- Email: [email protected]
Clinical prognostic factors associated with outcome in patients with renal cell cancer with prior tyrosine kinase inhibitors or immunotherapy treated with everolimus
Urologic Oncology: Seminars and Original Investigations ] (2013) ∎∎∎–∎∎∎
Review article
Clinical prognostic factors associated with outcome in patients with renal cell cancer with prior tyrosine kinase inhibitors or immunotherapy treated with everolimus Robert J. Amato, D.O.a,*, Amber Flaherty, M.D.b, Yufeng Zhangc, Fangqian Ouyangd, Virginia Mohlerec a
Division of Oncology, Department of Internal Medicine, The University of Texas Health Science Center at Houston (Medical School), Memorial Hermann Cancer Center, Houston, TX b The University of Texas Health Science Center at Houston, Houston, TX c Division of Oncology, The University of Texas Health Science Center at Houston, Houston, TX d The University of Texas School of Public Health, Houston, TX Received 20 June 2013; received in revised form 3 September 2013; accepted 3 September 2013
Abstract Background: The mTOR inhibitor, everolimus, is approved for the treatment of metastatic renal cell carcinoma (RCC). However, prognostic models are needed to determine the patients who would most benefit from this therapy. We have developed a model based on clinical parameters and patient stratification into risk groups to predict patients with RCC who will derive the most benefit from treatment with everolimus. Methods: We assessed retrospective data on 57 patients with RCC who received everolimus after previous treatment with immunotherapy or tyrosine kinase inhibitors to identify prognostic factors for progression-free survival (PFS) and overall survival (OS). In the original phase II study, patients received 10 mg of everolimus daily without interruption and were assessed every other week for the first 8 weeks on therapy and every 4 weeks thereafter. Kaplan-Meier analysis was used to calculate OS and PFS. Univariate and multivariate analyses were constructed using the Cox proportional hazards model and a stepwise procedure to validate the data. Results: We grouped patients according to risk: 0 prognostic factors indicated favorable risk, 1 to 2 factors intermediate risk, and Z 3 factors poor risk. We found notable differences in median OS (29.6 mo for favorable risk, 14.3 mo for intermediate risk, and 7.2 mo for poor risk). Three risk factors (prior radiation treatment, no lung metastasis present at the start of treatment, and lymphocytes o 25 cells/ml) in the multivariate analysis were found to be associated with PFS, and 4 risk factors were found to be associated with OS (bone metastasis at start of treatment, LDH 4 1.5 upper limit of normal, alkaline phosphatase 4 120 U/l, and lymphocytes o 25 cells/ml). Conclusions: Our prognostic model includes 3 readily available clinical parameters for PFS and 4 readily available clinical parameters for OS to help stratify patients into poor, intermediate, and favorable prognosis groups for the treatment of everolimus in clear cell RCC. These intriguing results warrant further study in a larger patient population to validate the findings. r 2013 Elsevier Inc. All rights reserved. Keywords: Everolimus; Prognostic model; Renal cell carcinoma
1. Introduction The incidence of renal cell cancer (RCC) has increased at a steady rate for the past decade. Although localized disease has a good prognosis, RCC is usually resistant to both chemotherapy and radiation therapy, and the 5-year survival * Corresponding author. Tel.: þ1-832-325-7702; fax: þ1-713-512-7132. E-mail address: [email protected] (R.J. Amato).
1078-1439/$ – see front matter r 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.urolonc.2013.09.008
rate for patients who go on to develop distant metastases is only 11.1% [1]. Despite this, advances in understanding the pathophysiology of carcinogenesis of clear cell RCC (cRCC) have allowed for the development of novel therapies. For example, the von Hippel-Lindau gene has loss of function in many RCC cases, which helped identify the critical role of mammalian target of rapamycin (mTOR) (Fig. 1). mTOR is a biological switch that responds to upstream signals of changing cellular environments by
2
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Fig. 1. mTOR mechanism of action. (Color version of the figure is available online.)
turning on and off a variety of cellular processes that regulate a host of key cell functions. Abundant evidence points to dysregulated signaling through mTOR in RCC, especially in cRCC. mTOR is also located downstream of other RCC therapy targets, such as vascular endothelial growth factor (VEGF) and VEGF tyrosine kinase inhibitors (TKIs), which makes it a good target for combination and salvage therapies [1,2]. First-line therapy for cRCC usually includes TKIs, such as sunitinib, sorafenib, or pazopanib. Everolimus (Afinitor) was approved for patients with metastatic cRCC who develop resistance to these treatments. Everolimus is a synthetic analogue of rapamycin that effectively inhibits mTOR activation. mTOR activation stimulates the production of angiogenic factors. For patients with RCC, higher levels of angiogenic factors are associated with notably worse survival than for patients with lower levels of these circulating factors [3,4]. A handful of studies have sought to create risk-factor models to identify which patients with RCC might respond best to which targeted therapies. For example, Motzer et al. [5,6] from Memorial Sloan-Kettering Cancer Center presented 2 studies in which patients were categorized according to the number of risk factors as having favorable, intermediate, or poor risk. Both studies showed that median overall survival (OS) was markedly lower in patients with more risk factors. In the first study, the prognostic factors were associated with improved survival were low Karnofsky performance status (o 80%), high lactate dehydrogenase (LDH; 4 1.5 the upper limit of normal [ULN]), low serum hemoglobin (less than the lower limit of normal [LLN]), high corrected serum calcium (410 mg/dl), and no prior nephrectomy. In the second study by Motzer et al. previous treatment with radiation; interleukin-2 or interferon-α or both; low Karnofsky performance status (o 80%); low
serum hemoglobin (4 LLN); and high corrected calcium (Z 10 mg/dl) were determined to be associated with worse survival. Later this method of grouping patients according to risk factors was verified by the Cleveland Clinic in patients with no prior systemic therapy enrolled in immunotherapy clinical trials [7]. That study found similarly reduced OS for patients with a worse risk profile. Studies of prediction models have continued past the immunotherapy era into the present standard of targeted therapy. One large retrospective study, the RECORD-1 trial, examined data from 264 patients with metastatic RCC to determine factors indicating prognosis. The researchers found that tumor size, as indicated by the sum of the longest tumor diameters, and progression-free survival (PFS) predicted longer OS. Further analysis to determine which aspect of PFS drove the influence on OS, progression of nontarget lesions (according to Response Evaluation Criteria In Solid Tumors) criteria significantly (P o 0.05) influenced OS [8]. However, despite these advances in understanding and treatment, response rates and prognosis remain poor for patients with metastatic RCC. In addition, although both risk groups and predictive factors have been studied, the 2 have rarely been combined in 1 analysis. We previously conducted a phase II study of everolimus as salvage treatment for RCC (41 patients), along with 25 additional patients from an expansion study [9]. We used this information to develop a clinical model that combines the features of previous studies, using both risk groups and predictive clinical and treatment factors to strengthen the evidence for choosing patients who stand to gain the most benefit from treatment with everolimus.
