Original research  
A simple prognostic model for overall survival in metastatic renal  
cell carcinoma  
1
,2  
3
4
5
Hazem I. Assi, MD; Francois Patenaude, MD; Ethan Toumishey, BSc; Laura Ross, BSc;  
5
2,6,7  
Mahmoud Abdelsalam, MD; Tony Reiman, MD  
1
2
Department of Internal Medicine, Faculty of Medicine, American University of Beirut Medical Centre, American University of Beirut, Lebanon; Department of Medicine, Dalhousie University, Halifax, NS,  
3
4
Canada; Segal Cancer Centre, Jewish General Hospital, Department of Oncology and Department of Medicine, Hematology Division, Montreal, QC, Canada; Dalhousie Medicine New Brunswick, Saint  
John, NB, Canada; Division of Medical Oncology, The Moncton Hospital, Moncton, NB, Canada; Department of Oncology, Saint John Regional Hospital, Saint John, NB, Canada; Department of Biology,  
University of New Brunswick, Fredericton, NB, Canada  
5
6
7
Cite as: Can Urol Assoc J 2016;10(3-4):113-9. http://dx.doi.org/10.5489/cuaj.3351  
tor receptor 2 (VEGF-2). These novel targeted therapies have  
transformed the management of metastatic RCC (mRCC).  
4
,5  
Abstract  
More than 80% of patients achieve clinical benefit in the  
form of objective response to treatment or disease stabiliza-  
tion with tyrosine kinase inhibitors. Additionally, median  
overall survival with the targeted therapies is now greater  
than two years, which is more than double the overall sur-  
Introduction: The primary purpose of this study was to develop  
a simpler prognostic model to predict overall survival for patients  
treated for metastatic renal cell carcinoma (mRCC) by examining  
variables shown in the literature to be associated with survival.  
Methods: We conducted a retrospective analysis of patients treat-  
ed for mRCC at two Canadian centres. All patients who started  
first-line treatment were included in the analysis. A multivari -  
ate Cox proportional hazards regression model was constructed  
using a stepwise procedure. Patients were assigned to risk groups  
depending on how many of the three risk factors from the final  
multivariate model they had.  
6
vival seen in the interferon-α era.  
Prognostic models that can be used to guide clinical trial  
design, patient counseling, and treatment decisions have  
7
been developed. The Memorial Sloan-Kettering Cancer  
8
Centre (MSKCC) model was first published in 1999 and  
remains the standard against which subsequent models for  
9
advanced RCC have been assessed. The authors of the  
Results: There were three risk factors in the final multivariate  
model: hemoglobin, prior nephrectomy, and time from diagno-  
sis to treatment. Patients in the high-risk group (two or three risk  
factors) had a median survival of 5.9 months, while those in the  
intermediate-risk group (one risk factor) had a median survival of  
model identified Karnofsky performance status, serum lac-  
tate dehydrogenase, hemoglobin, corrected serum calcium,  
and prior nephrectomy (later replaced with time from diag-  
10  
nosis to treatment) as pre-treatment factors predictive of  
survival and used these factors to categorize patients into  
three different risk groups. The MSKCC model has since  
been validated in the era of vascular endothelial growth fac-  
1
6.2 months, and those in the low-risk group (no risk factors) had  
a median survival of 50.6 months.  
Conclusions: In multivariate analysis, shorter survival times were  
associated with hemoglobin below the lower limit of normal,  
absence of prior nephrectomy, and initiation of treatment within  
one year of diagnosis.  
1
1
tor (VEGF)-targeted therapies. Another widely used model  
was developed by the International Metastatic Renal Cell  
1
2
Carcinoma Database Consortium in 2009, and it has also  
1
1
been externally validated. This model, often referred to as  
the Heng model, includes four of the five prognostic factors  
from the MSKCC model (hemoglobin, corrected serum cal-  
cium, Karnofsky performance status, and time from diagnosis  
to treatment), along with two additional ones: neutrophil  
Introduction  
12  
Renal cell carcinoma (RCC) is an aggressive disease, recur-  
ring in up to 40% of patients who are initially treated for  
count and platelet count.  
