Preoperative serum cholesterol is an independent prognostic factor for patients with renal cell carcinoma (RCC) Michela de Martino, Carmen V. Leitner, Christoph Seemann, Sebastian L. Hofbauer, Ilaria Lucca, Andrea Haitel*, Shahrokh F. Shariat and Tobias Klatte Department of Urology, Comprehensive Cancer Center of the Medical University of Vienna and *Clinical Institute of Pathology General Hospital Vienna, Medical University of Vienna – General Hospital, Vienna, Austria M.D.M and C.V.L. contributed equally to this work
Objective To assess the prognostic role of preoperative serum cholesterol in patients with renal cell carcinoma (RCC), as increasing evidence suggests that alterations in the lipid profile are associated with the development, progression and prognosis of various cancers.
Patients and Methods We analysed 867 patients, who underwent radical or partial nephrectomy for RCC between 2002 and 2012. Preoperative total cholesterol levels were determined in serum using colorimetric analysis (CHOD-PAP method). The association with cancer-specific survival (CSS) was assessed with Cox models. Discrimination was quantified with the C-index. The median follow-up was 52 months.
Results The median (interquartile range) serum cholesterol was 195 (166–232) mg/dL. Decreasing serum cholesterol was associated with more advanced T, N and M stages (P < 0.001), higher grades (P = 0.001) and presence of tumour necrosis (P =
Introduction RCC is the third most common malignancy of the urinary tract with an estimated 115 000 new cases and 31 000 cancer-related deaths annually in Europe [1]. Surgery is the mainstay of therapy for clinically localised disease, but 10–20% of patients will develop metastases after being considered disease free [2,3]. Although significant improvements have been made in systemic therapy of metastatic disease, prognosis remains poor and most patients ultimately succumb to their disease [4,5]. Evidence suggests that lipid metabolism is crucially involved in the biology of RCC. Preoperative surrogate markers, e.g. © 2014 The Authors BJU International © 2014 BJU International | doi:10.1111/bju.12767 Published by John Wiley & Sons Ltd. www.bjui.org
0.002). Continuously coded cholesterol was associated with CSS in both univariable (hazard ratio [HR] 0.87, P < 0.001) and multivariable analyses (HR 0.93, P = 0.001). The discrimination of a multivariable base model increased significantly from 88.3% to 89.2% following inclusion of cholesterol (P = 0.006). In patients with clinically localised disease (T1–3N0/+M0), cholesterol remained associated with CSS in multivariable analysis (HR 0.90, P = 0.002) and increased the discrimination from 74.6% to 76.9% (P = 0.002).
Conclusions Preoperative serum cholesterol is an independent prognostic factor for patients with RCC, with lower levels being associated with worse survival. Its use increases the discrimination of established prognostic factors. As cholesterol is a broadly available routine marker, its use may provide a meaningful adjunct in clinical practice. The biological rationale underlying this association remains to be clarified.
Keywords cholesterol, prognosis, survival, concordance, localised, regression
body mass index (BMI) and the nutritional status, have been linked with RCC stage and prognosis [6–8]. Serum cholesterol is another possible surrogate biomarker of lipid metabolism that is broadly available and routinely assessed in many patients. Although widely used to predict outcomes of patients with coronary heart disease, there are few studies that focus on its role as a prognostic marker in cancer [9,10]. Ohno et al. [11] recently suggested that higher preoperative blood levels of cholesterol may be associated with significantly better cancer-specific survival (CSS) in 364 patients with clear cell RCC. Moreover, in patients with metastatic RCC that were treated with temsirolimus, both higher baseline cholesterol and an increase in cholesterol
BJU Int 2014 wileyonlinelibrary.com
de Martino et al.
during the course of treatment were associated with improved overall survival [12]. The aim of the present study was to validate cholesterol as a prognostic biomarker in a large consecutive series of patients treated with surgery at a single institution.