2. Materials and methods The patient population for this analysis consisted of patients with metastatic cRCC who were treated with everolimus in a prospective phase II trial, along with an extension group to expand the population of patients who had received TKIs [9]. These patients were enrolled at Memorial Hermann Cancer Center/Methodist Hospital during May 2005 to October 2006. The key eligibility criteria were RCC composed of Z 75% clear cells, measureable progressive metastatic disease, a history of r 2 prior therapy regimens for RCC other than an mTOR inhibitor, Zubrod performance status of r 2, and adequate organ function. Patients received 10-mg everolimus daily for 28 days (1 treatment cycle) without interruption. Patients were assessed every other week for the first 8 weeks on therapy and every 4 weeks thereafter. Restaging occurred every 8 weeks. Baseline demographic, clinical, and laboratory data were collected for all patients. Laboratory values were standardized against institutional upper and lower limits of normal values when appropriate. Outcome data on PFS and OS were collected. This study was approved by the institutional review board and was conducted in accordance
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
with the Declaration of Helsinki and good clinical practice. All patients provided signed informed consent. 2.1. Statistical analysis OS was measured from the date of initiation of therapy until the date of death or last follow-up. Patients who were alive or lost to follow-up were censored at the date of their last follow-up. PFS was calculated from the date of initiation of therapy until the date of death or date of progression. Patients whose disease did not progress were censored. Distributions of OS and PFS were estimated by using the Kaplan-Meier product-limit method; median OS and PFS and their 95% confidence intervals were generated. In the univariate analysis, the correlations between OS and PFS and each of the variables were analyzed using the log-rank test and the Cox proportional hazards model. For multivariate analyses, the Cox proportional hazards model with stepwise variable selection at a significance level at 0.15 for entering and removing variables was used to assess multiple factors simultaneously. The proportional hazards assumption were tested using the ASSESS statement in PROC PHREG (SAS Institute; http://www.sas.com/). Once the risk factors were determined and the final model was formed, each patient was assigned to 1 of 3 risk groups: 0 risk factors (favorable risk), 1 risk factor (intermediate risk), or Z 2 risk factors (poor risk). Survival curves for each of these groups were estimated, and the groups were compared using the log-rank test. The accuracy of our prediction model was measured using the overall concordance index with 200 bootstrap samples in the rms R package (http://cran.r-project.org/web/packages/rms) [10,11]. For both analyses, the cutoff points for categorization were based on those used in previous studies (Table 1). 3. Results
3
were associated with a lower OS: prior radiation therapy, bone metastasis, hepatic metastasis, white blood cell count 4 10.5 cells/ml, neutrophils Z 69 cells/ml, lymphocytes o 25 cells/ml, platelets Z 300 cells/ml, LDH 4 1.5 ULN, and corrected calcium o8.5 or 410.5 mmol/l (Table 2). Five risk factors were associated with a lower PFS: prior radiation, no lung metastasis at the start of treatment, lymphocytes o 25 cells/ml, platelets Z 300 cells/ml, and LDH 4 1.5 ULN. Prior radiation, lymphocytes, platelets, and LDH were significant (P o 0.05) in the univariate analysis for both PFS and OS. 3.3. Multivariate analysis As described in the Methods section, patients were segregated into 3 risk categories based on the prognostic factors identified in the univariate analysis: favorable (0 risk factors), intermediate (1–2 risk factors), or poor (Z 3 risk factors) risk. For the OS model, 4 risk factors were kept in the final model: platelets o 300 cells/ml, LDH 4 1.5 ULN, the presence of hepatic metastasis at the start of everolimus therapy, and PFS r 6 months (Table 3). Patients with favorable risk (n ¼ 17) had a median OS of 39.5 months, those with intermediate risk (n ¼ 33) had a median OS of 13.3 months, and those with poor risk (n ¼ 7) had a median OS of 3.3 months (Fig. 4). One-year survival rates were 87.9% for patients with favorable risk, 58.7% for those with intermediate risk, and 7.7% for those with poor risk. For the PFS model, 3 variables were kept in the final model: lymphocytes o 25 cells/ml, LDH 4 1.5 ULN, and the absence of lung metastases (Table 4). Patients with favorable risk (n ¼ 37) had a median PFS of 9.7 months (Fig. 5). Patients with poor risk (n ¼ 20) had a median PFS of 2.8 months. The bootstrapped C-index was 0.63 for both PFS and OS.
3.1. Patient characteristics and outcomes
4. Discussion
Sixty-six patients were enrolled in the study, and 57 patients were evaluated for response after we removed those who had not received prior local therapy. Thirty-five patients (61%) had received TKIs, and 21 patients (37%) had received prior immunotherapy (1 patient's prior therapy was unknown). At the time of the analysis, all patients had discontinued their everolimus treatment. The median PFS for the entire population was 5.9 months (range, 0.4–34.5 mo; Fig. 2), and the median OS was 16.1 months (range, 9.7–23.8 mo; Fig. 3).
We analyzed clinical data from 57 patients treated with everolimus as second- or third-line therapy for metastatic cRCC to identify prognostic factors related to PFS and OS. Our multivariate analysis revealed 3 prognostic risk factors for determining PFS and 4 for determining OS. The median PFS for the entire population was 5.9 months, and the median OS was 16.1 months. Other researchers have sought to develop a better prognostic model for RCC. Although the prognostic factors vary by study, a few factors have been consistently identified: elevated platelets, elevated neutrophils, and time from diagnosis to treatment. In addition to the studies previously described by Motzer et al. [5,6], the Cleveland Clinic validated Memorial Sloan-Kettering Cancer Center criteria in a study involving 353 patients with no prior
3.2. Univariate analysis Demographic, clinical, and laboratory variables for PFS and OS were used in a univariate analysis. Nine risk factors
4
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Table 1 Univariate analysis of demographic and clinical patient characteristics and overall survival Characteristic Age, y r 60 4 60 Involved kidney Left Right Tumor size r 8.5 mm o 8.5 mm Time, diagnosis to treatment, and months o 24 Z 24 Prior therapy TKI No TKI Radiation No radiation ECOG performance status 0 Z1 Highest Furman grade 1/2 3/4 Lung metastases Yes No Bony metastases Yes No Mediastinal metastases Yes No Hepatic metastases Yes No Lymph node metastases Yes No Adrenal metastases Yes No Other metastases Yes No Number of sites r2 42 Hemoglobin level, g/dl o 14 Z 14 White blood cell count, cells 109/l r 10.5 4 10.5 Neutrophils, cells/ml o 69 Z 69 Lymphocytes, cells/ml o 25 Z 25 Platelets, cells/ml o 300 Z 300
No. of patients (no. who died)
Median OS, mo (95% CI)
30 (21) 27 (17)
14.3 (6.5–23.8) 16.1 (11.4–33.8)
27 (19) 28 (17)
17.6 (7.8–26.4) 13.0 (9.0–33.8)
28 (18) 23 (14)
19.2 (11.4–32.7) 17.6 (7.8–1)
27 (20) 27 (17)
10.2 (6.3–16.1) 21.7 (12.7–29.6)
35 22 21 36
18.7 13.3 12.7 21.7
Log-rank P value for each pairing 0.3348
0.62
0.8677 0.2107
(20) (18) (18) (20)
(7.8–1) (1.2–29.6) (6.3–16.5) (10.2–33.8)
0.8695 0.0153 0.45
50 (32) 7 (6)
17.6 (9.5–23.8) 12.2 (7.8–39.5)
13 (7) 39 (27)
13.0 (3.1–32.7) 16.5 (9.5–23.8)
42 (29) 15 (9)
13.6 (9.7–21.7) 16.5 (7.8–1)
29 (24) 27 (13)
11.4 (7.3–14.3) 32.7 (10.2–1)
17 (12) 39 (25)
13.6 (4.9–26.4) 16.5 (9.0–29.6)
15 (12) 41 (25)
8.4 (3.1–12.7) 19.2 (13.3–33.8)
31 (21) 26 (17)
13.3 (7.3–19.2) 21.7 (9.5–32.7)
11 (7) 46 (31)
9.0 (7.3–39.5) 16.5 (10.2–23.8)
22 (16) 35 (21)
13.3 (7.2–21.7) 16.5 (9.5–32.7)
36 (21) 21 (17)
23.8 (12.7–32.7) 7.8 (6.5–13.6)
43 (31) 14 (7)
13.3 (7.8–23.8) 20.2 (7.8–1)
53 (35) 4 (3)
16.5 (10.2–36.4) 9.6 (2.8–12.7)
34 (20) 21 (16)
21.7 (13.3–33.8) 7.8 (4.2–19.2)
34 (25) 19 (9)
12.7 (7.3–19.2) 33.8 (17.6–1)
40 (24) 15 (12)
21.7 (13.6–33.8) 7.2 (3.1–13.0)
0.9435
0.8966
0.0089
0.9743
0.0075
0.5043
0.6967
0.6
0.0852
0.1070
0.0342
0.0148
0.0033
0.0003
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
5
Table 1 Continued Characteristic Creatinine, mmol/l r 1.5 4 1.5 LDH ULN r 1.5 4 1.5 Alkaline phosphatase, U/l r 120 4 120 Albumin, g/l o4 Z4 Calcium, mmol/l 8.5–10.5 o8.5 or 410.5 Corrected calcium, mmol/l 8.5–10 o 8.5 or 4 10 Effect of PFS on OS r 5 mo 4 5 mo r 6.5 mo 4 6.5 mo r 14.3 mo 4 14.3 mo
No. of patients (no. who died)
Median OS, mo (95% CI)
50 (32) 7 (6)
14.3 (9.5–23.8) 19.2 (1.3–33.8)
50 (31) 7 (6)
18.7 (13.0–29.6) 4.9 (2.8–9.5)
39 (23) 18 (15)
18.7 (13.0–29.6) 7.5 (4.2–32.7)
5 (4) 52 (34)
6.5 (3.1–29.6) 16.1 (10.2–23.8)
51 (33) 6 (5)
17.6 (12.7–26.4) 6.2 (1.3–10.2)
49 (31) 8 (7)
17.6 (12.7–29.6) 5.7 (1.3–26.4)
20 37 30 27 41 16
7.2 21.7 7.8 32.7 9.7 39.5
Log-rank P value for each pairing 0.7394
o0.0001
0.0576
0.3982
0.0003
0.0012
(15) (23) (23) (15) (32) (6)
(4.1–13.6) (13.3–33.8) (6.8–13.6) (16.1–1) (7.3–13.3) (32.7–1)
o0.0001 o0.0001 o0.0001
Note: Not all data were available for all patients. P values were determined by chi-square test. Boldface values indicate statistically significant results. ULN ¼ upper limit of normal.