A high neutrophil to lymphocyte ratio (NLR), an index of  
systemic inflammation, has recently been found in multivari-  
ate analyses to be an independent factor for both progres-  
1
a localized tumour. About one-third of patients with RCC  
2
,3  
have metastatic disease at diagnosis. Advances in our  
understanding of the biology of RCC and particularly the  
role of angiogenesis in the progress of the clear cell subtype  
have led to the development of oral tyrosine kinase inhibitors  
with activity mostly against vascular endothelial growth fac-  
1
3-18  
sion-free survival and overall survival.  
The primary purpose of this study was to develop a sim-  
pler prognostic model to predict overall survival for patients  
who are treated for mRCC by examining variables shown  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
2016 Canadian Urological Association  
113  
©
aꢀꢀꢁ ꢂt ꢃꢄ.  
in the literature8,12 to be associated with survival, including  
NLR. Secondary aims were to compare our model with the  
MSKCC and Heng models and to analyze survival in the  
subset of patients on first-line sunitinib therapy.  
were drawn with replacement from the observed data. The  
model building strategy described above was conducted on  
all 500 bootstrapped data sets. Finally, the frequency with  
which each predictor variable appeared in the final model  
from all 500 samples was counted. Variables appearing in  
>
50% of the models were retained. In the next step, an  
Methods  
additional 500 bootstrap samples were obtained. With each  
sample, a Cox proportional hazards model was fit using the  
retained variables from the first step. Using the results from  
the 500 estimated models, mean parameter estimates, haz-  
ard ratios, and confidence intervals (CIs) were calculated.  
A risk group variable was created by summing the num-  
ber of risk factors each patient had from the final model.  
The area under the receiver operating characteristic curve  
(ROC) was used to determine the predictive accuracy of the  
risk group variable. Additional risk group variables were  
Patients  
Retrospective data were obtained from the medical records  
of all patients diagnosed with mRCC at two Canadian cen-  
tres, from July 2007 until December 2011. The inclusion  
9
criteria of Heng and colleagues were used.  
We recorded basic demographic, survival, and clinical  
data, including all variables that have been shown in the  
literature to be important prognostic factors in mRCC. We  
8
,12  
calculated using prognostic models from the literature.  
1
9
used an NLR categorization from the literature (≤3 vs. >3).  
The predictive accuracy of the risk group variables from dif-  
ferent prognostic models was assessed with the area under  
the ROC. Kaplan Meier curves for each of the different risk  
group variable were calculated.  
We retrospectively reviewed data from 120 patients, and  
included the 89 patients who received at least one cycle of  
active treatment; 31 patients did not receive treatment and  
were excluded from the analysis. Approval for this study  
was obtained from the Horizon Health Network Research  
Ethics Board.  
All analyses were conducted using R version 3.0.1.  
Results  
Statistical analysis  
Patient characteristics, treatment, and survival  
All patients who started first-line treatment, mostly with  
sunitinib, were included in the analysis. The main outcome,  
overall survival, was defined as time from treatment initia-  
tion until death, otherwise censored at last their last followup  
or contact. The survival distribution and median survival  
were assessed via Kaplan Meier estimates. Univariate asso-  
ciations between overall survival and baseline demographic  
and clinical factors were examined. Significance was taken  
at p<0.05. Log-rank tests were used to test the presence of  
a significant difference between survival among categor-  
ical variables and overall survival. For prognostic purposes,  
during the model-building phase all continuous variables  
were dichotomized at the upper or lower levels of normal  
except for age, which was dichotomized at >65 years and  
Baseline characteristics are provided in Table 1. Eighty-nine  
patients were treated for mRCC during the study period.  
First-line treatments included sunitinib, being the most com-  
mon (n=71, 79.8%), followed by interferon alpha (n=14,  
15.7%), pazopanib (n=2, 2.2%), sorafenib (n=1, 1.1%), and  
temsirolimus (n=1, 1.1%). Mean patient age was 63.1 years  
(standard deviation [SD] 9.9 years, range 38‒88). Thirty-  
eight out of the 89 patients (42.7%) were metastatic at diag-  
nosis. There were 17 patients who did not undergo nephrec-  
tomy. Of these 17 patients, seven were due to comorbidities;  
the remaining 10 patients did not have nephrectomy for  
unknown reasons.  
