A new prognostic model for cancer-specific survival after radical cystectomy including pretreatment thrombocytosis and standard pathological risk factors

Authors


Georgios Gakis, Department of Urology, Eberhard-Karls University, Hoppe-Seyler Strasse 3, D-72076 Tübingen, Germany. e-mail: georgios.gakis@web.de

Abstract

Study Type – Prognosis (cohort series)

Level of Evidence 2a

What's known on the subject? and What does the study add?

Preoperative thrombocytosis has been identified as a predictor of poor outcome in various cancer types. However, the prognostic role of platelet count in patients with invasive bladder cancer undergoing radical cystectomy is unknown.

The present study demonstrates that preoperative thrombocytosis is an independent risk factor for decreased cancer-specific survival after radical treatment of invasive bladder cancer. We developed a new prognostic scoring model for cancer-specific outcomes after radical cystectomy including platelet count and established pathological risk factors. Consideration of platelet count in the final model increased its predictive accuracy significantly. Thrombocytosis may be a useful parameter to include within established international bladder cancer nomograms.

OBJECTIVE

  • • To investigate the oncological significance of preoperative thrombocytosis in patients with invasive bladder cancer undergoing radical cystectomy, as it has been reported as a marker for aggressive tumour biology in a variety of solid tumours.

PATIENTS AND METHODS

  • • The series comprised 258 patients undergoing radical cystectomy between 1999 and 2010 in whom different clinical and histopathological parameters were assessed.
  • • Elevated platelet count was defined as >450 × 109/L.
  • • Based on regression estimates of significant parameters in multivariable analysis a new weighted scoring model was developed to predict cancer-specific outcomes.

RESULTS

  • • The median follow-up was 30 months (6–116).
  • • Of the 258 patients, 26 (10.1%) had elevated and 232 (89.9%) had normal platelet count. The 3-year cancer-specific survival in patients with normal and elevated platelet count was 61.5% and 32.7%, respectively (P < 0.001).
  • • In multivariable analysis, cancer-specific survival was significantly lower in patients with locally advanced disease (≥pT3a) (relative risk 2.91, 1.54–5.65; P= 0.001), positive soft tissue surgical margins (4.03, 1.99–7.92; P= 0.001) and thrombocytosis (2.68, 1.26–5.14; P= 0.011).
  • • The 3-year cancer-specific survival in patients with a score 0 (low risk), 1–2 (intermediate risk) and 3–5 (high risk) was 81.0%, 54.8% and 8.2%, respectively (P < 0.001).
  • • Consideration of preoperative platelet count in the final model increased its predictive accuracy by 1.8% with a concordance index of 0.745 (P= 0.040).

CONCLUSIONS

  • • The presence of thrombocytosis at radical cystectomy portends unfavourable prognosis.
  • • We constructed a simple weighted prognostic model for cancer-specific outcomes after radical cystectomy based on pretreatment platelet count and established pathological risk factors.
  • • These data warrant external validation and may allow for tailored monitoring and selection of appropriate patients for neoadjuvant and adjuvant trials.
Abbreviations
RC

radical cystectomy

PLT

platelet

TURBT

transurethral bladder resection

STSM

soft-tissue surgical margin

CSS

cancer-specific survival

RR

relative risk

INTRODUCTION

Up to 40% of patients undergoing radical cystectomy (RC) for muscle-invasive bladder cancer experience recurrence within 10 years [1,2]. As the disease course varies, it is essential to identify risk factors associated with poor survival to improve the choice of adjuvant therapies. In addition to classical clinical and histopathological risk factors (such as the TNM stage) associated with poor outcome, various molecular markers have also been identified [3]. Furthermore, haematological and inflammatory changes have been shown to indicate poor survival [4].