systemic therapy who were enrolled in immunotherapy clinical trials [7]. Median OS durations in that study were 28.6 months for patients with no risk factors, 14.6 months for those with 1 to 2 risk factors, and 4.5 months for those with 3 to 4 risk factors.
After targeted therapies came into use, Choueiri et al. [12] studied 120 patients with metastatic cRCC receiving VEGF-targeted therapy (bevacizumab, sorafenib, sunitinib, or axitinib). They found that 5 risk factors were associated with poor outcome: initial Eastern Cooperative Oncology
Fig. 2. Overall survival among all groups. (Color version of the figure is available online.)
6
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Fig. 3. Progression-free survival among all groups. (Color version of the figure is available online.)
Group performance status (4 0), time from diagnosis to treatment (o 2 y), baseline corrected serum calcium (o 8.5 or 4 10 mg/dl), and baseline platelet and neutrophil counts (4 300 and 4 4.5 K/ml, respectively). Three prognostic subgroups were identified: favorable risk (0 or 1 risk factor), intermediate risk (2 risk factors), and poor risk (Z 3 risk factors), with a median PFS of 20.1, 13, and 3.9 months, respectively. Motzer et al. [13] developed a model predicting the probability of 12-month PFS for patients who received sunitinib therapy based on the results of a randomized, phase III trial of sunitinib as first-line therapy in 375 patients with metastatic cRCC. Six parameters were identified as having an effect on 12-month PFS: corrected serum calcium, the number of metastatic sites, presence of hepatic metastases, thrombocytomsis, serum LDH, and time from diagnosis to treatment. However, risk groups were not identified in that study. In our univariate analysis, we also found that serum LDH and calcium (but not corrected calcium) significantly affected PFS. Heng et al. [14] evaluated the baseline characteristics and outcomes of 645 patients with anti-VEGF therapy-naïve metastatic RCC. Patients were treated with sunitinib, sorafenib, or bevacizumab plus interferon. Patients were categorized as having favorable (0 risk factors), intermediate (1–2 risk factors), or poor (3–6 risk factors) risk based on 6 factors found to be statistically significant: hemoglobin o LLN, corrected calcium 4 ULN, Karnofsky performance status 4 80%, time from diagnosis to treatment o 1 year, platelets 4 ULN, and neutrophils 4 ULN. Median OS had not yet been reached for the favorable risk group. Median OS was 27 months for the intermediate risk group and 8.8 months for the poor risk group. The median OS
numbers were similar to what we found for our favorable (29.6 mo) and poor (7.2) risk groups, even though the analyses shares only 1 clinical parameter (platelet levels). The International Kidney Cancer Working Group conducted a large retrospective study to determine prognostic factors in metastatic RCC. As with previous studies, patients were divided into 3 risk groups, and risk factors were identified that were associated with median OS. They found widely varying survival in each group: 26.9 months for those with favorable risk, 11.5 months for those with intermediate risk, and 4.2 months for those with poor risk. Treatment, performance status, number of metastatic sites, time from diagnosis to treatment, and pretreatment levels of hemoglobin, white blood cells, LDH, alkaline phosphatase, and serum calcium were the risk factors most predictive in the model [15]. Again, median OS was similar to what we found in our model, and the prognostic factors in that study were similar to ours. A phase III trial of everolimus as second-line treatment for RCC also included a prognostic model. Although risk groups were not used, prognostic factors indicative of shorter OS in that study included low performance status, high corrected calcium, low hemoglobin, and previous sunitinib treatment [16]. Clinical factors predictive of PFS in the multivariate analysis were receipt of prior radiation therapy, no lung metastasis present at the start of therapy, and lymphocytes o25 cells/ml. However, only the absence of lung metastasis reached statistical significance (median PFS). Although our PFS analysis revealed factors different from those in the literature, our OS and PFS curves are similar to one another, which demonstrates the consistency of our data.
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
7
Table 2 Univariate analysis of demographic and clinical patient characteristics and progression-free survival Characteristic Age, y r 60 4 60 Involved kidney Left Right Time, diagnosis to treatment, months o 24 Z 24 Prior therapy TKI No TKI Radiation No radiation ECOG performance status 0 Z1 Highest Furman grade 1/2 3/4 Lung metastases Yes No Bony metastases Yes No Mediastinal metastases Yes No Hepatic metastases Yes No Lymph node metastases Yes No Adrenal metastases Yes No Other metastases Yes No Number of sites r2 42 Hemoglobin level, g/dl o 14 Z 14 White blood cell count, cells 109/l r 10.5 4 10.5 Neutrophils, cells/ml o 69 Z 69 Lymphocytes, cells/ml o 25 Z 25 Platelets, cells/ml o 300 Z 300 Creatinine, mmol/l r 1.5 4 1.5
No. of patients (no. whose disease progressed)
Median OS, mo (95% CI)
30 (29) 27 (26)
502 (2.8–7.7) 9.1 (5.5–14.8)
27 (27) 28 (26)
5.4 (2.8–8.3) 7.7 (5.3–13.0)
27 (26) 27 (26)
5.3 (2.8–8.3) 7.2 (3.5–12.0)
35 22 21 36
5.5 8.3 5.0 8.3
Log-rank P value for each pairing 0.1460
0.1918
0.8758
(35) (20) (21) (34)
(4.1–7.7) (2.0–16.5) (1.9–7.5) (5.3–16.1)
0.0783 0.0094 0.6776
50 (48) 7 (7)
5.9 (5.0–8.3) 6.0 (0.4–13.0)
13 (13) 39 (37)
7.8 (1.8–13.0) 7.2 (4.2–12.0)
42 (40) 15 (15)
7.4 (5.3–9.7) 4.1 (0.8–7.7)
29 (28) 27 (26)
5.5 (2.8–7.7) 8.3 (4.2–16.5)
17 (15) 39 (39)
8.3 (3.5–16.5) 5.6 (4.1–7.8)
15 (15) 41 (39)
5.3 (1.8–6.0) 7.5 (4.6–13.0)
31 (29) 26 (26)
5.6 (1.8–8.3) 7.6 (2.7–14.8)
11 (11) 46 (44)
5.5 (1.8–9.1) 6.0 (4.2–8.3)
22 (21) 35 (34)
5.4 (3.5–9.1) 7.2 (5.0–12.0)
36 (36) 21 (19)
6.6 (4.2–9.8) 5.6 (3.5–9.7)
43 (42) 14 (13)
5.6 (4.2–7.7) 10.9 (3.6–20.2)
53 (51) 4 (4)
6.0 (5.0–9.7) 4.4 (2.7–7.8)
34 (33) 21 (20)
9.4 (5.0–16.1) 5.5 (2.7–7.2)
34 (33) 19 (18)
5.5 (2.8–7.5) 12.0 (5.9–19.3)
40 (38) 15 (15)
7.4 (5.4–12.0) 2.8 (1.3–7.8)
50 (48) 7 (7)
5.6 (4.2–8.3) 8.3 (1.3–19.3)
0.5054
0.0363
0.1517
0.1088
0.0768
0.9393
0.4729
0.6874
0.5399
0.1679
0.1812
0.1289
0.0452
0.0193
0.7673
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
8 Table 2 Continued Characteristic LDH ULN r 1.5 4 1.5 Alkaline phosphatase, U/l r 120 4 120 Albumin, g/l o4 Z4 Calcium, mmol/l 8.5–10.5 o 8.5 or 4 10.5 Corrected calcium, mmol/l 8.5–10 o 8.5 or 4 10
No. of patients (no. whose disease progressed)
Median OS, mo (95% CI)
Log-rank P value for each pairing 0.0524
50 (48) 7 (7)
6.6 (5.0–9.8) 3.5 (1.9–7.7)
39 (38) 18 (17)
7.2 (5.0–12.0) 5.2 (2.0–7.8)
5 (5) 52 (50)
2.8 (0.8–16.5) 6.6 (5.0–9.8)
51 (49) 6 (6)
6.0 (5.0–9.8) 3.2 (1.3–8.3)
49 (47) 8 (8)
7.2 (5.3–9.7) 3.2 (1.3–8.3)
0.6952
0.2012
0.0389
0.0728
Note: Not all data were available for all patients. P values were determined by chi-square test. Boldface values indicate statistically significant results. ULN ¼ upper limit of normal.