At a median followup of 24.6 months (95% CI 19.2, 39.7)  
for the entire cohort, the median overall survival was 20.9  
months (95% CI 14.2–50.6) (Fig. 1). By the end of the study  
period, 44 patients (49.4%) had died. One-year survival was  
63.7% (95% CI 0.54–0.76).  
≤
65 years.  
A multivariate Cox proportional hazards regression model  
was constructed using a stepwise procedure to identify the  
most significant variables affecting the disease-free sur -  
vival. The stepwise algorithm used the Aikake Information  
Criterion (AIC), a penalized likelihood measure, to choose a  
final model. When comparing two models, the model with a  
lower AIC value more closely resembles reality. The propor-  
tional hazards assumption of the final model was examined  
with a global test of proportionality. The test examines cor-  
relations between model residuals and log (time).  
Univariate analysis  
Factors that were significantly associated with poor overall  
survival were age >65 years, absence of prior nephrectomy,  
non-clear-cell histology, presence of two or more metastatic  
sites, presence of brain metastases, time interval from diag-  
nosis to treatment of <1 year, and hemoglobin below the  
lower limit of normal (Table 2).  
The internal validity of the final model was examined with  
a two-step bootstrap procedure. In the first step, 500 samples  
1
14  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
Pꢅoꢆꢇoꢀtꢁꢈ modꢂꢄ foꢅ Os ꢁꢇ mrcc  
Table 1. Participant baseline characteristics  
Categorical variables  
n
%
No. of patients  
89  
Gender  
Female  
26  
63  
29.2  
70.8  
Male  
Province  
New Brunswick  
80  
8
89.9  
9.0  
Nova Scotia  
Prince Edward Island  
1
1.1  
Treatment  
Sunitinib  
71  
18  
79.8  
20.2  
Other  
Nephrectomy  
No  
17  
72  
19.1  
80.9  
Yes  
Fig. 1. Overall survival (with 95% confidence limits) of patients in this study.  
Vertical lines indicate last followup.  
Histology  
Clear-cell  
72  
17  
80.9  
19.1  
Non-clear-cell  
with worse survival. Serum lactate dehydrogenase did not  
improve the fit of the model in multivariate analysis and  
was, therefore, not included in the final model.  
Patients were assigned to risk groups depending on how  
many of these three risk factors they had. Patients with no risk  
factors were assigned to the favourable-risk group. Patients  
with one risk factor were assigned to the intermediate-risk  
group, and patients with two or three risk factors were  
assigned to the high-risk group. A survival plot based on  
these risk groups is shown in Fig 2. There were 26 patients  
Radiation therapy  
No  
52  
33  
61.2  
38.8  
Yes  
Number of metastatic sites*  
1
41  
35  
12  
46.6  
39.8  
13.6  
2
3
Site of metastatic disease  
Lung  
63  
27  
70.8  
30.3  
13.5  
15.7  
24.7  
10.1  
SD  
Node  
(
29.2%) in the favourable-risk group, 31 patients (34.8%)  
Liver  
12  
in the intermediate-risk group, and 32 patients (36%) in the  
high-risk group.  
Patients in the high-risk group had a median survival of  
Renal bed  
14  
Bone  
22  
Brain  
9
5.9 months (95% CI 5.9–17.3), while those in the intermedi-  
Continuous variables  
Age  
Mean  
63.1  
125.8  
2.6  
ate-risk group had a median survival of 16.2 months (95%  
CI 11.2–NA), and those in the low-risk group had a median  
survival of 50.6 months (95% CI 49.3–NA) (Fig. 2).  
9.9  
Hemoglobin (g/L)  
Corrected calcium (mmol/L)  
Lactate dehydrogenase (U/L)  
Alkaline phosphatase (U/L)  
Neutrophil-to-lymphocyte ratio  
Absolute neutrophil count  
Platelets  
20.2  
0.2  
250.1  
119.9.  
5.3  
284.2  
108.3  
5.5  
Comparison between our new model and the MKSCC model  
The MKSCC model classifies patients into three risk categor-  
ies according to their number of risk factors: favourable-risk  
5.80  
297.4  
90.5  
2.5  
169.6  
10.1  
(
no risk factors), intermediate-risk (one or two risk factors),  
Karnofsky performance status score  
and poor (three, four, or five risk factors). Two of our patients  
were classified in the favourable group. Grouping these two  
patients with the intermediate-risk patients would result in  
*Information on number of metastatic sites was missing for one patient. The percentages  
in the second column for number of metastatic sites were calculated only for the group of  
patients for whom data on number of metastatic sites were available.  