Elevated platelet (PLT) count is frequently observed in patients with cancer and has been reported as a prognostic factor in several tumour types including lung cancer [5], glioblastoma [6], renal cancer [7] and several gastrointestinal tumours [8]. Patients suffering from changes in multiple haematological lineages have an even worse prognosis in some tumour types [8]. The predictive value of elevated PLT count in patients with bladder cancer is unknown. This retrospective study is the first to evaluate the prevalence and prognostic significance of pretreatment PLT count in patients with invasive bladder cancer undergoing RC.

PATIENTS AND METHODS

In this retrospective study (approved by the local ethics committee Tübingen, No. 417/2010A) we reviewed the clinical and pathological records of 258 patients who underwent RC and bilateral pelvic lymphadenectomy without neoadjuvant treatment at our hospital for invasive bladder cancer between 1999 and 2010. Blood count was regularly performed 1–3 days before cystectomy, with an elevated PLT count defined as >450 × 109/L. For postoperative assessment of PLT count, data sets from postoperative blood count analyses at day 10 were reviewed. The following clinical and pathological parameters were assessed: gender, age at RC, number of transurethral bladder resections (TURBTs) before RC, tumour stage and grade, presence of carcinoma in situ, urinary diversion, lymphovascular invasion, lymph node tumour involvement, non-pure urothelial carcinoma pathology and preoperative hydronephrosis, tumour size, soft-tissue surgical margins (STSMs), leucocyte count, haemoglobin level (normal level >14 g/dL in men and >12 g/dL in women) and postoperative chemotherapy.

HISTOLOGICAL ASSESSMENT

The histopathological evaluation was performed at a single pathology department and was based on the TNM classification approved by the American Joint Committee on Cancer [9]. The pathological macroscopic and microscopic evaluation of RC specimens included cross-sectioning of the entire specimen with immunohistochemical staining for cytokeratin-5/6 and −20 to identify the presence of urothelial carcinoma. Positive surgical margins were defined as the microscopic presence of malignant cells at the resection margins.

FOLLOW-UP

Patient charts and physician records were reviewed to determine clinical outcome. Patients generally were seen postoperatively at least every 3–4 months for the first year, semi-annually for the second and third years, and annually thereafter. Follow-up examinations included radiological imaging with CT in all patients. In addition to physical examination with laboratory testing, intravenous pyelography, cystoscopy, urine cytology, urethral washings and bone scintigraphy were carried out if indicated. Local recurrence was defined as recurrence in the surgical bed, distant as recurrence at distant organs. Clinical outcomes were measured from the date of cystectomy to the date of first documented recurrence at CT, the date of death, or the date of last follow-up when the patient had not experienced disease recurrence.

STATISTICAL ANALYSIS AND DEVELOPMENT OF PROGNOSTIC MODEL

For univariable analysis, the Fisher exact test was used for nominal data and Student's t test for scaled data. For multivariable analysis, the Cox proportional hazard analysis was carried out to evaluate risk factors for cancer-specific death. Kaplan–Meier plots with the log-rank test were used to estimate cancer-specific survival (CSS). Of the 258 patients, five died perioperatively within the postoperative 30-day interval and eight were lost to follow-up and were therefore excluded from survival analysis. A total of 245 patients were considered for final survival analysis.

A weighted prognostic model for CSS was developed based on the regression coefficients from the final multivariable Cox proportional hazard analysis. The model-building algorithm was based on the 13 clinical and pathological risk factors for cancer-specific death after RC [1,10]. For multivariable Cox regression analysis, only the most significant parameters of univariable analysis were considered. Following multivariable analysis, independent risk factors were used to create a simple prognostic model as has been carried out in a similar analysis for renal cell carcinoma patients [11]. In the final model, which included 245 patients, the regression coefficient of each parameter was divided by the coefficient of the parameter with the highest regression coefficient, multiplied by 3 and rounded to the nearest integer [11]. The predictive ability of the final model was analysed using the concordance index [12]. Receiver–operator curves were constructed to evaluate the predictive accuracy of the final model including standard pathological risk factors with and without the inclusion of pretreatment thrombocytosis. In this context, a concordance index of 1.0 indicates that the model discriminates perfectly between patients with different clinical outcomes whereas a value of 0.5 is equivalent to a toss of a coin indicating that the model contains no predictive information [13]. P values are two-sided and P < 0.05 is considered as significant. Statistical analysis was performed using JMP® 8.0.2. Values are given as the mean ±sem for all continuous variables or as the median (range) for non-continuous variables.