Table 3 Multivariate analysis of risk factors associated with overall survival Factor, level
Parameter estimate ⫾ SE
Hazard ratio
95% confidence interval
P value
Hepatic metastases Platelets 4 300 cells/ml LDH 41.5 ULN PFS r 6 months
0.91 ⫾ 0.40 1.18 ⫾ 0.44 1.90 ⫾ 0.58 1.88 ⫾ 0.41
2.49 3.25 6.75 6.55
1.13–5.46 1.38–7.64 1.84–12.91 2.93–14.65
0.02 0.007 0.001 o 0.001
Note: P values were determined by chi-square test. Boldface values indicate statistically significant results. ULN, upper limit of normal.
The clinical features that we identified as predictive of OS in everolimus-treated patients with metastatic cRCC are in line with those of previous studies. In multivariate analysis, the risk factors with statistically significant effect OS were LDH 4 1.5 ULN, hepatic metastasis at start of treatment, platelets Z 300 cells/ml, and PFS r 6 months. The effect of PFS on OS is notable in light of the same finding in the much larger RECORD-1 trial [8]. The median OS of patients with LDH 41.5 ULN was 4.9 months, compared with 18.7 months for those with lower levels. Hepatic metastasis meant an OS of 8.4 months, compared with 19.2 in patients without, and patients with PFS r 6 months survived for a median of only 7.8 months, compared with 32.7 months for those whose disease progressed more slowly. These differences suggest that the clinical factors we identified could be useful indicators of prognosis.
The major limitations of this study were the small population of 57 patients and the high number of patients categorized as having a favorable prognosis (0 risk factors) for OS vs. other categories. Additionally, because this was a retrospective analysis, not all patient data were available. However, our prognostic model differs from others in that we included predictions for both PFS and OS as well as both predictive factors and risk groups, making this study, to our knowledge, the most complete prognostic model to date for advanced RCC. In our institution, we will conduct a large, 164-patient study of pathway-specific agents for the treatment of RCC using quantitative proteomics for precise patient selection. We plan to use the prognostic models developed in the present study in that cohort in the hope of validating the results.
Table 4 Multivariate analysis of risk factors associated with progression-free survival Factor, level
Parameter estimate ⫾ SE
Hazard ratio
95% confidence interval
Received prior radiation therapy Absence of lung metastases Lymphocytes o 25 cells/ml
0.53 ⫾ 0.32 0.85 ⫾ 0.32 0.54 ⫾ 0.33
1.69 2.35 1.72
0.90–3.19 1.24–4.43 0.91–3.26
Note: P values were determined by chi-square test. Boldface values indicate statistically significant results.
P value 0.103 0.008 0.119
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Fig. 4. Overall survival probability per risk group. (Color version of the figure is available online.)
Fig. 5. Progression-free survival probability per risk group. (Color version of the figure is available online.)
9
10
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
References [1] Elfiky AA, Aziz SA, Conrad PJ, et al. Characterization and targeting of phosphatidylinositol-e kinase (PI3K) and mammalian target of rapamycin (mTOR) in renal cell cancer. J Transl Med 2011;9:133–43. [2] Escudier B, Thompson JA. Mechanism of action of everolimus in renal cell carcinoma. Med Oncol 2009;26:32–9. [3] Pantuck AJ, Seligson DB, Klatte T, et al. Prognostic relevance of the mTOR pathway in renal cell carcinoma: implications for molecular patient selection for targeted therapy. Cancer 2007;109:2258–65. [4] Schultz L, Chaux A, Albadine R, et al. Immunoexpression status and prognostic value of mTOR and hypoxia-induced pathway members in promary and metastatic clear cell renal cell carcinoma. Am J Surg Pathol 2011;35:1549–56. [5] Motzer RJ, Mazumdar M, Bacik J, Berg W, Amsterdam A, Ferrara J. Survival and prognostic stratification of 670 patients with advanced renal cell carcinoma. J Clin Oncol 1999;17:2530–40. [6] Motzer RJ, Bacik J, Schwartz LH, Reuter V, Russo P, Marion S, Mazumdar M. Prognostic factors for survival in previously treated patients with metastatic renal cell carcinoma. J Clin Oncol 2004;22:454–63. [7] Bukowski RM, Negrier S, Elson P. Prognostic factors in patients with advanced renal cell carcinoma: development of an International Kidney Cancer Working Group. Clin Cancer Res 2004;10:6310S–6314SS. [8] Stein A, Bellmunt J, Escudier B, Kim D, Stergiopoulos SG, Mietlowski W, et al. RECORD-1 Trial Study Group: Survival prediciton in everolimus-treated patients with metastatic renal cell carcinoma incorporating tumor burden response in the RECORD-1
[9]
[10]
[11] [12]
[13]
[14]
[15]
[16]
trial. Eur Urol 2012 [pii: S0302-2838(12)01407-8. doi: 10.1016/j. eururo.2012.11.032; Epub ahead of print]. Amato RJ, Jac J, Giessinger S, Saxena S, Willis JP. A phase 2 study with a daily regimen of the oral mTOR inhibitor RAD001 (everolimus) in patients with metastatic clear cell renal cell cancer. Cancer 2009;115:2438–46. Pencina MJ, D'Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23:2109–23. Liu L, Forman S, Barton B. Fitting Cox model using PROC PHREG and beyond in SAS. SAS Global Forum; 2009. Choueiri TK, Garcia JA, Elson P, et al. Clinical factors associated with outcome in patients with metastatic clear-cell renal cell carcinoma treated with vascular endothelial growth factor-targeted therapy. Cancer 2007;110:543–50. Motzer RJ, Bukowski RM, Figlin RA, et al. Prognostic nomogram for sunitinib in patients with metastatic renal cell carcinoma. Cancer 2008;113:1552–8. Heng DYC, Xie W, Regan MM, et al. Prognostic factors for overall survival in patients with metastatic renal cell carcinoma treated wtih vascular endothelial growth factor-targeted agents: results from a large, multicenter study. J Clin Oncol 2009;27:5794–9. Manola J, Royston P, Elson P, et al. Prognostic model for survival in patients with metastatic renal cell carcinoma: results from the International Kidney Cancer Working Group. Clin Cancer Res 2011;17:5443–50. Mozter RJ, Escudier B, Oudard S, et al. Phase 3 trial of everolimus for metastatic renal cell carcinoma. Cancer 2010;116:4256–65.
