older patients because of concerns about adverse effects or overall clinical status,51,52 and we were not able to determine the degree to which this may have occurred or whether this may have affected the survival outcomes we observed. For example, some adverse effects have been found to be more common among older clinical trial participants, and we do not know whether adverse effects may have led to dose reductions or discontinuation in our Medicare population.53 In addition, rather than analyzing the average treatment effects for a group of patients (as in a randomized trial), the instrumental variable analysis estimate focuses on the treatment effect among marginal patients, or those whose choice of treatment is affected by the instrumental variable.22 In our study, this means that our results are generalizable only to patients whose treatment assignment was influenced by the year of their mRCC diagnosis.

Conclusions

Targeted therapies may have expanded treatment options for a complex US patient population covered by Medicare. In this study, the use of these therapies was associated with modest survival advantages over nontargeted therapies.

Supplement.

eMethods. Additional Information About the Study Sample and Instrumental Variable Analysis

eFigure 1. Sample Selection Diagram

eFigure 2. Proportion of Treated Patients Receiving Targeted Therapy, by Year of Initial Metastatic Renal Cell Carcinoma Diagnosis

eTable 1. FDA-Approved Targeted Therapies Available During the Study Period

eTable 2. Logistic Regression: Year of Metastatic Renal Cell Carcinoma Diagnosis and Treatment Choice (Targeted Therapy vs Nontargeted Therapy)

eTable 3. Test of Independent Association Between Year of Diagnosis and Survival

eTable 4. Strength of Instrumental Variable and Overidentification Test for Instrumental Variable

eReferences

Click here for additional data file. (187.9KB, pdf)