4
2 of our patients (47.2%) being classified in the intermedi-  
Multivariate modelling and risk stratification in a new prognostic model  
ate group and 47 (52.8%) in the poor group. There was no  
difference in survival between the two groups.  
There were three risk factors in the final multivariate model  
(
Table 3): hemoglobin, prior nephrectomy, and time from  
Comparison between our new model and the Heng model  
diagnosis to treatment. Hemoglobin below the lower limit  
of normal, absence of prior nephrectomy, and having treat-  
ment begin within one year of diagnosis were all associated  
The Heng model stratifies patients into three risk categories  
according to their number of risk factors: favourable-risk (no  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
115  
aꢀꢀꢁ ꢂt ꢃꢄ.  
Table 2. Univariate analysis of categorical variables and  
overall survival  
Table 2. Univariate analysis of categorical variables and  
overall survival (cont’d)  
Overall  
Overall  
Variable  
%
% dead  
survival  
p value  
Variable  
%
% dead  
survival  
p value  
(median)  
(median)  
Gender  
Male  
Brain  
70.8  
29.2  
44.4  
61.5  
3.2  
1.5  
0.304  
No  
89.8  
10.2  
46.8  
77.8  
2.29  
0.88  
0.003  
Female  
Age  
Yes  
Time to treatment  
≤1 year  
<0.001  
≤65 years  
56.2  
43.8  
46.0  
53.8  
4.11  
1.18  
0.042  
<0.001  
0.693  
0.228  
0.035  
0.045  
60.5  
39.5  
59.6  
38.2  
0.92  
4.11  
>65 years  
>1 year  
Nephrectomy  
Hemoglobin  
<LLN  
<0.001  
0.839  
0.361  
0.153  
No  
19.1  
80.9  
76.5  
43.1  
0.35  
3.19  
42.0  
58.0  
67.6  
37.3  
0.75  
4.11  
Yes  
≥LLN  
Radiation  
No  
Corrected calcium  
>ULN  
61.2  
38.8  
48.1  
51.5  
1.92  
1.50  
12.1  
87.9  
37.5  
51.7  
NA  
Yes  
≤ULN  
1.92  
Treatment  
Sunitinib  
Other  
Lactate dehydrogenase  
>ULN  
79.8  
20.2  
43.7  
72.2  
2.29  
1.18  
29.3  
70.7  
50.0  
48.3  
1.50  
2.29  
≤ULN  
Histology  
Clear-cell  
Non-clear-cell  
No. of metastatic sites  
1
Alkaline phosphatase  
>ULN  
80.9  
19.1  
48.6  
52.9  
1.92  
0.58  
18.3  
81.7  
53.3  
47.7  
2.29  
1.74  
≤ULN  
Neutrophil-to-  
lymphocyte ratio  
0
.014  
.429  
46.6  
53.4  
43.9  
55.3  
3.19  
1.18  
>ULN  
24.4  
75.6  
61.9  
46.2  
0.46  
2.29  
≥2  
≤ULN  
Metastatic sites  
Absolute neutrophil  
count  
Lung  
No  
0
28.4  
71.6  
28.0  
58.7  
NA  
0.051  
0.755  
0.691  
0.547  
0.889  
>
ULN  
ULN  
Platelets  
ULN  
ULN  
14.0  
86.0  
71.4  
53.5  
0.35  
1.35  
Yes  
1.35  
≤
Node  
No  
0.483  
0.342  
70.5  
29.5  
51.6  
46.2  
1.90  
1.40  
>
11.8  
88.2  
50.0  
57.8  
NA  
Yes  
≤
1.26  
Liver  
No  
Karnofsky  
performance status  
86.4  
13.6  
52.6  
33.3  
1.50  
NA  
Yes  
>80  
70.3  
29.7  
50.0  
54.5  
1.74  
1.14  
Renal bed  
No  
≤80  
84.1  
15.9  
50.0  
50.0  
1.50  
4.11  
Note: Signiꢀcant p values are bolded. LLN: lower limit of normal; ULN: upper limit of  
normal; NA: not applicable.  