Review article
Clinical prognostic factors associated with outcome in patients with renal cell cancer with prior tyrosine kinase inhibitors or immunotherapy treated with everolimus Robert J. Amato, D.O.a,*, Amber Flaherty, M.D.b, Yufeng Zhangc, Fangqian Ouyangd, Virginia Mohlerec a
Division of Oncology, Department of Internal Medicine, The University of Texas Health Science Center at Houston (Medical School), Memorial Hermann Cancer Center, Houston, TX b The University of Texas Health Science Center at Houston, Houston, TX c Division of Oncology, The University of Texas Health Science Center at Houston, Houston, TX d The University of Texas School of Public Health, Houston, TX Received 20 June 2013; received in revised form 3 September 2013; accepted 3 September 2013
Abstract Background: The mTOR inhibitor, everolimus, is approved for the treatment of metastatic renal cell carcinoma (RCC). However, prognostic models are needed to determine the patients who would most benefit from this therapy. We have developed a model based on clinical parameters and patient stratification into risk groups to predict patients with RCC who will derive the most benefit from treatment with everolimus. Methods: We assessed retrospective data on 57 patients with RCC who received everolimus after previous treatment with immunotherapy or tyrosine kinase inhibitors to identify prognostic factors for progression-free survival (PFS) and overall survival (OS). In the original phase II study, patients received 10 mg of everolimus daily without interruption and were assessed every other week for the first 8 weeks on therapy and every 4 weeks thereafter. Kaplan-Meier analysis was used to calculate OS and PFS. Univariate and multivariate analyses were constructed using the Cox proportional hazards model and a stepwise procedure to validate the data. Results: We grouped patients according to risk: 0 prognostic factors indicated favorable risk, 1 to 2 factors intermediate risk, and Z 3 factors poor risk. We found notable differences in median OS (29.6 mo for favorable risk, 14.3 mo for intermediate risk, and 7.2 mo for poor risk). Three risk factors (prior radiation treatment, no lung metastasis present at the start of treatment, and lymphocytes o 25 cells/ml) in the multivariate analysis were found to be associated with PFS, and 4 risk factors were found to be associated with OS (bone metastasis at start of treatment, LDH 4 1.5 upper limit of normal, alkaline phosphatase 4 120 U/l, and lymphocytes o 25 cells/ml). Conclusions: Our prognostic model includes 3 readily available clinical parameters for PFS and 4 readily available clinical parameters for OS to help stratify patients into poor, intermediate, and favorable prognosis groups for the treatment of everolimus in clear cell RCC. These intriguing results warrant further study in a larger patient population to validate the findings. r 2013 Elsevier Inc. All rights reserved. Keywords: Everolimus; Prognostic model; Renal cell carcinoma
1. Introduction The incidence of renal cell cancer (RCC) has increased at a steady rate for the past decade. Although localized disease has a good prognosis, RCC is usually resistant to both chemotherapy and radiation therapy, and the 5-year survival * Corresponding author. Tel.: þ1-832-325-7702; fax: þ1-713-512-7132. E-mail address: [email protected] (R.J. Amato).
1078-1439/$ – see front matter r 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.urolonc.2013.09.008
rate for patients who go on to develop distant metastases is only 11.1% [1]. Despite this, advances in understanding the pathophysiology of carcinogenesis of clear cell RCC (cRCC) have allowed for the development of novel therapies. For example, the von Hippel-Lindau gene has loss of function in many RCC cases, which helped identify the critical role of mammalian target of rapamycin (mTOR) (Fig. 1). mTOR is a biological switch that responds to upstream signals of changing cellular environments by
2
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Fig. 1. mTOR mechanism of action. (Color version of the figure is available online.)
turning on and off a variety of cellular processes that regulate a host of key cell functions. Abundant evidence points to dysregulated signaling through mTOR in RCC, especially in cRCC. mTOR is also located downstream of other RCC therapy targets, such as vascular endothelial growth factor (VEGF) and VEGF tyrosine kinase inhibitors (TKIs), which makes it a good target for combination and salvage therapies [1,2]. First-line therapy for cRCC usually includes TKIs, such as sunitinib, sorafenib, or pazopanib. Everolimus (Afinitor) was approved for patients with metastatic cRCC who develop resistance to these treatments. Everolimus is a synthetic analogue of rapamycin that effectively inhibits mTOR activation. mTOR activation stimulates the production of angiogenic factors. For patients with RCC, higher levels of angiogenic factors are associated with notably worse survival than for patients with lower levels of these circulating factors [3,4]. A handful of studies have sought to create risk-factor models to identify which patients with RCC might respond best to which targeted therapies. For example, Motzer et al. [5,6] from Memorial Sloan-Kettering Cancer Center presented 2 studies in which patients were categorized according to the number of risk factors as having favorable, intermediate, or poor risk. Both studies showed that median overall survival (OS) was markedly lower in patients with more risk factors. In the first study, the prognostic factors were associated with improved survival were low Karnofsky performance status (o 80%), high lactate dehydrogenase (LDH; 4 1.5 the upper limit of normal [ULN]), low serum hemoglobin (less than the lower limit of normal [LLN]), high corrected serum calcium (410 mg/dl), and no prior nephrectomy. In the second study by Motzer et al. previous treatment with radiation; interleukin-2 or interferon-α or both; low Karnofsky performance status (o 80%); low
serum hemoglobin (4 LLN); and high corrected calcium (Z 10 mg/dl) were determined to be associated with worse survival. Later this method of grouping patients according to risk factors was verified by the Cleveland Clinic in patients with no prior systemic therapy enrolled in immunotherapy clinical trials [7]. That study found similarly reduced OS for patients with a worse risk profile. Studies of prediction models have continued past the immunotherapy era into the present standard of targeted therapy. One large retrospective study, the RECORD-1 trial, examined data from 264 patients with metastatic RCC to determine factors indicating prognosis. The researchers found that tumor size, as indicated by the sum of the longest tumor diameters, and progression-free survival (PFS) predicted longer OS. Further analysis to determine which aspect of PFS drove the influence on OS, progression of nontarget lesions (according to Response Evaluation Criteria In Solid Tumors) criteria significantly (P o 0.05) influenced OS [8]. However, despite these advances in understanding and treatment, response rates and prognosis remain poor for patients with metastatic RCC. In addition, although both risk groups and predictive factors have been studied, the 2 have rarely been combined in 1 analysis. We previously conducted a phase II study of everolimus as salvage treatment for RCC (41 patients), along with 25 additional patients from an expansion study [9]. We used this information to develop a clinical model that combines the features of previous studies, using both risk groups and predictive clinical and treatment factors to strengthen the evidence for choosing patients who stand to gain the most benefit from treatment with everolimus.