References

  • 1.Motzer RJ, Jonasch E, Agarwal N, et al. Kidney cancer, version 2.2017, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2017;15(6):-. doi: 10.6004/jnccn.2017.0100 [DOI] [PubMed] [Google Scholar]
  • 2.Escudier B, Eisen T, Stadler WM, et al. ; TARGET Study Group . Sorafenib in advanced clear-cell renal-cell carcinoma. N Engl J Med. 2007;356(2):125-134. doi: 10.1056/NEJMoa060655 [DOI] [PubMed] [Google Scholar]
  • 3.Escudier B, Pluzanska A, Koralewski P, et al. ; AVOREN Trial Investigators . Bevacizumab plus interferon alfa-2a for treatment of metastatic renal cell carcinoma: a randomised, double-blind phase III trial. Lancet. 2007;370(9605):2103-2111. doi: 10.1016/S0140-6736(07)61904-7 [DOI] [PubMed] [Google Scholar]
  • 4.Motzer RJ, Hutson TE, Tomczak P, et al. Sunitinib versus interferon alfa in metastatic renal-cell carcinoma. N Engl J Med. 2007;356(2):115-124. doi: 10.1056/NEJMoa065044 [DOI] [PubMed] [Google Scholar]
  • 5.Motzer RJ, Escudier B, McDermott DF, et al. ; CheckMate 025 Investigators . Nivolumab versus everolimus in advanced renal-cell carcinoma. N Engl J Med. 2015;373(19):1803-1813. doi: 10.1056/NEJMoa1510665 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Iacovelli R, Sternberg CN, Porta C, et al. Inhibition of the VEGF/VEGFR pathway improves survival in advanced kidney cancer: a systematic review and meta-analysis. Curr Drug Targets. 2015;16(2):164-170. doi: 10.2174/1389450115666141120120145 [DOI] [PubMed] [Google Scholar]
  • 7.Choueiri TK, Motzer RJ. Systemic therapy for metastatic renal-cell carcinoma. N Engl J Med. 2017;376(4):354-366. doi: 10.1056/NEJMra1601333 [DOI] [PubMed] [Google Scholar]
  • 8.Flanigan RC, Salmon SE, Blumenstein BA, et al. Nephrectomy followed by interferon alfa-2b compared with interferon alfa-2b alone for metastatic renal-cell cancer. N Engl J Med. 2001;345(23):1655-1659. doi: 10.1056/NEJMoa003013 [DOI] [PubMed] [Google Scholar]
  • 9.Motzer RJ, Tannir NM, McDermott DF, et al. ; CheckMate 214 Investigators . Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma. N Engl J Med. 2018;378(14):1277-1290. doi: 10.1056/NEJMoa1712126 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Garcia JA, Rini BI. Recent progress in the management of advanced renal cell carcinoma. CA Cancer J Clin. 2007;57(2):112-125. doi: 10.3322/canjclin.57.2.112 [DOI] [PubMed] [Google Scholar]
  • 11.Ratta R, Verzoni E, Di Maio M, et al. Exposure to multiple lines of treatment and survival of patients with metastatic renal cell carcinoma: a real-world analysis. Clin Genitourin Cancer. 2018;16(4):e735-e742. doi: 10.1016/j.clgc.2018.01.016 [DOI] [PubMed] [Google Scholar]
  • 12.Nazha S, Tanguay S, Kapoor A, et al. Use of targeted therapy in patients with metastatic renal cell carcinoma: clinical and economic impact in a Canadian real-life setting. Curr Oncol. 2018;25(6):e576-e584. doi: 10.3747/co.25.4103 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Lakomy R, Poprach A, Bortlicek Z, et al. Utilization and efficacy of second-line targeted therapy in metastatic renal cell carcinoma: data from a national registry. BMC Cancer. 2017;17(1):880. doi: 10.1186/s12885-017-3901-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Lindskog M, Wahlgren T, Sandin R, et al. Overall survival in Swedish patients with renal cell carcinoma treated in the period 2002 to 2012: update of the RENCOMP study with subgroup analysis of the synchronous metastatic and elderly populations. Urol Oncol. 2017;35(9):541.e15-541.e22. doi: 10.1016/j.urolonc.2017.05.013 [DOI] [PubMed] [Google Scholar]
  • 15.Bennett DA. An introduction to instrumental variables analysis: part 1. Neuroepidemiology. 2010;35(3):237-240. doi: 10.1159/000319455 [DOI] [PubMed] [Google Scholar]
  • 16.Noone A, Howlader N, Krapcho M, et al. SEER cancer statistics review, 1975-2015. https://seer.cancer.gov/csr/1975_2015/. Updated September 10, 2018. Accessed February 18, 2019.
  • 17.Kaldany A, Blum KA, Paulucci DJ, et al. An evaluation of race, ethnicity, age, and sex-based representation in phase I to II renal cell carcinoma clinical trials in the United States. Urol Oncol. 2018;36(8):363.e1-363.e6. doi: 10.1016/j.urolonc.2018.05.006 [DOI] [PubMed] [Google Scholar]
  • 18.Unger JM, Coltman CA Jr, Crowley JJ, et al. Impact of the year 2000 Medicare policy change on older patient enrollment to cancer clinical trials. J Clin Oncol. 2006;24(1):141-144. doi: 10.1200/JCO.2005.02.8928 [DOI] [PubMed] [Google Scholar]
  • 19.Ng CS, Wood CG, Silverman PM, Tannir NM, Tamboli P, Sandler CM. Renal cell carcinoma: diagnosis, staging, and surveillance. AJR Am J Roentgenol. 2008;191(4):1220-1232. doi: 10.2214/AJR.07.3568 [DOI] [PubMed] [Google Scholar]
