Long-term kidney function and mortality after radical cystectomy and ileal conduit formation.

Background: Treatment for bladder cancer includes radical cystectomy (RC) and urinary diversion, RC is associated with long-term morbidity, kidney function deterioration and mortality. Our aim was to identify risk factors associated with postoperative long-term kidney function decline and mortality. Methods: Retrospective study of patients with RC and urinary diversion in Beaumont Hospital from 1996 to 2016. We included patients who had follow up data of at least two years post procedure. We assessed the following outcomes: estimated glomerular ltration rate (eGFR) preoperatively, at rst and second year post-procedure, kidney function decline >10 ml/min/1.73m 2 , dialysis commencement and mortality. Logistic regression analyses were applied to assess risk factors associated, a p-value <0.05 was considered signicant. Results: We included 264 patients, with median age 68.3 years, 73,7% males. The most common diagnosis was bladder cancer 93.3%, TNM stages were grouped in T ≥ 2 75.9%, N ≥ 1 47.6% and M1 28%. The median eGFR preoperative was 65.8 ml/min/1.73m 2 and after 2 years 58.2 ml/min/1.73m 2 (p:0.009), 5.6% required chronic dialysis and 32.8% had a decrease >10 ml/min/1.73m 2 . Risk factors associated with ESKD included; age (HR:1.13, CI95% 1.05-1.22), and pre-operative eGFR (HR:1.04, CI95% 1.01-1.07). Overall mortality was 43.2% and 75.9% at 5 and 10 years respectively, risk factors for which were age (HR:1.1, CI95% 1.04-1.18), preoperative eGFR (HR:1.03, CI95% 1.01-1.06) and male gender (HR:14.8, CI95% 1.1-192). Conclusions: Patients RC kidney function and elevated


Background
The international gold standard treatment for muscle invasive bladder cancer with curative intent is a multimodal approach involving radical cystectomy (RC), resection of regional lymph nodes, urinary diversion and neo-adjuvant chemotherapy (1). The two most commonly used procedures for urinary diversion are ileal conduit diversion (ICD) and orthotopic neobladder formation (2). Although surgery itself is associated with signi cant peri-operative morbidity and mortality, the 5-year relative survival rate has improved greatly in recent years (3,4). It has resulted in consideration of the long-term impact of surgery on morbidity, quality of life and it has been shown that patients after RC are at increased lifetime risk of deterioration in kidney function and progression to chronic kidney disease (3)(4)(5)(6)(7).
Chronic kidney disease itself is an independent risk factor for all-cause mortality, and cardiovascular events (8,9). Multiple factors have been implicated in the deterioration of kidney function postoperatively, these can be divided into patient factors such as advancing age, central obesity, diabetes mellitus, hypertension, pre-operative kidney impairment, and iatrogenic factors such as use nephrotoxic chemotherapy, perioperative blood loss, post-operative urinary tract infection or obstruction (10)(11)(12)(13)(14)(15)(16). The trajectory of the deterioration of kidney function is thought to be biphasic: an accelerated decline is seen in the rst two postoperative years with a moderate loss occurring gradually over later years (12).
In this study, we analysed data from all patients who underwent radical cystectomy in our hospital, a tertiary referral centre for urological cancer, from 1996 to 2016. The aim was to identify potential risk factors associated with postoperative and long-term kidney function decline and mortality.

Methods
Following institutional review board approval, a retrospective study was conducted of patients undergoing radical cystectomy from January 1996 to December 2016. We identi ed a total of 338 consecutive adult patients underwent radical cystectomy and urinary diversion attended in Beaumont Hospital Urology department. We excluded patients with partial cystectomy (n = 33), without urinary diversion due to dialysis (n = 2) and patients whose analysis data were unavailable or incomplete (n = 39). We included 264 patients who had undergone total radical cystectomy and ileal conduit diversion or cutaneous ureterostomy who had on going assessment for at least two years post procedure, with clinical and laboratory. We con rmed the mortality status with the National Cancer Centre O ce database, which is maintained prospectively.
The preoperative factors analysed were age at RC, gender, smoking history, concomitant comorbidity (DM and HTN), body mass index (BMI), preoperative kidney function, type of surgical technique, indication for surgery and tumour stage. The main outcomes were mortality, kidney function deterioration and end stage kidney disease (ESKD). We assessed kidney function using the CKD-EPI equation for estimated glomerular ltration rate (eGFR) (17). Deterioration in kidney function was de ned as clinically relevant if there was a >10 ml/min/1.73m 2 decrease in eGFR at 2 years compared to the baseline value before surgery. This threshold was chosen rather than 5ml/min/year because this rate of deterioration has been shown to be superior and is considered more clinically meaningful (14,18).