2. Materials and methods The patient population for this analysis consisted of patients with metastatic cRCC who were treated with everolimus in a prospective phase II trial, along with an extension group to expand the population of patients who had received TKIs [9]. These patients were enrolled at Memorial Hermann Cancer Center/Methodist Hospital during May 2005 to October 2006. The key eligibility criteria were RCC composed of Z 75% clear cells, measureable progressive metastatic disease, a history of r 2 prior therapy regimens for RCC other than an mTOR inhibitor, Zubrod performance status of r 2, and adequate organ function. Patients received 10-mg everolimus daily for 28 days (1 treatment cycle) without interruption. Patients were assessed every other week for the first 8 weeks on therapy and every 4 weeks thereafter. Restaging occurred every 8 weeks. Baseline demographic, clinical, and laboratory data were collected for all patients. Laboratory values were standardized against institutional upper and lower limits of normal values when appropriate. Outcome data on PFS and OS were collected. This study was approved by the institutional review board and was conducted in accordance
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
with the Declaration of Helsinki and good clinical practice. All patients provided signed informed consent. 2.1. Statistical analysis OS was measured from the date of initiation of therapy until the date of death or last follow-up. Patients who were alive or lost to follow-up were censored at the date of their last follow-up. PFS was calculated from the date of initiation of therapy until the date of death or date of progression. Patients whose disease did not progress were censored. Distributions of OS and PFS were estimated by using the Kaplan-Meier product-limit method; median OS and PFS and their 95% confidence intervals were generated. In the univariate analysis, the correlations between OS and PFS and each of the variables were analyzed using the log-rank test and the Cox proportional hazards model. For multivariate analyses, the Cox proportional hazards model with stepwise variable selection at a significance level at 0.15 for entering and removing variables was used to assess multiple factors simultaneously. The proportional hazards assumption were tested using the ASSESS statement in PROC PHREG (SAS Institute; http://www.sas.com/). Once the risk factors were determined and the final model was formed, each patient was assigned to 1 of 3 risk groups: 0 risk factors (favorable risk), 1 risk factor (intermediate risk), or Z 2 risk factors (poor risk). Survival curves for each of these groups were estimated, and the groups were compared using the log-rank test. The accuracy of our prediction model was measured using the overall concordance index with 200 bootstrap samples in the rms R package (http://cran.r-project.org/web/packages/rms) [10,11]. For both analyses, the cutoff points for categorization were based on those used in previous studies (Table 1). 3. Results
3
were associated with a lower OS: prior radiation therapy, bone metastasis, hepatic metastasis, white blood cell count 4 10.5 cells/ml, neutrophils Z 69 cells/ml, lymphocytes o 25 cells/ml, platelets Z 300 cells/ml, LDH 4 1.5 ULN, and corrected calcium o8.5 or 410.5 mmol/l (Table 2). Five risk factors were associated with a lower PFS: prior radiation, no lung metastasis at the start of treatment, lymphocytes o 25 cells/ml, platelets Z 300 cells/ml, and LDH 4 1.5 ULN. Prior radiation, lymphocytes, platelets, and LDH were significant (P o 0.05) in the univariate analysis for both PFS and OS. 3.3. Multivariate analysis As described in the Methods section, patients were segregated into 3 risk categories based on the prognostic factors identified in the univariate analysis: favorable (0 risk factors), intermediate (1–2 risk factors), or poor (Z 3 risk factors) risk. For the OS model, 4 risk factors were kept in the final model: platelets o 300 cells/ml, LDH 4 1.5 ULN, the presence of hepatic metastasis at the start of everolimus therapy, and PFS r 6 months (Table 3). Patients with favorable risk (n ¼ 17) had a median OS of 39.5 months, those with intermediate risk (n ¼ 33) had a median OS of 13.3 months, and those with poor risk (n ¼ 7) had a median OS of 3.3 months (Fig. 4). One-year survival rates were 87.9% for patients with favorable risk, 58.7% for those with intermediate risk, and 7.7% for those with poor risk. For the PFS model, 3 variables were kept in the final model: lymphocytes o 25 cells/ml, LDH 4 1.5 ULN, and the absence of lung metastases (Table 4). Patients with favorable risk (n ¼ 37) had a median PFS of 9.7 months (Fig. 5). Patients with poor risk (n ¼ 20) had a median PFS of 2.8 months. The bootstrapped C-index was 0.63 for both PFS and OS.
3.1. Patient characteristics and outcomes
4. Discussion
Sixty-six patients were enrolled in the study, and 57 patients were evaluated for response after we removed those who had not received prior local therapy. Thirty-five patients (61%) had received TKIs, and 21 patients (37%) had received prior immunotherapy (1 patient's prior therapy was unknown). At the time of the analysis, all patients had discontinued their everolimus treatment. The median PFS for the entire population was 5.9 months (range, 0.4–34.5 mo; Fig. 2), and the median OS was 16.1 months (range, 9.7–23.8 mo; Fig. 3).
We analyzed clinical data from 57 patients treated with everolimus as second- or third-line therapy for metastatic cRCC to identify prognostic factors related to PFS and OS. Our multivariate analysis revealed 3 prognostic risk factors for determining PFS and 4 for determining OS. The median PFS for the entire population was 5.9 months, and the median OS was 16.1 months. Other researchers have sought to develop a better prognostic model for RCC. Although the prognostic factors vary by study, a few factors have been consistently identified: elevated platelets, elevated neutrophils, and time from diagnosis to treatment. In addition to the studies previously described by Motzer et al. [5,6], the Cleveland Clinic validated Memorial Sloan-Kettering Cancer Center criteria in a study involving 353 patients with no prior
3.2. Univariate analysis Demographic, clinical, and laboratory variables for PFS and OS were used in a univariate analysis. Nine risk factors
4
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Table 1 Univariate analysis of demographic and clinical patient characteristics and overall survival Characteristic Age, y r 60 4 60 Involved kidney Left Right Tumor size r 8.5 mm o 8.5 mm Time, diagnosis to treatment, and months o 24 Z 24 Prior therapy TKI No TKI Radiation No radiation ECOG performance status 0 Z1 Highest Furman grade 1/2 3/4 Lung metastases Yes No Bony metastases Yes No Mediastinal metastases Yes No Hepatic metastases Yes No Lymph node metastases Yes No Adrenal metastases Yes No Other metastases Yes No Number of sites r2 42 Hemoglobin level, g/dl o 14 Z 14 White blood cell count, cells 109/l r 10.5 4 10.5 Neutrophils, cells/ml o 69 Z 69 Lymphocytes, cells/ml o 25 Z 25 Platelets, cells/ml o 300 Z 300
No. of patients (no. who died)
Median OS, mo (95% CI)
30 (21) 27 (17)
14.3 (6.5–23.8) 16.1 (11.4–33.8)
27 (19) 28 (17)
17.6 (7.8–26.4) 13.0 (9.0–33.8)
28 (18) 23 (14)
19.2 (11.4–32.7) 17.6 (7.8–1)
27 (20) 27 (17)
10.2 (6.3–16.1) 21.7 (12.7–29.6)
35 22 21 36
18.7 13.3 12.7 21.7
Log-rank P value for each pairing 0.3348
0.62
0.8677 0.2107
(20) (18) (18) (20)
(7.8–1) (1.2–29.6) (6.3–16.5) (10.2–33.8)
0.8695 0.0153 0.45
50 (32) 7 (6)
17.6 (9.5–23.8) 12.2 (7.8–39.5)
13 (7) 39 (27)
13.0 (3.1–32.7) 16.5 (9.5–23.8)
42 (29) 15 (9)
13.6 (9.7–21.7) 16.5 (7.8–1)
29 (24) 27 (13)
11.4 (7.3–14.3) 32.7 (10.2–1)
17 (12) 39 (25)
13.6 (4.9–26.4) 16.5 (9.0–29.6)
15 (12) 41 (25)
8.4 (3.1–12.7) 19.2 (13.3–33.8)
31 (21) 26 (17)
13.3 (7.3–19.2) 21.7 (9.5–32.7)
11 (7) 46 (31)
9.0 (7.3–39.5) 16.5 (10.2–23.8)
22 (16) 35 (21)
13.3 (7.2–21.7) 16.5 (9.5–32.7)
36 (21) 21 (17)
23.8 (12.7–32.7) 7.8 (6.5–13.6)
43 (31) 14 (7)
13.3 (7.8–23.8) 20.2 (7.8–1)
53 (35) 4 (3)
16.5 (10.2–36.4) 9.6 (2.8–12.7)
34 (20) 21 (16)
21.7 (13.3–33.8) 7.8 (4.2–19.2)
34 (25) 19 (9)
12.7 (7.3–19.2) 33.8 (17.6–1)
40 (24) 15 (12)
21.7 (13.6–33.8) 7.2 (3.1–13.0)
0.9435
0.8966
0.0089
0.9743
0.0075
0.5043
0.6967
0.6
0.0852
0.1070
0.0342
0.0148
0.0033
0.0003
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
5
Table 1 Continued Characteristic Creatinine, mmol/l r 1.5 4 1.5 LDH ULN r 1.5 4 1.5 Alkaline phosphatase, U/l r 120 4 120 Albumin, g/l o4 Z4 Calcium, mmol/l 8.5–10.5 o8.5 or 410.5 Corrected calcium, mmol/l 8.5–10 o 8.5 or 4 10 Effect of PFS on OS r 5 mo 4 5 mo r 6.5 mo 4 6.5 mo r 14.3 mo 4 14.3 mo
No. of patients (no. who died)
Median OS, mo (95% CI)
50 (32) 7 (6)
14.3 (9.5–23.8) 19.2 (1.3–33.8)
50 (31) 7 (6)
18.7 (13.0–29.6) 4.9 (2.8–9.5)
39 (23) 18 (15)
18.7 (13.0–29.6) 7.5 (4.2–32.7)
5 (4) 52 (34)
6.5 (3.1–29.6) 16.1 (10.2–23.8)
51 (33) 6 (5)
17.6 (12.7–26.4) 6.2 (1.3–10.2)
49 (31) 8 (7)
17.6 (12.7–29.6) 5.7 (1.3–26.4)
20 37 30 27 41 16
7.2 21.7 7.8 32.7 9.7 39.5
Log-rank P value for each pairing 0.7394
o0.0001
0.0576
0.3982
0.0003
0.0012
(15) (23) (23) (15) (32) (6)
(4.1–13.6) (13.3–33.8) (6.8–13.6) (16.1–1) (7.3–13.3) (32.7–1)
o0.0001 o0.0001 o0.0001
Note: Not all data were available for all patients. P values were determined by chi-square test. Boldface values indicate statistically significant results. ULN ¼ upper limit of normal.