  • 20.Saad AM, Gad MM, Al-Husseini MJ, Ruhban IA, Sonbol MB, Ho TH. Trends in renal-cell carcinoma incidence and mortality in the United States in the last 2 decades: a SEER-based study. Clin Genitourin Cancer. 2019;17(1):46-57.e5. doi: 10.1016/j.clgc.2018.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344-349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
  • 22.Newhouse JP, McClellan M. Econometrics in outcomes research: the use of instrumental variables. Annu Rev Public Health. 1998;19:17-34. doi: 10.1146/annurev.publhealth.19.1.17 [DOI] [PubMed] [Google Scholar]
  • 23.National Cancer Institute Overview of the SEER program. https://seer.cancer.gov/about/overview.html. Accessed February 18, 2019.
  • 24.Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care. 2002;40(8)(suppl):IV-3-IV-18. [DOI] [PubMed] [Google Scholar]
  • 25.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373-383. doi: 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]
  • 26.National Cancer Institute NCI comorbidity index overview. https://healthcaredelivery.cancer.gov/seermedicare/considerations/comorbidity.html. Updated April 30, 2019. Accessed February 18, 2019.
  • 27.Davidoff AJ, Zuckerman IH, Pandya N, et al. A novel approach to improve health status measurement in observational claims-based studies of cancer treatment and outcomes. J Geriatr Oncol. 2013;4(2):157-165. doi: 10.1016/j.jgo.2012.12.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Davidoff AJ, Gardner LD, Zuckerman IH, Hendrick F, Ke X, Edelman MJ. Validation of disability status, a claims-based measure of functional status for cancer treatment and outcomes studies. Med Care. 2014;52(6):500-510. doi: 10.1097/MLR.0000000000000122 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Hansen RN, Hackshaw MD, Nagar SP, et al. Health care costs among renal cancer patients using pazopanib and sunitinib. J Manag Care Spec Pharm. 2015;21(1):37-44, 44a. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Terza JV, Basu A, Rathouz PJ. Two-stage residual inclusion estimation: addressing endogeneity in health econometric modeling. J Health Econ. 2008;27(3):531-543. doi: 10.1016/j.jhealeco.2007.09.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Terza JV, Bradford WD, Dismuke CE. The use of linear instrumental variables methods in health services research and health economics: a cautionary note. Health Serv Res. 2008;43(3):1102-1120. doi: 10.1111/j.1475-6773.2007.00807.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Tan HJ, Norton EC, Ye Z, Hafez KS, Gore JL, Miller DC. Long-term survival following partial vs radical nephrectomy among older patients with early-stage kidney cancer. JAMA. 2012;307(15):1629-1635. doi: 10.1001/jama.2012.475 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Bower JE. Cancer-related fatigue—mechanisms, risk factors, and treatments. Nat Rev Clin Oncol. 2014;11(10):597-609. doi: 10.1038/nrclinonc.2014.127 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Aizer AA, Chen MH, McCarthy EP, et al. Marital status and survival in patients with cancer. J Clin Oncol. 2013;31(31):3869-3876. doi: 10.1200/JCO.2013.49.6489 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Motzer RJ, Hutson TE, Cella D, et al. Pazopanib versus sunitinib in metastatic renal-cell carcinoma. N Engl J Med. 2013;369(8):722-731. doi: 10.1056/NEJMoa1303989 [DOI] [PubMed] [Google Scholar]
  • 36.Li P, Wong YN, Armstrong K, et al. Survival among patients with advanced renal cell carcinoma in the pretargeted versus targeted therapy eras. Cancer Med. 2016;5(2):169-181. doi: 10.1002/cam4.574 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Kanesvaran R, Le Saux O, Motzer R, et al. Elderly patients with metastatic renal cell carcinoma: position paper from the International Society of Geriatric Oncology. Lancet Oncol. 2018;19(6):e317-e326. doi: 10.1016/S1470-2045(18)30125-6 [DOI] [PubMed] [Google Scholar]
  • 38.Quivy A, Daste A, Harbaoui A, et al. Optimal management of renal cell carcinoma in the elderly: a review. Clin Interv Aging. 2013;8:433-442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Wong MK, Mohamed AF, Hauber AB, et al. Patients rank toxicity against progression free survival in second-line treatment of advanced renal cell carcinoma. J Med Econ. 2012;15(6):1139-1148. doi: 10.3111/13696998.2012.708689 [DOI] [PubMed] [Google Scholar]