Patients and Methods We retrospectively studied the prospective kidney cancer database of the Medical University of Vienna, Vienna, Austria, which had abstracted clinical, pathological, laboratory and follow-up data of 1203 consecutive patients, who were treated for a renal tumour between 2002 and 2012. Patients with benign renal tumours, malignant tumours other than RCC, unclassified or collecting duct RCC (n = 211), those that were not treated surgically (n = 20), with hereditary renal tumour syndromes or bilateral disease (n = 32), with known conditions or medications that interfere with serum cholesterol (thyroid diseases, nephrotic syndrome, chronic liver disease, statins for secondary prevention of coronary heart disease, n = 49), and with incomplete data on preoperative cholesterol levels (n = 24) were excluded. In all, 867 patients were finally analysed. The study was approved by the Institutional Review Board. Study Variables Clinical, pathological, follow-up data and preoperative serum cholesterol levels were extracted from the kidney cancer database. Clinical data included Eastern Cooperative Oncology Group Performance Status (ECOG PS), age, gender, symptoms, BMI, and presence of diabetes mellitus. A radical nephrectomy was performed in 561 patients (64.7%), while 306 (35.3%) underwent a partial nephrectomy. A concomitant lymph node dissection was carried out in 164 patients (18.9%). All surgical specimens were evaluated by one pathologist with specialisation in genitourinary pathology (A.H.). Tumour size was determined pathologically according to the largest diameter. Staging was performed according to the 2010 TNM classification. T stage was assigned pathologically and M stage clinically. N stage was assigned clinically (cN0 = pNx) or pathologically (pN0 or pN+). Patients with clinically positive retroperitoneal nodes (size >1 cm on imaging, enlarged or palpable nodes during surgery) had a node dissection. All N+ cases reported in this study had pathological confirmation. Preoperative clinical staging included CT of the abdomen and CT or X-ray of the chest. Further staging was performed if the patient was symptomatic or at surgeon preference. Nuclear grade was evaluated according to Fuhrman criteria. The postoperative surveillance protocol in non-metastatic disease followed guideline recommendations [13]. Blood for serum cholesterol measurement was taken by clean venipuncture at the time of admission for surgery within 1–7 © 2014 The Authors 2 BJU International © 2014 BJU International
days preoperatively. Blood was drawn into a serum Vacuette tube without anticoagulant (Greiner Bio-One). Total serum cholesterol was determined using the standard enzymatic colorimetric CHOD-PAP method (Roche Diagnostics). Statistical Analysis Serum cholesterol was first evaluated as a continuous variable and is presented as the median and interquartile range (IQR). Using the Shapiro–Wilk test, we rejected the hypothesis that serum cholesterol is normally distributed (z = 11.4, P < 0.001). Thus, Mann–Whitney U-tests and Kruskal–Wallis tests were used to assess differences in serum cholesterol between independent groups. Correlations with other continuously coded variables, e.g. tumour size, BMI, were assessed with Spearman’s rank correlation coefficients. The outcome measure of this study was CSS, which was calculated from the date of surgery to the date of death from RCC or last follow-up. Cholesterol was then coded as categorical variable, and the corresponding threshold was identified by the root node in a univariable recursive partitioning based survival tree analysis. Survival functions were estimated with the Kaplan–Meier method. Group differences in CSS of categorically coded cholesterol were determined with log-rank tests. Associations with CSS were further evaluated with univariable and multivariable Cox proportional hazards models. Cox models were stratified by presence of diabetes mellitus and gender. For multivariable Cox models, a backward variable selection approach was used that relied on Akaike’s information criterion. Discrimination of Cox models was assessed with Harrell’s concordance index. Discriminations of nested models were compared with likelihood ratio tests. Sub-analyses were performed according to the histological subtype and in patients with clinically localised disease (T1–3N0/+M0). All statistical testing was two-sided and statistical significance was defined as P < 0.05. Statistical analyses were performed with R 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria).
Results The final analysis population consisted of 867 patients with a median (IQR) age of 64 (54–72) years. The median (IQR) serum cholesterol was 195 (166–232) mg/dL. Patient and tumour characteristics are given in Table 1. Cholesterol and Clinical/Pathological Variables Associations and correlations are shown in Table 1. Serum cholesterol decreased with increasing T stage (P < 0.001) and nuclear grades (P = 0.001), and was lower in patients with
Serum cholesterol is an independent prognostic factor for patients with RCC
Table 1 Serum cholesterol levels in 867 patients with RCC and its (A) associations with categorically coded variables and (B) correlations with continuously coded variables. A Variable
Gender: Female Male Diabetes mellitus: No Yes ECOG PS: 0 1–2 Symptoms: No Yes T stage: pT1–2 pT3–4 N stage: pNx/N0 pN+ M stage: M0 M1 Grade: G1–2 G3–4 Tumour necrosis: No Yes Subtype: Clear cell Papillary Chromophobe
N (%)
Median (IQR) cholesterol, mg/dL
P
308 (35.5) 559 (64.5)
202 (175–238) 193 (163–229)
0.008
789 (91.0) 78 (9.0)
196 (167–234) 179 (155–216)
0.003
562 (64.8) 305 (35.2)
201 (176–238) 180 (148–215)
Objective To assess the prognostic role of preoperative serum cholesterol in patients with renal cell carcinoma (RCC), as increasing evidence suggests that alterations in the lipid profile are associated with the development, progression and prognosis of various cancers.