systemic therapy who were enrolled in immunotherapy clinical trials [7]. Median OS durations in that study were 28.6 months for patients with no risk factors, 14.6 months for those with 1 to 2 risk factors, and 4.5 months for those with 3 to 4 risk factors.
After targeted therapies came into use, Choueiri et al. [12] studied 120 patients with metastatic cRCC receiving VEGF-targeted therapy (bevacizumab, sorafenib, sunitinib, or axitinib). They found that 5 risk factors were associated with poor outcome: initial Eastern Cooperative Oncology
Fig. 2. Overall survival among all groups. (Color version of the figure is available online.)
6
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Fig. 3. Progression-free survival among all groups. (Color version of the figure is available online.)
Group performance status (4 0), time from diagnosis to treatment (o 2 y), baseline corrected serum calcium (o 8.5 or 4 10 mg/dl), and baseline platelet and neutrophil counts (4 300 and 4 4.5 K/ml, respectively). Three prognostic subgroups were identified: favorable risk (0 or 1 risk factor), intermediate risk (2 risk factors), and poor risk (Z 3 risk factors), with a median PFS of 20.1, 13, and 3.9 months, respectively. Motzer et al. [13] developed a model predicting the probability of 12-month PFS for patients who received sunitinib therapy based on the results of a randomized, phase III trial of sunitinib as first-line therapy in 375 patients with metastatic cRCC. Six parameters were identified as having an effect on 12-month PFS: corrected serum calcium, the number of metastatic sites, presence of hepatic metastases, thrombocytomsis, serum LDH, and time from diagnosis to treatment. However, risk groups were not identified in that study. In our univariate analysis, we also found that serum LDH and calcium (but not corrected calcium) significantly affected PFS. Heng et al. [14] evaluated the baseline characteristics and outcomes of 645 patients with anti-VEGF therapy-naïve metastatic RCC. Patients were treated with sunitinib, sorafenib, or bevacizumab plus interferon. Patients were categorized as having favorable (0 risk factors), intermediate (1–2 risk factors), or poor (3–6 risk factors) risk based on 6 factors found to be statistically significant: hemoglobin o LLN, corrected calcium 4 ULN, Karnofsky performance status 4 80%, time from diagnosis to treatment o 1 year, platelets 4 ULN, and neutrophils 4 ULN. Median OS had not yet been reached for the favorable risk group. Median OS was 27 months for the intermediate risk group and 8.8 months for the poor risk group. The median OS
numbers were similar to what we found for our favorable (29.6 mo) and poor (7.2) risk groups, even though the analyses shares only 1 clinical parameter (platelet levels). The International Kidney Cancer Working Group conducted a large retrospective study to determine prognostic factors in metastatic RCC. As with previous studies, patients were divided into 3 risk groups, and risk factors were identified that were associated with median OS. They found widely varying survival in each group: 26.9 months for those with favorable risk, 11.5 months for those with intermediate risk, and 4.2 months for those with poor risk. Treatment, performance status, number of metastatic sites, time from diagnosis to treatment, and pretreatment levels of hemoglobin, white blood cells, LDH, alkaline phosphatase, and serum calcium were the risk factors most predictive in the model [15]. Again, median OS was similar to what we found in our model, and the prognostic factors in that study were similar to ours. A phase III trial of everolimus as second-line treatment for RCC also included a prognostic model. Although risk groups were not used, prognostic factors indicative of shorter OS in that study included low performance status, high corrected calcium, low hemoglobin, and previous sunitinib treatment [16]. Clinical factors predictive of PFS in the multivariate analysis were receipt of prior radiation therapy, no lung metastasis present at the start of therapy, and lymphocytes o25 cells/ml. However, only the absence of lung metastasis reached statistical significance (median PFS). Although our PFS analysis revealed factors different from those in the literature, our OS and PFS curves are similar to one another, which demonstrates the consistency of our data.
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
7
Table 2 Univariate analysis of demographic and clinical patient characteristics and progression-free survival Characteristic Age, y r 60 4 60 Involved kidney Left Right Time, diagnosis to treatment, months o 24 Z 24 Prior therapy TKI No TKI Radiation No radiation ECOG performance status 0 Z1 Highest Furman grade 1/2 3/4 Lung metastases Yes No Bony metastases Yes No Mediastinal metastases Yes No Hepatic metastases Yes No Lymph node metastases Yes No Adrenal metastases Yes No Other metastases Yes No Number of sites r2 42 Hemoglobin level, g/dl o 14 Z 14 White blood cell count, cells 109/l r 10.5 4 10.5 Neutrophils, cells/ml o 69 Z 69 Lymphocytes, cells/ml o 25 Z 25 Platelets, cells/ml o 300 Z 300 Creatinine, mmol/l r 1.5 4 1.5
No. of patients (no. whose disease progressed)
Median OS, mo (95% CI)
30 (29) 27 (26)
502 (2.8–7.7) 9.1 (5.5–14.8)
27 (27) 28 (26)
5.4 (2.8–8.3) 7.7 (5.3–13.0)
27 (26) 27 (26)
5.3 (2.8–8.3) 7.2 (3.5–12.0)
35 22 21 36
5.5 8.3 5.0 8.3
Log-rank P value for each pairing 0.1460
0.1918
0.8758
(35) (20) (21) (34)
(4.1–7.7) (2.0–16.5) (1.9–7.5) (5.3–16.1)
0.0783 0.0094 0.6776
50 (48) 7 (7)
5.9 (5.0–8.3) 6.0 (0.4–13.0)
13 (13) 39 (37)
7.8 (1.8–13.0) 7.2 (4.2–12.0)
42 (40) 15 (15)
7.4 (5.3–9.7) 4.1 (0.8–7.7)
29 (28) 27 (26)
5.5 (2.8–7.7) 8.3 (4.2–16.5)
17 (15) 39 (39)
8.3 (3.5–16.5) 5.6 (4.1–7.8)
15 (15) 41 (39)
5.3 (1.8–6.0) 7.5 (4.6–13.0)
31 (29) 26 (26)
5.6 (1.8–8.3) 7.6 (2.7–14.8)
11 (11) 46 (44)
5.5 (1.8–9.1) 6.0 (4.2–8.3)
22 (21) 35 (34)
5.4 (3.5–9.1) 7.2 (5.0–12.0)
36 (36) 21 (19)
6.6 (4.2–9.8) 5.6 (3.5–9.7)
43 (42) 14 (13)
5.6 (4.2–7.7) 10.9 (3.6–20.2)
53 (51) 4 (4)
6.0 (5.0–9.7) 4.4 (2.7–7.8)
34 (33) 21 (20)
9.4 (5.0–16.1) 5.5 (2.7–7.2)
34 (33) 19 (18)
5.5 (2.8–7.5) 12.0 (5.9–19.3)
40 (38) 15 (15)
7.4 (5.4–12.0) 2.8 (1.3–7.8)
50 (48) 7 (7)
5.6 (4.2–8.3) 8.3 (1.3–19.3)
0.5054
0.0363
0.1517
0.1088
0.0768
0.9393
0.4729
0.6874
0.5399
0.1679
0.1812
0.1289
0.0452
0.0193
0.7673
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
8 Table 2 Continued Characteristic LDH ULN r 1.5 4 1.5 Alkaline phosphatase, U/l r 120 4 120 Albumin, g/l o4 Z4 Calcium, mmol/l 8.5–10.5 o 8.5 or 4 10.5 Corrected calcium, mmol/l 8.5–10 o 8.5 or 4 10
No. of patients (no. whose disease progressed)
Median OS, mo (95% CI)
Log-rank P value for each pairing 0.0524
50 (48) 7 (7)
6.6 (5.0–9.8) 3.5 (1.9–7.7)
39 (38) 18 (17)
7.2 (5.0–12.0) 5.2 (2.0–7.8)
5 (5) 52 (50)
2.8 (0.8–16.5) 6.6 (5.0–9.8)
51 (49) 6 (6)
6.0 (5.0–9.8) 3.2 (1.3–8.3)
49 (47) 8 (8)
7.2 (5.3–9.7) 3.2 (1.3–8.3)
0.6952
0.2012
0.0389
0.0728
Note: Not all data were available for all patients. P values were determined by chi-square test. Boldface values indicate statistically significant results. ULN ¼ upper limit of normal.