  • 40.González JM, Doan J, Gebben DJ, Boeri M, Fishman M. Comparing the relative importance of attributes of metastatic renal cell carcinoma treatments to patients and physicians in the United States: a discrete-choice experiment. Pharmacoeconomics. 2018;36(8):973-986. doi: 10.1007/s40273-018-0640-7 [DOI] [PubMed] [Google Scholar]
  • 41.Doshi JA, Miller Sonet E, Puckett JT, Glick H The need for a new patient-centered decision tool for value-based treatment choices in oncology. Health Affairs Blog. https://www.healthaffairs.org/do/10.1377/hblog20180309.241877/full/. Published March 19, 2018. Accessed May 9, 2019.
  • 42.Lakdawalla DN, Romley JA, Sanchez Y, Maclean JR, Penrod JR, Philipson T. How cancer patients value hope and the implications for cost-effectiveness assessments of high-cost cancer therapies. Health Aff (Millwood). 2012;31(4):676-682. doi: 10.1377/hlthaff.2011.1300 [DOI] [PubMed] [Google Scholar]
  • 43.Garrison LP Jr, Kamal-Bahl S, Towse A. Toward a broader concept of value: identifying and defining elements for an expanded cost-effectiveness analysis. Value Health. 2017;20(2):213-216. doi: 10.1016/j.jval.2016.12.005 [DOI] [PubMed] [Google Scholar]
  • 44.Shafrin J, Schwartz TT, Okoro T, Romley JA. Patient versus physician valuation of durable survival gains: implications for value framework assessments. Value Health. 2017;20(2):217-223. doi: 10.1016/j.jval.2016.11.028 [DOI] [PubMed] [Google Scholar]
  • 45.Pal SK, Ghate SR, Li N, et al. Real-world survival outcomes and prognostic factors among patients receiving first targeted therapy for advanced renal cell carcinoma: a SEER-Medicare database analysis. Clin Genitourin Cancer. 2017;15(4):e573-e582. doi: 10.1016/j.clgc.2016.12.005 [DOI] [PubMed] [Google Scholar]
  • 46.Cook JP, Golec JH, Vernon JA, Pink GH. Real option value and path dependence in oncology innovation. Int J Econ Bus. 2011;18(2):225-238. doi: 10.1080/13571516.2011.584428 [DOI] [Google Scholar]
  • 47.Lakdawalla DN, Doshi JA, Garrison LP Jr, Phelps CE, Basu A, Danzon PM. Defining elements of value in health care-a health economics approach: an ISPOR special task force report [3]. Value Health. 2018;21(2):131-139. doi: 10.1016/j.jval.2017.12.007 [DOI] [PubMed] [Google Scholar]
  • 48.Jacobs JM, Pensak NA, Sporn NJ, et al. Treatment satisfaction and adherence to oral chemotherapy in patients with cancer. J Oncol Pract. 2017;13(5):e474-e485. doi: 10.1200/JOP.2016.019729 [DOI] [PubMed] [Google Scholar]
  • 49.Shen C, Zhao B, Liu L, Shih YT. Adherence to tyrosine kinase inhibitors among Medicare Part D beneficiaries with chronic myeloid leukemia. Cancer. 2018;124(2):364-373. doi: 10.1002/cncr.31050 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Olszewski AJ, Dusetzina SB, Eaton CB, Davidoff AJ, Trivedi AN. Subsidies for oral chemotherapy and use of immunomodulatory drugs among Medicare beneficiaries with myeloma. J Clin Oncol. 2017;35(29):3306-3314. doi: 10.1200/JCO.2017.72.2447 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Foster JA, Salinas GD, Mansell D, Williamson JC, Casebeer LL. How does older age influence oncologists’ cancer management? Oncologist. 2010;15(6):584-592. doi: 10.1634/theoncologist.2009-0198 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Zeng C, Wen W, Morgans AK, Pao W, Shu XO, Zheng W. Disparities by race, age, and sex in the improvement of survival for major cancers: results from the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program in the United States, 1990 to 2010. JAMA Oncol. 2015;1(1):88-96. doi: 10.1001/jamaoncol.2014.161 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Hutson TE, Bukowski RM, Rini BI, et al. Efficacy and safety of sunitinib in elderly patients with metastatic renal cell carcinoma. Br J Cancer. 2014;110(5):1125-1132. doi: 10.1038/bjc.2013.832 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement.

eMethods. Additional Information About the Study Sample and Instrumental Variable Analysis

eFigure 1. Sample Selection Diagram

eFigure 2. Proportion of Treated Patients Receiving Targeted Therapy, by Year of Initial Metastatic Renal Cell Carcinoma Diagnosis

eTable 1. FDA-Approved Targeted Therapies Available During the Study Period

eTable 2. Logistic Regression: Year of Metastatic Renal Cell Carcinoma Diagnosis and Treatment Choice (Targeted Therapy vs Nontargeted Therapy)

eTable 3. Test of Independent Association Between Year of Diagnosis and Survival

eTable 4. Strength of Instrumental Variable and Overidentification Test for Instrumental Variable

eReferences

Click here for additional data file. (187.9KB, pdf)

Articles from JAMA Network Open are provided here courtesy of American Medical Association

RESOURCES