RESULTS

Of the 258 patients, 26 (10.1%) had elevated PLT count preoperatively and 232 (89.1%) normal PLT count (Table 1). By contrast to preoperative values, 10 days after surgery, 157 patients (60.9%) had elevated and 101 patients (39.1%) normal PLT counts (P≤ 0.001). In univariable analysis, preoperatively elevated PLT count was associated with elevated leucocyte count (defined as >10 000/mm3), decreased haemoglobin level (defined as male < 14 g/dL, female < 12 g/dL), increased tumour size (all P < 0.001) and mixed carcinoma pathology (P= 0.018). No significant differences were found between patients with normal and elevated PLT count with regard to gender, age at RC, T stage, lymph node tumour involvement, tumour grade, presence of carcinoma in situ, postoperative chemotherapy, positive STSMs, lymphovascular invasion, number of TURBTs before RC, preoperative hydronephrosis and number of retrieved lymph nodes (see Table 1).

Table 1. Univariable analysis of clinical and pathological parameters in 258 bladder cancer patients with normal or elevated platelet count
 Normal PLT, (150–450) × 109/LElevated PLT, >450 × 109/L P
  1. UC, urothelial carcinoma.

Number of patients (%)232 (89.9)26 (10.1) 
Patient characteristics   
 Gender   
  Male180 (77.5%)21 (80.1%)0.80
  Female52 (22.5%)5 (19.9%) 
 Age at RC (years)   
  Median68.066.10.81
  Range38.9–86.145.8–81.1 
Clinical variables   
 Number of retrieved lymph nodes   
  Mean18.119.80.45
  Median1820 
  Range1–531–34 
 Time interval between last TURBT and RC   
 Median (days)32410.69
  Range (days)4–47610–231 
 Postoperative chemotherapy38 (16.4%)3 (11.5%)0.77
 Continent diversion137 (59.1%)11 (42.3%)0.09
 Number of TURBTs before RC   
  Mean2.01.30.62
  Median11 
  Range1–12 (9.1%)1–5 
 Anaemia present (<12 g/dL in women; <14 g/dL in men)21 (9.1%)6 (23.1%) <0.001
 Elevated leucocyte count71 (30.6%)19 (73.1%) <0.001
 Preoperative hydronephrosis47 (20.3%)8 (30.1%)0.19
Histopathological variables   
 T stage   
  Organ-confined (≤pT2b)122 (52.6%)8 (30.8%)0.09
   pT05 (2.2%)0 
   pTa7 (5.7%)1 (3.8%) 
   pTis4 (1.7%)0 
   pT137 (30.3%)1 (3.8%) 
   pT2a41 (17.7%)2 (7.7%) 
   pT2b28 (12.1%)4 (15.4%) 
  Extravesical (≥pT3a)111 (47.8%)18 (69.2%) 
   pT3a39 (16.8%)4 (15.4%) 
   pT3b31 (13.4%)8 (30.8%) 
   pT4a31 (13.4%)7 (26.9%) 
   pT4b11 (4.7%)3 (11.5%) 
 Tumour grade   
  G102 (7.7%)0.63
  G262 (26.7%)5 (19.2%) 
  G3170 (73.3%)19 (73.1%) 
 Cancer in situ at RC   
  Present67 (28.9%)5 (19.3%)0.36
  Absent165 (71.1%)21 (80.7%) 
 Lymphovascular invasion present76 (32.8%)7 (26.9%)0.99
 Non-pure UC histology12 (5.2%)5 (19.2%) 0.018
 Squamous cell carcinoma5 (2.2%)3 (11.6%) 
 Adenocarcinoma3 (1.3%)0 
 Sarcomatoid differentiation2 (0.9%)1 (3.8%) 
 Neuroendocrine differentiation2 (0.9%)1 (3.8%) 
 Tumour size (cm)   
  Mean3.34.4 0.002
  Median34 
  Range0–121.5–10.5 
 Positive STSMs22 (9.5%)4 (15.4%)0.29
 Node-positive disease at RC62 (26.7%)8 (12.1%)0.63