Data analysis and statistical methods.
Continuous variables are shown using summary statistics and categorical variables as frequencies. We performed exploratory analyses for categorical variables using Fisher's exact or Chi square test and continuous with student's T test. The occurrence of ESKD requiring dialysis was estimated using the Kaplan-Meier method and compared using the Log rank test. Patients who had not experienced the event of interest were censored at the time of the last urologic follow-up or death not owing to events of interest.
Multivariate logistic regression was applied to identify independent risk factors and reported as hazard ratios (HR) with 95% CI. The covariates were selected among baseline variables and compared in a bivariable model, with the patient outcomes of mortality and ESKD. We selected variables to be included in regression models if the p-value was <0.10. We included age, sex and eGFR perioperative in all models.
To achieve model parsimony and stability, we analysed all possible models and selected the best model using the criteria of a change of <10% in the full model B coe cient, and applying a selection procedure based on the dropout criterion of P>0.1. All probabilities were two sided, and a p-value <0.05 was considered signi cant. All analyses were performed using SPSS 19 (IBM-California).

Results
We evaluated 264 patients, the clinic pathological characteristics and distributions are shown in Table 1 The median eGFR pre-operative was 65.8 ml/min/1.73m 2 (IQR 48.5 to 81.5), and after 2 years decreased to 58.2 ml/min/1.73m 2 (IQR 45.3 to 79.2), the difference being statistically signi cant (p: 0.009) (Graphic 01). Fourteen patients (5.6%) went on to develop CKD stage V and required chronic dialysis, 2 of whom required dialysis immediately, median time dialysis commencement was 3.3 years (IQR 8.0). 41.3% of patients had moderate to severe CKD (stage III-V) pre surgery; after 2 years postoperatively this percentage increased to 53.9%. A decrease in eGFR greater than 10 ml/min/1.73m 2 at two years was present in 32.8%. It is noteworthy that 63.3% of patients had a reduction in their eGFR after 2 years, although 36.7% maintained stable kidney function or had a mild increase in their eGFR.
We found that the majority of patients experienced a decrease in kidney function during long-term follow up. The initial eGFR in our cohort was 65.8 ml/min/1.73m 2 , which is similar to previous reports (10,(12)(13)(14)20) with initial eGFR ranging from 65 to 69.7 ml/min, with the exception of Samuel et al, which reported an eGFR of 77. This was most likely due to the fact that the cohort in Samuel et al was younger.
The eGFR after two years was 55 ml/min, again similar to previous reports with ranges from 55 to 59 (11)(12)(13)(14)20), being comparable. 62.8% of patients had a decline of their kidney function during follow up, and 32.8% of patients had a decrease of >10 ml/min/1.73m 2 at 2 years. Previous studies have shown a high prevalence 70% (11,21) at 10 years, and 49% at 5 years (11), however there are also reports with a lower prevalence 34 to 51% (12)(13)(14)(15). The difference in the prevalence across the different studies could be explained by the variety in de nitions used to categorize the decline in kidney function, as well as different follow up times, and different methods of GFR estimation.
There is a predictable decline in the kidney function related to aging; the annual age-related decline in eGFR from age 30 years onward is believed to be ±1 mL/min/1.73 m 2 in a healthy population (22)(23)(24).
This GFR declines by about 8 -10 ml/min/1.73 m 2 per decade (25,26), however a linear decline in eGFR over time is often observed (27) in patients with diabetes or HTN. This decline in persons with diabetes can range from ±2-3 mL/min/1.73m 2 per year (28). In patients with RC and ICD, previous reports have used de nitions such as a decrease of 1 ml/min/year (5,11,12) over a time period of 5 to 10 years, a decrease of >25% (15,20) in eGFR from baseline, or a decline of >10% in eGFR (11,14). The de nition of a clinically signi cant deterioration in kidney function in our study was a decrease of >10 ml/min/1.73m 2 in the rst two years. This was selected because the majority of the observed decline in kidney function in previous studies occurs in the rst 2 years (12,20). This represents the period of major risk for kidney function deterioration, which is likely associated with RC and of clinical signi cance.
A further difference relates to the method used to measure kidney function in patients with RC. The most common methods used are the MDRD (12,14,19), CKD-EPI equations (10,11) and isotopic (5,19) methods. These equations represent a practical way to measure kidney function, which is reliable, easy to use and reproducible. The gold standard to measure GFR is the use of isotopes or inulin; these methods are costly and not widely available in clinical practice. There are reports that suggest that methods like MDRD and CKD-EPI overestimate the GFR comparing this with isotopic methods (19) in patients post RC. Therefore, the differences between these series can be explained by variations in patient selection and the variety of methods used to calculate eGFR.