Yes  
Bone  
No  
Fig. 2 also presents a survival plot for patients in our  
study, using the three risk categories of the Heng model.  
Twenty-one of our patients (23.6%) were in the favourable-  
risk group according to the Heng model, 61 (68.5%) were  
in the intermediate-risk group, and seven patients (7.9%)  
were in the high-risk group. The log-rank test showed a  
statistically significant difference between survival curves  
(chi-square=9.8, degrees of freedom=2, p=0.007). Median  
survival was 49.3 months for the favourable-risk group, 14.2  
months for the intermediate-risk group, and 7.4 months for  
the high-risk group.  
75.0  
25.0  
53.0  
40.9  
1.74  
1.35  
Yes  
Note: Signiꢀcant p values are bolded. LLN: lower limit of normal; ULN: upper limit of  
normal; NA: not applicable.  
risk factors), intermediate-risk (one or two risk factors), and  
high-risk (more than two risk factors).  
1
2
All 21 patients classified in the favourable-risk category  
according to the Heng model (Table 4) were also in the  
favourable-risk group in our model. One of the patients in  
the intermediate-risk group in our model was classified in  
the high-risk group of the Heng model. Of the 32 patients  
classified in our high-risk group, 26 were classified in the  
Heng model’s intermediate-risk group and six were classified  
in the Heng model’s high-risk group.  
Two of the three risk factors in our model (hemoglobin  
and time from diagnosis to treatment) were also in the Heng  
model. Both the Heng model and our model were able  
to distinguish survival curves between the risk groups they  
1
16  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
Pꢅoꢆꢇoꢀtꢁꢈ modꢂꢄ foꢅ Os ꢁꢇ mrcc  
Fig. 2. Side-by-side comparison of survival of patients in our study, stratified according to the three risk groups developed in our model (left) and the Heng model  
(
right). Vertical lines indicate last followup.  
produced. Our more parsimonious model was slightly better  
at predicting survival, as determined by the C-index. The  
bootstrap corrected C-index was 0.635 for the Heng model  
and 0.761 for our model.  
All 21 patients classified in the favourable-risk category  
according to the Heng model were also in the favourable-  
risk group in our model. One of the patients in the inter-  
mediate-risk group in the Heng model were classified in  
the high-risk group of our model and 17 were classified  
in our favourable-risk group. Five of the patients classified  
in our high-risk group were classified in the Heng model’s  
intermediate-risk group.  
survival was 67.3% (95% CI 56.3–80.5), whereas two-year  
survival was 53.1% (95% CI 40.7–69.2). Survival up to one  
year was predicted by the same variables as in our univari-  
ate analysis above: NLR, hemoglobin, prior nephrectomy,  
and time from treatment to diagnosis. This is not surprising,  
given the proportional hazards assumption. Median follo-  
wup for patients receiving sunitinib was 21 months (95%  
CI 18.1–30.4); it was 24.6 months (95% CI 19.2, 39.7) for  
the entire cohort.  
Discussion  
In our multivariate analysis, shorter survival times were asso-  
ciated with hemoglobin below the lower limit of normal,  
absence of prior nephrectomy, and initiation of treatment  
within one year of diagnosis.  
Sub-analysis of patients treated with sunitinib  
Of the 89 patients in the study, 71 were treated with sunitinib  
(
3
Table 1). Overall survival of these patients is shown in Fig.  
. Median survival for this group was 2.29 years. One-year  
The NLR has been shown elsewhere to be an independent  
predictor of survival in patients with mRCC  
1
3, 14  
and, to our  
Table 3. Results of multivariate survival regression and risk group regression  
Multivariate Cox proportional hazards regression  
Cox proportional hazards regression  
95% CI  
Predictor  
HR  
LB  
UB  
Risk groups  
HR  
Prior nephrectomy  
0.26  
0.36  
0.50  
(0.14,  
(0.15,  
(0.23,  
0.60)  
0.52)  
1.10)  
Intermediate vs. favourable  
High vs. favourable  
2.70  
7.99  
Hemoglobin <LLN  
Time from diagnosis to treatment <1 year  
CI: conꢀdence interval; HR: hazard ratio; LB, lower bound; LLN: lower limit of normal; UB: upper bound.  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
117  
aꢀꢀꢁ ꢂt ꢃꢄ.  