The median follow-up was 30 months (range 6–116). The median time between last TURBT and RC was 41 days (range 10–231) in patients with elevated PLT count and 32 days (4–476) in patients with normal PLT count (P= 0.69). The CSS rate at 3 and 5 years in the total patient cohort was 59% and 55%, respectively. Cancer-specific death occurred in 61 patients with preoperative PLT count ≤ 450 × 109/L and 13 patients with >450 × 109/L. The 3-year CSS in patients with normal and elevated pretreatment PLT count was 61.5% and 32.7%, respectively (P < 0.001) (see Fig. 1). No difference was observed in median 3-year CSS in patients with elevated and normal postoperative PLT count (56.6% and 61.7%; P= 0.46).

Figure 1.

Kaplan–Meier analysis for CSS in patients with normal (≤450 × 109/L) and elevated (>450 × 109/L) PLT count (P < 0.001) (CS, cancer-specific).

Number of patients at risk at given intervals (%)
Time (in months)01224364860
PLT count ≤ 450 × 109/L22113699583520
PLT count > 450 × 109/L24108322

In univariable Cox regression analysis, CSS was decreased in patients with preoperatively elevated PLT counts (relative risk [RR] 2.71, 1.42–4.77; P= 0.003), extravesical disease (RR 4.06, 2.49–6.91; P < 0.000), lymph node tumour involvement (RR 3.95, 2.49–6.27; P < 0.001), positive STSMs (RR 7.39, 4.25–12.35; P < 0.001), lymphovascular invasion (RR 2.72, 1.73–4.26; P < 0.001), preoperative anaemia (RR 2.09, 1.27–3.52, P= 0.004), postoperative chemotherapy (RR 2.20, 1.34–3.53; P= 0.002), non-pure urothelial carcinoma pathology (RR 3.16, 1.46–6.02; P= 0.005), tumour size >3 cm (RR 1.70, 1.07–2.71; P= 0.025) and age >65 years (RR 1.65, 1.04–2.71; P= 0.034) (see Table 2). The presence of postoperative thrombocytosis did not affect CSS (RR 1.17, 0.72–1.94; P= 0.52).

Table 2. Univariable Cox regression analysis of risk factors for cancer-specific death in patients with invasive bladder cancer
ParameterRR (95% CI) P
  1. Bold indicates a statistically significant difference.

Elevated vs normal PLT count2.71 (1.42–4.77) 0.004
≥pT3a vs ≤pT2b4.06 (2.49–6.91) <0.001
pN+ vs pN03.95 (2.49–6.27) <0.001
Positive vs negative STSMs7.39 (4.25–12.35) <0.001
Lymphovascular invasion present vs not2.72 (1.73–4.26) <0.001
Anaemia present vs absent2.09 (1.27–3.52) 0.004
Postoperative chemotherapy performed vs not performed2.20 (1.34–3.53) 0.002
Non-pure vs pure UC histology3.16 (1.46–6.02) 0.005
Tumour size >3 cm vs ≤3 cm1.70 (1.07–2.71) 0.025
Age > 65 years vs ≤65 years1.65 (1.04–2.71) 0.034
Elevated vs normal leukocytes1.52 (0.95–2.41)0.077
Hydronephrosis1.43 (0.85–2.31)0.17
Gender male vs female0.74 (0.45–1.25)0.25