Patients and Methods We analysed 867 patients, who underwent radical or partial nephrectomy for RCC between 2002 and 2012. Preoperative total cholesterol levels were determined in serum using colorimetric analysis (CHOD-PAP method). The association with cancer-specific survival (CSS) was assessed with Cox models. Discrimination was quantified with the C-index. The median follow-up was 52 months.
Results The median (interquartile range) serum cholesterol was 195 (166–232) mg/dL. Decreasing serum cholesterol was associated with more advanced T, N and M stages (P < 0.001), higher grades (P = 0.001) and presence of tumour necrosis (P =
Introduction RCC is the third most common malignancy of the urinary tract with an estimated 115 000 new cases and 31 000 cancer-related deaths annually in Europe [1]. Surgery is the mainstay of therapy for clinically localised disease, but 10–20% of patients will develop metastases after being considered disease free [2,3]. Although significant improvements have been made in systemic therapy of metastatic disease, prognosis remains poor and most patients ultimately succumb to their disease [4,5]. Evidence suggests that lipid metabolism is crucially involved in the biology of RCC. Preoperative surrogate markers, e.g. © 2014 The Authors BJU International © 2014 BJU International | doi:10.1111/bju.12767 Published by John Wiley & Sons Ltd. www.bjui.org
0.002). Continuously coded cholesterol was associated with CSS in both univariable (hazard ratio [HR] 0.87, P < 0.001) and multivariable analyses (HR 0.93, P = 0.001). The discrimination of a multivariable base model increased significantly from 88.3% to 89.2% following inclusion of cholesterol (P = 0.006). In patients with clinically localised disease (T1–3N0/+M0), cholesterol remained associated with CSS in multivariable analysis (HR 0.90, P = 0.002) and increased the discrimination from 74.6% to 76.9% (P = 0.002).
Conclusions Preoperative serum cholesterol is an independent prognostic factor for patients with RCC, with lower levels being associated with worse survival. Its use increases the discrimination of established prognostic factors. As cholesterol is a broadly available routine marker, its use may provide a meaningful adjunct in clinical practice. The biological rationale underlying this association remains to be clarified.
Keywords cholesterol, prognosis, survival, concordance, localised, regression
body mass index (BMI) and the nutritional status, have been linked with RCC stage and prognosis [6–8]. Serum cholesterol is another possible surrogate biomarker of lipid metabolism that is broadly available and routinely assessed in many patients. Although widely used to predict outcomes of patients with coronary heart disease, there are few studies that focus on its role as a prognostic marker in cancer [9,10]. Ohno et al. [11] recently suggested that higher preoperative blood levels of cholesterol may be associated with significantly better cancer-specific survival (CSS) in 364 patients with clear cell RCC. Moreover, in patients with metastatic RCC that were treated with temsirolimus, both higher baseline cholesterol and an increase in cholesterol
BJU Int 2014 wileyonlinelibrary.com
de Martino et al.
during the course of treatment were associated with improved overall survival [12]. The aim of the present study was to validate cholesterol as a prognostic biomarker in a large consecutive series of patients treated with surgery at a single institution.