lation-based study, the introduction of first-line sunitinib was  
associated with a doubling of overall survival compared  
2
3
with patients treated with interferon alone and it has a  
24  
manageable safety profile. We were particularly interested  
in prognostic factors for patients receiving sunitinib. Survival  
up to one year was predicted by the same variables as in our  
univariate analysis for the entire study population: hemoglo-  
bin, prior nephrectomy, and time from treatment to diagno-  
sis. Barnias etal produced a model for patients treated with  
sunitinib with three prognostic factors: time from diagnosis  
to initiation of treatment (as in our analysis), number of  
6
metastatic sites, and performance status.  
There are limitations to this study, including its retro -  
spective design and the relatively small number of patients  
included in the analysis. There were missing data for some  
of the prognostic variables in the patient charts, which may  
have biased our results.  
Fig. 3. Survival of the 71 patients (with 95% confidence limits) treated with  
sunitinib. Vertical lines indicate last followup.  
Our model is simpler and could be validated in a large  
data bank registry, such as the International Metastatic  
Renal-Cell Carcinoma Database Consortium or the Canadian  
Kidney Cancer Information System.  
knowledge, it has not yet been considered in any prognostic  
models. In contrast to recently published articles, our uni-  
variate analysis did not demonstrate the independent prog-  
1
3-18  
nostic value of NLR;  
perhaps this is because only 24%  
of our study population had an elevated NLR. This finding  
should be confirmed with a much larger cohort of patients.  
Our model was slightly better at predicting survival  
Conclusion  
Prognostic models using clinical and laboratory-based vari-  
ables remain the primary tool for predicting outcomes in  
mRCC. Our study adds a new set of real-world data to the  
international efforts to develop better prognostic models.  
(
C-index 0.761) than the Heng model (0.631). Two of the  
three risk factors in our model (hemoglobin and time from  
1
2
diagnosis to treatment) were in the Heng model, as well  
1
0
as the 2002 version of the MKSCC model. Our third risk  
factor, prior nephrectomy, has been shown to be independ-  
ently associated with overall survival in patients receiving  
Competing interests: Dr. Assi has been an Advisory Board member for Pꢀzer. Dr. Patenaude has  
received grants/honoraria from BMS, Novartis, Pꢀzer, and Roche; and has participated in clinical  
trials for BMS, Merck, Novartis, Pꢀzer, and Roche. Dr. Toumishey has participated in clinical trials for  
Celgene. Ms. Ross has participated in numerous clinical trials. Dr. Abdelsalam has been an Advisory  
Board member for Amgen, Astelllas, BI, Celgene, Eli Lily, Innomar Strategies, Janssen, Johnson &  
Johnson, Merck, Novartis, and Sanoꢀ; has received grants/honoraria from Amgen, Astellas, BI,  
Eli Lily, Janssen, Johnson & Johnson, and Roche; and has participated in clinical trials for Amgen,  
Aragon, Astra Zeneca, BI, BMI, Exelixis, GSK, Merck, Merrimack, NCIC, Novartis, QC Clinical Research  
Organization in Cancer, Quintiles, Roche, and Serono. Dr. Reiman has received grants/honoraria  
from Celgene and Roche; and has participated in clinical trials for AstraZeneca, Celgene, Eli Lilly,  
Genentech, GSK, Takeda, and Roche.  