In multivariable analysis, adjusted for most significant parameters of univariable analysis, CSS was significantly lower in patients with locally advanced disease (≥pT3a) (RR 2.91, 1.54–5.65; P= 0.001), positive STSMs (4.03, 1.99–7.92; P= 0.001) and thrombocytosis (2.68, 1.26–5.14; P= 0.011) (see Table 3). A weighted prognostic model was constructed for CSS after RC using the regression coefficients of the final multivariable model. The final model consisted of pT stage, resection margin status and pretreatment PLT count (see Table 2). The score was calculated as 2 (if ≥pT3a) + 2 (if ≥positive STSMs) + 1 (if PLT count > 450 × 109/L) and 0 (otherwise).

Table 3. Multivariable Cox proportional hazard analysis for cancer-specific death in patients with invasive bladder cancer
VariableFull modelFinal model
Regression coefficientRisk ratio (95% CI) P Regression coefficientRisk ratio (95% CI) P
  1. Bold indicates a statistically significant difference.

Elevated vs normal PLT count0.4942.68 (1.26–5.14) 0.011 0.5122.71 (1.41–4.89) 0.004
≥pT3a vs ≤pT2b0.5342.91 (1.54–5.65) <0.001 0.6123.49 (1.99–6.38) <0.001
Positive vs negative STSMs0.6974.03 (1.99–7.92) <0.001 0.7194.21 (2.35–7.34) <0.001
pN+ vs pN00.1971.48 (0.77–2.81)0.23
Lymphovascular invasion present vs not0.061.13 (0.61–2.09)0.68

The median score of the final model was 1 (mean ±sem 1.47 ± 0.06, range 0–5). The 3-year/5-year CSS in patients with a score 0 (low risk), 1–2 (intermediate risk) and 3–5 (high risk) was 81.0%/78.1%, 54.8%/51.9% and 8.2%/0%, respectively (P < 0.001; Fig. 2). Consideration of preoperative PLT count in the final model increased its predictive accuracy by 1.8% with a concordance index of 0.745 (P= 0.040).

Figure 2.

CSS by score categories in the final prognostic model (differences between subgroups, P < 0.001; CS, cancer-specific).

Number of patients at risk at given intervals
Time (in months)01224364860
0 (low risk)1208160372314
1–2 (intermediate risk)905945251511
3–5 (high risk)3595200
Mean ±sem % CSS (% of patients at risk)
Score categoriesNo. of patients (%)Postoperative year 3Postoperative year 5
0 (low risk)120 (49.0)81.0 ± 4.7 (36)78.1 ± 5.3 (12)
1–2 (intermediate risk)90 (36.7)54.8 ± 6.8 (24)51.9 ± 7.0 (10)
3–5 (high risk)35 (14.3)8.2 ± 5.4 (2)0 (0)

DISCUSSION

Haematological and inflammatory changes have been identified as prognostic factors in muscle-invasive bladder cancer [4]. In contrast to other malignancies, where thrombocytosis has been shown to be associated with poor survival, the role of PLTs in bladder cancer has not been investigated yet. We therefore aimed to evaluate the significance of PLTs in patients undergoing RC for muscle-invasive bladder cancer.

In the present study, the rate of preoperatively elevated PLT count in patients with bladder cancer undergoing RC was 10.1%. The presence of thrombocytosis varies in different tumour types. Compared with reported rates in bronchial cancer (53%) [5], renal cancer (20%–57%) [14,15] and glioblastoma (19%) [6], our 10.1% incidence is low. However, in these studies thrombocytosis was defined as >400 × 109/L and some of these included a high proportion of patients with advanced and metastatic disease.

In univariable analysis thrombocytosis was associated with increased tumour size, preoperative anaemia, leucocytosis and mixed pathology. In general, non-urothelial carcinomas exhibit a more aggressive tumour biology leading to increased risk for progression and death [16] and this might be due to a higher capability to secrete growth factors and cytokines thereby promoting systemic inflammation and thrombocytosis. Subtypes of mixed carcinomas of the bladder show rapid tumour growth and are capable of causing paraneoplastic syndromes including leukemoid reactions and thrombocytosis [17,18].