Table 3 Multivariate analysis of risk factors associated with overall survival Factor, level
Parameter estimate ⫾ SE
Hazard ratio
95% confidence interval
P value
Hepatic metastases Platelets 4 300 cells/ml LDH 41.5 ULN PFS r 6 months
0.91 ⫾ 0.40 1.18 ⫾ 0.44 1.90 ⫾ 0.58 1.88 ⫾ 0.41
2.49 3.25 6.75 6.55
1.13–5.46 1.38–7.64 1.84–12.91 2.93–14.65
0.02 0.007 0.001 o 0.001
Note: P values were determined by chi-square test. Boldface values indicate statistically significant results. ULN, upper limit of normal.
The clinical features that we identified as predictive of OS in everolimus-treated patients with metastatic cRCC are in line with those of previous studies. In multivariate analysis, the risk factors with statistically significant effect OS were LDH 4 1.5 ULN, hepatic metastasis at start of treatment, platelets Z 300 cells/ml, and PFS r 6 months. The effect of PFS on OS is notable in light of the same finding in the much larger RECORD-1 trial [8]. The median OS of patients with LDH 41.5 ULN was 4.9 months, compared with 18.7 months for those with lower levels. Hepatic metastasis meant an OS of 8.4 months, compared with 19.2 in patients without, and patients with PFS r 6 months survived for a median of only 7.8 months, compared with 32.7 months for those whose disease progressed more slowly. These differences suggest that the clinical factors we identified could be useful indicators of prognosis.
The major limitations of this study were the small population of 57 patients and the high number of patients categorized as having a favorable prognosis (0 risk factors) for OS vs. other categories. Additionally, because this was a retrospective analysis, not all patient data were available. However, our prognostic model differs from others in that we included predictions for both PFS and OS as well as both predictive factors and risk groups, making this study, to our knowledge, the most complete prognostic model to date for advanced RCC. In our institution, we will conduct a large, 164-patient study of pathway-specific agents for the treatment of RCC using quantitative proteomics for precise patient selection. We plan to use the prognostic models developed in the present study in that cohort in the hope of validating the results.
Table 4 Multivariate analysis of risk factors associated with progression-free survival Factor, level
Parameter estimate ⫾ SE
Hazard ratio
95% confidence interval
Received prior radiation therapy Absence of lung metastases Lymphocytes o 25 cells/ml
0.53 ⫾ 0.32 0.85 ⫾ 0.32 0.54 ⫾ 0.33
1.69 2.35 1.72
0.90–3.19 1.24–4.43 0.91–3.26
Note: P values were determined by chi-square test. Boldface values indicate statistically significant results.
P value 0.103 0.008 0.119
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
Fig. 4. Overall survival probability per risk group. (Color version of the figure is available online.)
Fig. 5. Progression-free survival probability per risk group. (Color version of the figure is available online.)
9
10
R.J. Amato et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–10
References [1] Elfiky AA, Aziz SA, Conrad PJ, et al. Characterization and targeting of phosphatidylinositol-e kinase (PI3K) and mammalian target of rapamycin (mTOR) in renal cell cancer. J Transl Med 2011;9:133–43. [2] Escudier B, Thompson JA. Mechanism of action of everolimus in renal cell carcinoma. Med Oncol 2009;26:32–9. [3] Pantuck AJ, Seligson DB, Klatte T, et al. Prognostic relevance of the mTOR pathway in renal cell carcinoma: implications for molecular patient selection for targeted therapy. Cancer 2007;109:2258–65. [4] Schultz L, Chaux A, Albadine R, et al. Immunoexpression status and prognostic value of mTOR and hypoxia-induced pathway members in promary and metastatic clear cell renal cell carcinoma. Am J Surg Pathol 2011;35:1549–56. [5] Motzer RJ, Mazumdar M, Bacik J, Berg W, Amsterdam A, Ferrara J. Survival and prognostic stratification of 670 patients with advanced renal cell carcinoma. J Clin Oncol 1999;17:2530–40. [6] Motzer RJ, Bacik J, Schwartz LH, Reuter V, Russo P, Marion S, Mazumdar M. Prognostic factors for survival in previously treated patients with metastatic renal cell carcinoma. J Clin Oncol 2004;22:454–63. [7] Bukowski RM, Negrier S, Elson P. Prognostic factors in patients with advanced renal cell carcinoma: development of an International Kidney Cancer Working Group. Clin Cancer Res 2004;10:6310S–6314SS. [8] Stein A, Bellmunt J, Escudier B, Kim D, Stergiopoulos SG, Mietlowski W, et al. RECORD-1 Trial Study Group: Survival prediciton in everolimus-treated patients with metastatic renal cell carcinoma incorporating tumor burden response in the RECORD-1
[9]
[10]
[11] [12]
[13]
[14]
[15]
[16]
trial. Eur Urol 2012 [pii: S0302-2838(12)01407-8. doi: 10.1016/j. eururo.2012.11.032; Epub ahead of print]. Amato RJ, Jac J, Giessinger S, Saxena S, Willis JP. A phase 2 study with a daily regimen of the oral mTOR inhibitor RAD001 (everolimus) in patients with metastatic clear cell renal cell cancer. Cancer 2009;115:2438–46. Pencina MJ, D'Agostino RB. Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004;23:2109–23. Liu L, Forman S, Barton B. Fitting Cox model using PROC PHREG and beyond in SAS. SAS Global Forum; 2009. Choueiri TK, Garcia JA, Elson P, et al. Clinical factors associated with outcome in patients with metastatic clear-cell renal cell carcinoma treated with vascular endothelial growth factor-targeted therapy. Cancer 2007;110:543–50. Motzer RJ, Bukowski RM, Figlin RA, et al. Prognostic nomogram for sunitinib in patients with metastatic renal cell carcinoma. Cancer 2008;113:1552–8. Heng DYC, Xie W, Regan MM, et al. Prognostic factors for overall survival in patients with metastatic renal cell carcinoma treated wtih vascular endothelial growth factor-targeted agents: results from a large, multicenter study. J Clin Oncol 2009;27:5794–9. Manola J, Royston P, Elson P, et al. Prognostic model for survival in patients with metastatic renal cell carcinoma: results from the International Kidney Cancer Working Group. Clin Cancer Res 2011;17:5443–50. Mozter RJ, Escudier B, Oudard S, et al. Phase 3 trial of everolimus for metastatic renal cell carcinoma. Cancer 2010;116:4256–65.