Premium
Impact of clinical variables on predicting disease‐free survival of patients with surgically resected renal cell carcinoma
Author(s) -
BrookmanAmissah Sabine,
Kendel Friederike,
Spivak Inna,
Pflanz Sandra,
Roigas Jan,
Klotz Theodor,
May Matthias
Publication year - 2009
Publication title -
bju international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.773
H-Index - 148
eISSN - 1464-410X
pISSN - 1464-4096
DOI - 10.1111/j.1464-410x.2008.08233.x
Subject(s) - medicine , renal cell carcinoma , hazard ratio , proportional hazards model , confidence interval , nephrectomy , platelet , gastroenterology , surgery , oncology , urology , kidney
OBJECTIVE To determine the value of particular clinical variables for the preoperative prognostic Cindolo formula (PPCF) to predict disease‐free survival (DFS) of patients with surgically treated renal cell carcinoma (RCC). PATIENTS AND METHODS In all, 771 consecutive patients (T1‐4NxM0) who had radical or partial nephrectomy were reviewed retrospectively. For each patient with RCC, PPCF was constructed according to clinical size and clinical presentation. On the basis of PPCF, patients were divided into Cindolo good prognosis (CGP) and Cindolo poor prognosis (CPP) groups. We also analysed further clinical variables (Eastern Cooperative Oncology Group score, American Society of Anesthesiologists score, body mass index, hepatic dysfunction, night sweat, fever, value of blood platelets, leukocytes, haemoglobin level, gender, age and location). DFS was estimated using the Kaplan‐Meier method. Univariable and multivariable Cox proportional hazard regression models were fitted to determine associations between the PPCF, measured clinical features, and DFS. RESULTS Four of the variables emerged as statistically significant for DFS from the univariable analysis ( P < 0.001), i.e. clinical presentation, clinical tumour size, haemoglobin level and blood platelet count. In the multivariable analysis, only clinical tumour size and blood platelet count remained significant for DFS. By contrast, clinical presentation, used in the PPCF, had no significant influence. According to the PPCF we developed the preoperative Amissah Prognosis Score (PAPS) calculated as (0.19 × clinical size) + (0.492 × platelet count (≤400/nL = 0, >400/nL = 1) with a threshold between the two resulting prognosis groups at 1.76. The multivariable hazard ratio (95% confidence interval, CI) for the PAPS was 2.98 (2.15–4.12) ( P < 0.001) compared to a hazard ratio for the PPCF of 1.36 (0.99–1.87) ( P = 0.061). Furthermore, the predictive ability was greater when using the PAPS (area under the curve 0.721; 95% CI, 0.680–0.763; P ≤ 0.001) than the PPCF (0.690; 0.647–0.734; P ≤ 0.001). CONCLUSIONS Using preoperative prognostic models is reasonable to provide patients with pertinent information about their prognosis, and for tailoring the treatment to each patient’s needs. Applying the PPCF allows a prediction of the outcome of patients with surgically treated RCC on the basis of preoperatively available variables. However, clinical presentation, included in this model, had no significant influence on DFS in the present patients. By contrast, using the PAPS resulted in an improvement in the predictive value and in a greater discrimination between patients subdivided into a good and a poor prognosis group and hence is suitable for preoperative risk assignment.