In univariable Cox regression analysis CSS was significantly decreased in patients with thrombocytosis, extravesical tumour stage, lymph node tumour involvement, positive STSMs, lymphovascular invasion, preoperative anaemia, tumour size >3 cm and age >65 years. These findings are in concordance with prior studies reporting the prognostic significance of these clinical and pathological risk factors for cancer-specific death after RC [1,2]. With regard to the number of cancer-specific deaths occurring in the present study population, the Cox regression multivariable model can sufficiently evaluate not more than five risk factors [19], which consisted of the five most significant parameters of univariable analysis. However, multivariable analysis showed increased PLT count, extravesical disease and positive STSMs to be the only independent risk factors for decreased CSS in this patient cohort.

The validity of a multivariable Cox regression model encompassing clinicopathological and serological or molecular parameters has to be determined by receiver–operator curve analysis. It has to be proved that the addition of a given variable (i.e. thrombocytosis) increases the predictive accuracy of established clinical and pathological parameters significantly [20]. As only 10% of our patients displayed pretreatment thrombocytosis which was associated with non-pure urothelial pathology and even worse survival (median CSS at 3 years postoperatively 31%) the presence of thrombocytosis may be used for identifying those patients at highest risk for cancer-specific death.

Impaired prognosis of patients with thrombocytosis has also been demonstrated in other tumour entities [6,8,21]. However, a marker identified as an independent risk factor does not necessarily lead to better predictive accuracy when considered along with established predictors. Therefore, based on regression estimates of significant parameters in multivariable analysis [11], we developed a simple and user-friendly prognostic model to predict CSS after RC. This model accounts for tumour stage, resection margin status and the presence of preoperative thrombocytosis and yielded a predictive accuracy of 74.5%. Of note, the inclusion of thrombocytosis as an independent variable significantly improved the predictive accuracy by 1.8%, which corresponds to 18 out of 1000 patients who would be classified correctly when thrombocytosis is included as a risk factor in the final model. To avoid the problem of lack of standardization of threshold values, in multivariable analysis a biomarker can be used as a continuous coded parameter in order to strengthen its relevance as a prognostic marker. However, defining cut-off values in the final marker-based model, as has been carried out in this analysis, reduces the risk of overfitting the model to the data and facilitates its external validation. In fact, prespecified cut-offs can be deemed a hallmark of legitimate biomarker research instead of conducting a strategy of biomarker discovery based on fishing expeditions [20]. Interestingly, in multivariable analysis, node-positive disease lost its independent value when adjusted for thrombocytosis, tumour stage, STSMs, postoperative chemotherapy and preoperative anaemia. This finding underscores the independent significance of tumour stage, STSMs and thrombocytosis for predicting cancer-specific outcomes since patients with extravesical disease and positive STSMs inherit a higher risk of concomitant lymph node tumour involvement, and this may result in its loss of independent prognostic significance [1,22].

Preoperative thrombocytosis may also be used as an additional parameter to identify bladder cancer patients with advanced disease and poor prognosis. As in a high proportion of patients undergoing RC clinical understaging is present [1], additional parameters are required for identification of patients who might benefit from neoadjuvant chemotherapy. However, since RC is a major surgical procedure postoperative thrombocytosis might only be of a reactive nature and not cancer related anymore, and this might be associated with loss of its prognostic impact.