The use of estimated methods to assess the kidney function have some caveats and considerations. One important consideration is that these equations are dependent on the level of serum creatinine; the level of which re ects kidney function but serum creatinine can also be affected by muscle mass, weight, diet, exercise and other possible factors (29,30). For example, a decrease in body weight after a diagnosis of cancer is a frequent phenomenon, Meyerhardt et al, previously reported a decrease in 37.5% of patients (31). On the other hand, it is unclear whether there is signi cant reabsorption of urea and creatinine in ICD patients, as the contact time of urine is shorter and the reabsorbing surface of an ICD is small. Animal models suggest that much of the creatinine is reabsorbed by solvent drag, a glucose-dependent way of transport. Creatinine is reabsorbed less well by an active carrier mediated transport. As urine normally does not have large amounts of glucose, creatinine may be resorbed to a lesser degree (32).
It should be noted that changes in kidney function were not uniform across patients, in our analysis we showed that 36.7% of our patients has stable or mild increase in their eGFR, and 10.6% an increase of >10%, since previous reports showed similar ndings Rouanne (12)  Our study showed cumulative incidence of dialysis of 5.6% at 5 years; Jin et al (14), reported only 1.2% and Rouanne (12) et al only 2.5%. The low incidence in the Jin (14) et al report may be related to a younger cohort and patient selection. Additionally, the incidence of dialysis may be affected by high mortality since there is a higher incidence of comorbidity, and conservative treatment may have been adopted rather than dialysis.. Long-term kidney function after RC can be adversely affected by several factors, including age, potential nephrotoxic chemotherapy, comorbidities, and urine diversion-related factors. Our results are concordant with previous studies, in that age and preoperative eGFR associated signi cantly with postoperative kidney function on both univariate and multivariate analyses (11,12,33). However, there is not a clear consensus about other risk factors; neither chronic hypertension nor diabetes mellitus were associated with the decline eGFR (10,12,20,33), or different types of ureterointestinal anastomoses, such as Bricker and Wallace (12,33) or urinary infection (12), or chemotherapy (20,33). Nonetheless, other reports have shown associations with HTN (5,11,14), Diabetes (14), hydronephrosis post RC (11), urinary infection (5,11,14) and total subcutaneous fat (10).
While mortality at 10 years in the other cohorts was from 59 to 65% (12,14) in our study it was 76.5%.
After adjustment in multivariate analysis the main risk factors associated were age, pre operative eGFR and sex. The lack of association with previous well-known risk factors like HTN and DM are most likely related to the follow up, the nature of the cancer and a high mortality rate, that have a high impact as a competing event.
The present study is limited by its retrospective nature, and nonrandomized design and the single centre design may be associated with unknown biases. Additionally, we could not include well-known other clinical factors such as nutrition status due to unavailability of this data. Excluding patients without complete data may have introduced selection bias. Despite these limitations, we have analysed a signi cant number of patients, with regular follow up, and have con rmed mortality and dialysis status with the National Cancer Centre O ce and the National Kidney Registry databases which are regularly updated and maintained prospectively. We did not evaluate the gold standard measures of inulin clearance or use of isotope measured GFR, and while we recognise the potentially limited accuracy of measuring serum creatinine in this cohort, the use of the eGFR is the standard clinical practice due to simplicity, cost and availability, and eGFR is mainly used for management and therapeutic approach in clinical practice.

Conclusions
In conclusion, our study suggest that a high proportion of patients have some degree of renal impairment prior to RC and ileal conduit which will likely deteriorate in the post-operative period. Preoperative eGFR and age, correlate signi cantly with early postoperative kidney function deterioration, dialysis and mortality. Therefore, it is critical that from time of diagnosis these patients are optimised as their preoperative eGFR may dictate decisions about chemotherapy choices, their likelihood of requiring dialysis and their survival. Follow-up of postoperative kidney function is critical in order to minimise longterm morbidity and mortality. Finally, further investigation by well-designed prospective studies is necessary to assess the changes of postoperative kidney function decline in patients with bladder cancer.

Declarations
Ethics approval and consent to participate.
The present study was submitted to the ethical committee of the Beaumont Hospital and was approved.

Consent for publication.
Not applicable.
Availability of data materials.
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests.
All authors con rm that they have no con icts of interest related to this research and the results presented in this paper have not been published previously in whole or part, except in abstract format.

Funding.
This study has no external support and was totally done with own resources.
Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.