Patients and Methods We retrospectively studied the prospective kidney cancer database of the Medical University of Vienna, Vienna, Austria, which had abstracted clinical, pathological, laboratory and follow-up data of 1203 consecutive patients, who were treated for a renal tumour between 2002 and 2012. Patients with benign renal tumours, malignant tumours other than RCC, unclassified or collecting duct RCC (n = 211), those that were not treated surgically (n = 20), with hereditary renal tumour syndromes or bilateral disease (n = 32), with known conditions or medications that interfere with serum cholesterol (thyroid diseases, nephrotic syndrome, chronic liver disease, statins for secondary prevention of coronary heart disease, n = 49), and with incomplete data on preoperative cholesterol levels (n = 24) were excluded. In all, 867 patients were finally analysed. The study was approved by the Institutional Review Board. Study Variables Clinical, pathological, follow-up data and preoperative serum cholesterol levels were extracted from the kidney cancer database. Clinical data included Eastern Cooperative Oncology Group Performance Status (ECOG PS), age, gender, symptoms, BMI, and presence of diabetes mellitus. A radical nephrectomy was performed in 561 patients (64.7%), while 306 (35.3%) underwent a partial nephrectomy. A concomitant lymph node dissection was carried out in 164 patients (18.9%). All surgical specimens were evaluated by one pathologist with specialisation in genitourinary pathology (A.H.). Tumour size was determined pathologically according to the largest diameter. Staging was performed according to the 2010 TNM classification. T stage was assigned pathologically and M stage clinically. N stage was assigned clinically (cN0 = pNx) or pathologically (pN0 or pN+). Patients with clinically positive retroperitoneal nodes (size >1 cm on imaging, enlarged or palpable nodes during surgery) had a node dissection. All N+ cases reported in this study had pathological confirmation. Preoperative clinical staging included CT of the abdomen and CT or X-ray of the chest. Further staging was performed if the patient was symptomatic or at surgeon preference. Nuclear grade was evaluated according to Fuhrman criteria. The postoperative surveillance protocol in non-metastatic disease followed guideline recommendations [13]. Blood for serum cholesterol measurement was taken by clean venipuncture at the time of admission for surgery within 1–7 © 2014 The Authors 2 BJU International © 2014 BJU International
days preoperatively. Blood was drawn into a serum Vacuette tube without anticoagulant (Greiner Bio-One). Total serum cholesterol was determined using the standard enzymatic colorimetric CHOD-PAP method (Roche Diagnostics). Statistical Analysis Serum cholesterol was first evaluated as a continuous variable and is presented as the median and interquartile range (IQR). Using the Shapiro–Wilk test, we rejected the hypothesis that serum cholesterol is normally distributed (z = 11.4, P < 0.001). Thus, Mann–Whitney U-tests and Kruskal–Wallis tests were used to assess differences in serum cholesterol between independent groups. Correlations with other continuously coded variables, e.g. tumour size, BMI, were assessed with Spearman’s rank correlation coefficients. The outcome measure of this study was CSS, which was calculated from the date of surgery to the date of death from RCC or last follow-up. Cholesterol was then coded as categorical variable, and the corresponding threshold was identified by the root node in a univariable recursive partitioning based survival tree analysis. Survival functions were estimated with the Kaplan–Meier method. Group differences in CSS of categorically coded cholesterol were determined with log-rank tests. Associations with CSS were further evaluated with univariable and multivariable Cox proportional hazards models. Cox models were stratified by presence of diabetes mellitus and gender. For multivariable Cox models, a backward variable selection approach was used that relied on Akaike’s information criterion. Discrimination of Cox models was assessed with Harrell’s concordance index. Discriminations of nested models were compared with likelihood ratio tests. Sub-analyses were performed according to the histological subtype and in patients with clinically localised disease (T1–3N0/+M0). All statistical testing was two-sided and statistical significance was defined as P < 0.05. Statistical analyses were performed with R 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria).
Results The final analysis population consisted of 867 patients with a median (IQR) age of 64 (54–72) years. The median (IQR) serum cholesterol was 195 (166–232) mg/dL. Patient and tumour characteristics are given in Table 1. Cholesterol and Clinical/Pathological Variables Associations and correlations are shown in Table 1. Serum cholesterol decreased with increasing T stage (P < 0.001) and nuclear grades (P = 0.001), and was lower in patients with
Serum cholesterol is an independent prognostic factor for patients with RCC
Table 1 Serum cholesterol levels in 867 patients with RCC and its (A) associations with categorically coded variables and (B) correlations with continuously coded variables. A Variable
Gender: Female Male Diabetes mellitus: No Yes ECOG PS: 0 1–2 Symptoms: No Yes T stage: pT1–2 pT3–4 N stage: pNx/N0 pN+ M stage: M0 M1 Grade: G1–2 G3–4 Tumour necrosis: No Yes Subtype: Clear cell Papillary Chromophobe
N (%)
Median (IQR) cholesterol, mg/dL
P
308 (35.5) 559 (64.5)
202 (175–238) 193 (163–229)
0.008
789 (91.0) 78 (9.0)
196 (167–234) 179 (155–216)
0.003
562 (64.8) 305 (35.2)
201 (176–238) 180 (148–215)