Our results raise the question whether thrombocytosis is a result of aggressive tumour biology or whether PLTs influence tumour behaviour. Tumour cells secrete growth factors and cytokines stimulating PLT production. Thrombocytosis is assumed to be a result of the release of several growth factors and cytokines by tumour cells [23,24]. For example, interleukin-6, which is produced by malignant tumours and inflammatory tissues, potently stimulates PLT production [25]. On the other hand, PLTs can promote tumour cell growth by secreting growth factors such as vascular endothelial growth factor, leading to increased angiogenesis [26]. Furthermore, PLTs could be identified as promoters for metastases [27], and several mechanisms have been proposed. Tumour cells aggregated with PLTs may shield cancer cells from the host immune system, especially natural killer cells [28]. Furthermore, PLTs facilitate adhesion of cancer cells to endothelial cells, which may promote the essential step of extravasation in the metastatic process [29]. In metastatic bone disease, PLTs could be identified as indirect activators of osteoclasts [27]. Thus, PLTs promote tumour cell growth and cancer progression.

The correlation observed between preoperative thrombocytosis, anaemia and elevated leucocyte count indicates an effect of tumour cells on multiple haematological lineages. Several studies have shown that cytokines such as interleukin-6 and tumour necrosis factor α, which promote inflammation and PLT production, are also responsible for tumour-induced anaemia [30,31]. Anaemia is assumed to be a consequence of either inflammation or inhibition of transcription factors inducing erythropoesis [32]. These transcription factors are also involved in PLT production: for example, GATA-2 is a transcription factor that is downregulated during erythroid differentiation, and its overexpression leads to anaemia and stimulates differentiation of megakaryocytes [33].

In our opinion, our prognostic model serves for the treating urologist to address the need of perioperative chemotherapy in patients undergoing RC. First, preoperative thrombocytosis should alert physicians to the presence of non-pure urothelial carcinoma components in the bladder. However, as only approximately 20% of patients with pretreatment thrombocytosis displayed non-pure urothelial carcinoma histology in their RC specimens, thrombocytosis may also help to identify those patients with pure urothelial carcinoma histology at highest risk of recurrence and subsequent cancer-specific death. Both implications may have the same clinical consequence: the question whether a patient may benefit from neoadjuvant or adjuvant chemotherapy. Recently, it has been shown that patients with mixed histological features derive a survival benefit from neoadjuvant chemotherapy [34]. In addition, due to the fact that an unequivocal follow-up scheme in patients after RC has not been established so far [35], the proposed scoring system may help to risk-stratify an individual patient postoperatively, and this might help to tailor the frequency and modality of follow-up investigations within prospective studies.

The study has limitations inherent to any retrospective studies. First, thrombocytosis can be caused by inflammation and physical stress (e.g. prior TURBT, although we did control for this in univariable analysis) that may have an impact on the results. In this respect, prospective trials would be helpful to exclude this possible bias. Second, the results of this study await formal external validation before their implementation into clinical routine and validation of the model with established bladder cancer nomograms [36,37]. Since our median follow-up was 30 months, we focused on 3-year survival rates. With regard to the fact that the median time to local or systemic recurrence after RC ranges between 7 and 18 months [1], our follow-up time has to be regarded as sufficient to provide meaningful conclusions for cancer-specific outcomes.

This to our knowledge is the first study in the literature that reports the role of preoperative thrombocytosis in patients undergoing RC for bladder cancer as an independent prognosticator for decreased CSS. The inclusion of pretreatment thrombocytosis in a simple and user-friendly prognostic model increases the predictive accuracy of pathological risk factors significantly. In addition, routine preoperative assessment of PLT count in patients undergoing RC for bladder cancer is an easy method which may provide additional prognostic information and improved selection of patients for neoadjuvant and adjuvant trials. Moreover, the inclusion of PLT counts into current nomograms may improve their predictive accuracy.

This is the first study in the literature to demonstrate that preoperative thrombocytosis is an independent risk factor for decreased CSS after RC. We constructed a simple weighted prognostic model for cancer-specific outcomes after RC based on pretreatment PLT count. The incorporation of PLT count in this model showed a significant improvement in the predictive accuracy of established pathological risk factors. These data warrant external validation and may allow for tailored monitoring and selection of appropriate patients for neoadjuvant and adjuvant trials.

CONFLICT OF INTEREST

None declared.

Ancillary