Impact of serum sodium concentration on survival outcomes in patients with invasive bladder cancer without metastasis treated by cystectomy.

bladder cancer

body mass index

confidence interval

C‐reactive protein

cancer‐specific survival

clinical T

hemoglobin

hazard ratio

lactate dehydrogenase

neoadjuvant chemotherapy

radical cystectomy

transurethral resection of bladder cancer

Approximately 35% of local invasive BC patients who undergo RC develop distant metastasis and eventually die.1, 2 Thus, identification of poor prognostic factors is a major concern for clinicians. Recently, serum sodium concentration was reported to be associated with prognosis in localized cancers (N0M0), such as non‐small cell lung cancer,3 upper urinary tract cancer4 and renal cell carcinoma,5 even if it is within the normal range. However, there are no reports on BC. In the present study, we evaluated the prognostic significance of preoperative clinical, pathological and laboratory factors in local invasive BC after RC.

Between 2000 and 2015, 179 patients with locally invasive BC (pTany, pN0 cM0) who underwent RC with pelvic lymph node dissection in Osaka Rosai Hospital were retrospectively reviewed. Data collected on each patient included age, sex, BMI, cT stage, tumor nuclear grade, values of serum sodium, LDH and CRP, Hb concentrations, administration of NAC, and follow‐up data. All patients underwent TURBT before cystectomy and were histologically diagnosed as having urothelial carcinoma. No patients received adjuvant chemotherapy. Survival data were available for all patients, and laboratory data were obtained at the first visit before TURBT. Statistical analyses were carried out using jmp 12 (SAS Institute, Cary, NC, USA).

Perioperative information is summarized in Table S1. The median follow‐up duration was 50 months, and the 5‐year CSS estimate was 72.8%. Overall, 59 (33.6%) patients experienced recurrence and 49 (27.3%) patients died of BC.

Kaplan–Meyer analysis and log–rank tests identified cT stage (P = 0.0050), tumor nuclear grade (P = 0.0357), CRP (P = 0.0004) and serum sodium concentration (P = 0.0355) as showing significant associations with CSS (Fig. S1a–d). The optimal cut‐off values of CRP and serum sodium by receiver operating curve analysis were 0.4 mg/dL and 139 mEq/L, respectively. Thereafter, we divided the patients into two groups according to CRP >0.4 mg/dL (high‐CRP, n = 53) and CRP ≤0.4 mg/dL (low‐CRP, n = 126), and Na ≥140 mEq/L (high‐Na, n = 127) and Na ≤139 mEq/L (low‐Na, n = 52) for further study. We confirmed that CRP level and serum sodium concentration were most significantly related to CSS when comparing the two groups according to the optimal cut‐off values (Fig. S1e,f).

Univariate and multivariate Cox proportional hazards analyses were used to evaluate the association of the 10 preoperative factors with CSS, both when divided into two groups (Table 1) and when considered as a continuous variable (Table S2). In univariate analysis, cT stage, tumor nuclear grade, CRP level and serum sodium concentration were significantly associated with CSS by both analyses. In the multivariate models, serum sodium concentration and CRP level correlated significantly with CSS by two‐groups analysis (Table 1), and by continuous variable form, serum sodium concentration was also an independent risk factor for CSS (Table S2).

Table 1

Univariate and multivariate Cox regression analysis of 10 preoperative parameters associated with cancer‐specific survival according to the optimal cut‐off value

	n	Univariate analysis		Multivariate analysis
		HR (95% CI)	P‐value	HR (95% CI)	P‐value
Age (years)
<69  ≥69	90 89	Reference 1.15 (0.65–2.03)	0.6259
Sex
Male  Female	133 46	Reference 0.83 (0.40–1.57)	0.5779
BMI (kg/m2)
<24.0  ≥24.0	105 35	Reference 0.81 (0.35–1.69)	0.5960
cT stage
≤cT2  ≥cT3	101 78	Reference 2.22 (1.26‐4.02)	0.0056	Reference 1.62 (0.89–3.02)	0.1133
Grade
G2  G3	36 132	Reference 2.59 (1.13–7.50)	0.0224	Reference 2.06 (0.87–6.07)	0.1107
Na
≤139 mEq/L  ≥140 mEq/L	52 127	Reference 0.51 (0.29–0.93)	0.0280	Reference 0.54 (0.30–0.98)	0.0442
LDH
Normal  High	150 29	Reference 1.52 (0.72–2.93)	0.2585
CRP
≤0.4 mg/dL  >0.4 mg/dL	126 53	Reference 3.01 (1.71–5.30)	0.0002	Reference 2.68 (1.50–4.84)	0.0010
Hb
Normal  Anemia	102 77	Reference 1.29 (0.73–2.27)	0.3724
NAC
Yes  No	85 94	Reference 1.26 (0.72–2.25)	0.4217

Total n = 179. Optimal cut‐off value Na 139 mEq/L and CRP 0.4 mg/dL.

A prognostic model of risk classification was then constructed based on CRP level and serum sodium concentration as independent predictors of CSS (Table 1). Patients were divided into the low‐risk, moderate‐risk or high‐risk group according to the number of positive independent prognostic factors: CRP ≥0.4 mg/dL and Na ≤139 mEq/L. Patients with low‐CRP and high‐Na were classified into the low‐risk group (n = 94), patients with one prognostic factor into the moderate‐risk group (n = 65), and patients with high‐CRP and low‐Na into the high‐risk group (n = 20). Kaplan–Meyer analysis and log–rank tests showed the respective 5‐year CSS estimates to be 88.6%, 60.6% and 37.5% for the low‐risk, moderate‐risk and high‐risk groups (Fig. S2).

Some meta‐analyses showed that preoperative laboratory data, such as CRP and Hb, were predictive factors for prognosis after RC.6, 7 However, there is no report on serum sodium concentration. This is the first report to show preoperative Na ≤139 mEq/L to be related to poor prognosis after RC. Recently, high immune response was reported to lead to the development of hyponatremia because of pro‐inflammatory cytokines.8 Therefore, we speculated that low‐Na might affect the cancer‐related immune response. Our patients who presented with both Na ≤139 mEq/L and CRP >0.4 mg/dL had a poor prognosis, and therefore might have benefited from adjuvant chemotherapy or extended lymph node dissection. However, because the present retrospective study included only a small number of cases from a single institution, the results still need to be validated in other cohorts.

Conflict of interest

None declared.

Figure S1. The association of preoperative parameters with CSS. (a) The cT stage, (b) tumor nuclear grade and (c) CRP showed a significant association with CSS. (d) Serum sodium concentration showed a significant relationship to CSS when divided into two groups according to the median value. When divided into two groups according to the optimal cut‐off value, (e) CRP level and (f) serum sodium concentration showed a most significant relationship to CSS (Kaplan–Meier log–rank test).

Click here for additional data file.

Figure S2. CSS for risk classification (Kaplan–Meier log–rank test).

Click here for additional data file.

Table S1. Patients characteristics (n = 179).

Click here for additional data file.

Table S2. Univariate and multivariate Cox regression analysis of 10 preoperative parameters associated with CSS with continuous variable form (n = 179).

Click here for additional data file.