2
0
targeted therapy. As Tagawa points out, however, most  
patients in prospective clinical trials previously underwent  
nephrectomy, and thus it is difficult to draw conclusions  
about this factor even though retrospective studies such as  
2
1
ours suggest a benefit. An ongoing prospective clinical  
trial, the CARMENA trial, is evaluating the value of upfront  
nephrectomy followed by sunitinib vs. sunitinib alone with-  
out nephrectomy in metastatic clear-cell RCC.22  
Based on multiple prospective phase 3 trials, first-line  
oral therapy with tyrosine kinase inhibitors directed against  
VEGF signaling has become the standard of care for most  
4
,5  
patients with mRCC with clear-cell histology. In a popu-  
Table 4. Comparison of our model with the Heng model  
Our model  
Heng model  
No. of patients (%)  
26 (29.2)  
Median  
50.6  
95% CI  
49.3–NA  
11.1–NA  
3.1–17.3  
No. of patients (%)  
21 (23.6)  
Median  
49.3  
95% CI  
49.3–NA  
10.5–27.5  
3.6–NA  
Favourable-risk group  
Intermediate-risk group  
High-risk group  
31 (34.8)  
16.2  
61 (68.5)  
14.2  
32 (36.0)  
5.9  
7 (7.9)  
7.4  
Log-rank test  
Chi square=30.9, df=2, p<0.001  
0.761  
Chi square=9.8, df=2, p=0.007  
0.635  
C index  
CI: conꢀdence interval; df: degrees of freedom.  
1
18  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
Pꢅoꢆꢇoꢀtꢁꢈ modꢂꢄ foꢅ Os ꢁꢇ mrcc  
Acknowledgments: We thank Tracey Allen for data collection and the Community Health and  
Education Endowment Fund provided by The Friends of The Moncton Hospital. Dr. Reiman is the  
Canadian Cancer Society Research Chair at the University of New Brunswick.  
12. Heng DYC, Xie W, Regan MM, et al. Prognostic factors for overall survival in patients with metastatic  
renal cell carcinoma treated with vascular endothelial growth factor-targeted agents: Results from a large,  
multicentre study. J Clin Oncol 2009;27:5794-99. http://dx.doi.org/10.1200/JCO.2008.21.4809  
1
3. Cetin B, Berk V, Kaplan MA, et al. Is the pretreatment neutrophil to lymphocyte ratio an important prog-  
nostic parameter in patients with metastatic renal cell carcinoma? Clin Genitourin Cancer 2013;11:141-8.  
http://dx.doi.org/10.1016/j.clgc.2012.09.001  
This paper has been peer-reviewed.  
1
4. Keizman D, Ish-Shalom M, Huang P, et al. The association of pre-treatment neutrophil to lymphocyte  
ratio with response rate, progression free survival and overall survival of patients treated with sunitinib  
for metastatic renal cell carcinoma. Eur J Cancer 2012;48:202-8. http://dx.doi.org/10.1016/j.  
ejca.2011.09.001  
1
1
5. Pichler M, Hutterer GC, Stoeckigt C, et al. Validation of the pre-treatment neutrophil-lymphocyte ratio as a  
prognostic factor in a large European cohort of renal cell carcinoma patients. Br J Cancer 2013;108:901-7.  
http://dx.doi.org/10.1038/bjc.2013.28  
6. Cetin B, Berk V, Kaplan MA, et al. Is the pretreatment neutrophil to lymphocyte ratio an important  
prognostic parameter in patients with metastatic renal cell carcinoma? Clin Genitourin Cancer 2013;11:  
References  
1
2
.
.
Chin AI, Lam JS, Figlin RA, et al. Surveillance strategies for renal cell carcinoma patients following  
nephrectomy. Rev Urol 2006;8:1-7.  
Choueiri TK. Factors associated with outcome in patients with advanced renal cell carcinoma in the  
era of antiangionenic agents. Clin Genitourin Cancer 2008;9:15-20. http://dx.doi.org/10.3816/  
CGC.2008.n.002  
1
41-8. http://dx.doi.org/10.1016/j.clgc.2012.09.001  
1
1
7. Templeton AJ, Amir E, Aneja P, et al. Neutrophil to lymphocyte ratio and response to tyrosine kinase  
inhibitor therapy in metastatic renal cell carcinoma. J Clin Oncol 2014;32: abstr 477.  
3
4
.
.
Molina AM, Motzer RJ, Heng DY. Systemic treatment options for untreated patients with metastatic clear cell  
renal cancer. Semin Oncol 2013;40:436-43. http://dx.doi.org/10.1053/j.seminoncol.2013.05.013  
Pal SK, Nelson RA, Vogelzang N. Disease-speciꢀc survival in de novo metastatic renal cell carcinoma  
in the cytokine and targeted therapy era. PLoS One 2013;8:e63341. http://dx.doi.org/10.1371/  
journal.pone.0063341  
8. Ohno Y, Nakashima J, Ohori M, et al. Clinical variables for predicting metastatic renal cell carcinoma  
patients who might not beneꢀt from cytoreductive nephrectomy: Neutrophil-to-lymphocyte ratio and  
performance status. Int J Clin Oncol 2014;19:139-5. http://dx.doi.org/10.1007/s10147-012-0514-5  
9. Chen TM, Lin CC, Huang PT, et al. Neutrophil-to-lymphocyte ratio associated with mortality in early  
hepatocellular carcinoma patients after radiofrequency ablation. J Gastroenterol Hepatol 2012;27:553-61.  
http://dx.doi.org/10.1111/j.1440-1746.2011.06910.x  
1
2
5
.
.
Powles T, Albers P. Management of favourable-risk patients with metastatic renal cell carcinoma: When  
to start and when to stop targeted therapy. Clin Genitourin Cancer 2012;20:213-8. http://dx.doi.  
org/10.1016/j.clgc.2012.06.002  
Harshman LC, Xie W, Bjarnason GA, et al. Conditional survival of patients with metastatic renal-cell  
carcinoma treated with VEGF-targeted therapy: A population-based study. Lancet Oncol 2012;13:927-35.  
http://dx.doi.org/10.1016/S1470-2045(12)70285-1  
0. Choueiri TK, Xie W, Kollmannsberger C, et al. The impact of cytoreductive nephrectomy on survival of  
patients with metastatic renal cell carcinoma receiving vasscular endothelial growth factor targeted therapy.  
J Urol 2011;185:60-6. http://dx.doi.org/10.1016/j.juro.2010.09.012  
6
2
2
2
2
1. Tagawa ST. Boning up on prognostic factors in advanced renal cell carcinoma. Ann Oncol 2011;22:751-  
3
http://dx.doi.org/10.1093/annonc/mdq705  
7
8
9
.
.
.
Sun M, Shariat S, Cheng C, et al. Prognostic factors and predictive models in renal cell carcinoma: A  
contemporary review. Eur Urol 2011;60:644-61. http://dx.doi.org/10.1016/j.eururo.2011.06.041  
Motzer RJ, Mazumdar M, Bacik J, et al. Survival and prognostic stratiꢀcation of 670 patients with advanced  
renal cell carcinoma. J Clin Oncol 1999;17:2530-40.  
Bamias A, Karadimou A, Lampaki S, et al. Prognostic stratiꢀcation of patients with advanced renal cell  
carcinoma treated with sunitinib: Comparison with the Memorial Sloan-Kettering prognostic factors model.  
BMC Cancer 2010;10:45. http://dx.doi.org/10.1186/1471-2407-10-45  
2. ClinicalTrials.gov. Clinical trial to assess the importance of nephrectomy (CARMENA). http://clinicaltrials.  
gov/ct2/show/NCT00930033. Accessed February 23, 2016.  
3. Heng DY, Chi KN, Murray N, et al. A population-based study evaluating the impact of sunitinib on overall  
survival. Cancer 2009;115:776-83. http://dx.doi.org/10.1002/cncr.24051  
4. Gore ME, Szczylik C, Porta C, et al. Safety and efꢀcacy of sunitinib for metastatic renal-cell carci -  
noma: An expanded-access trial. Lancet Oncol 2009;10:757-63. http://dx.doi.org/10.1016/S1470-  
2
045(09)70162-7  
1
1
0. Motzer RJ, Bacik J, Murphy BA, et al. Interferon-alfa as a comparative treatment for clinical trials of  
new therapies against advanced renal cell carcinoma. J Clin Oncol 2002;20:289-96. http://dx.doi.  
org/10.1200/JCO.20.1.289  
1. Heng DY, Xie W, Regan MM, et al. External validation and comparison with other models of the  
International Metastatic Renal-Cell Carcinoma Database Consortium prognostic model: A population-  
based study. Lancet Oncol 2013;14:141-8. http://dx.doi.org/10.1016/S1470-2045(12)70559-4  
Correspondence: Dr. Hazem Assi, American University of Beirut Medical Center, American University  
of Beirut, Beirut, Lebanon; hazemassi@gmail.com  
CUAJ • March-April 2016 • Volume 10, Issues 3-4  
119